Merge branch 'master' into asm-shader-rework-1

Conflicts:
	src/mesa/shader/arbprogparse.c

616 files changed, 31483 insertions, 20929 deletions

diff --git a/Makefile b/Makefile
index 146c52093b2..6eed65e40b7 100644
--- a/Makefile
+++ b/Makefile
@@ -345,7 +345,9 @@ DRI_FILES = \
 	$(DIRECTORY)/src/mesa/drivers/dri/common/xmlpool/*.[ch]		\
 	$(DIRECTORY)/src/mesa/drivers/dri/common/xmlpool/*.po		\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/*.[chS]			\
+	$(DIRECTORY)/src/mesa/drivers/dri/*/*/*.[chS]			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/Makefile			\
+	$(DIRECTORY)/src/mesa/drivers/dri/*/*/Makefile			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/Doxyfile			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/server/*.[ch]
 
@@ -388,7 +390,17 @@ GLW_FILES = \
 	$(DIRECTORY)/src/glw/glw.pc.in			\
 	$(DIRECTORY)/src/glw/depend
 
+GLEW_FILES = \
+	$(DIRECTORY)/include/GL/glew.h			\
+	$(DIRECTORY)/include/GL/glxew.h			\
+	$(DIRECTORY)/include/GL/wglew.h			\
+	$(DIRECTORY)/src/glew/*.c			\
+	$(DIRECTORY)/src/glew/Makefile			\
+	$(DIRECTORY)/src/glew/SConscript		\
+	$(DIRECTORY)/src/glew/LICENSE.txt
+
 DEMO_FILES = \
+	$(GLEW_FILES)					\
 	$(DIRECTORY)/progs/beos/*.cpp			\
 	$(DIRECTORY)/progs/beos/Makefile		\
 	$(DIRECTORY)/progs/images/*.rgb			\
@@ -415,6 +427,7 @@ DEMO_FILES = \
 	$(DIRECTORY)/progs/glsl/*.c			\
 	$(DIRECTORY)/progs/glsl/*.frag			\
 	$(DIRECTORY)/progs/glsl/*.vert			\
+	$(DIRECTORY)/progs/glsl/*.shtest		\
 	$(DIRECTORY)/progs/windml/Makefile.ugl		\
 	$(DIRECTORY)/progs/windml/*.c			\
 	$(DIRECTORY)/progs/windml/*.bmp			\
diff --git a/bin/mklib b/bin/mklib
index 24449450068..db97087c0a6 100755
--- a/bin/mklib
+++ b/bin/mklib
@@ -971,5 +971,6 @@ esac
 #
 if [ ${INSTALLDIR} != "." ] ; then
     echo "mklib: Installing" ${FINAL_LIBS} "in" ${INSTALLDIR}
+    test -d ${INSTALLDIR} || mkdir -p ${INSTALLDIR}
     mv ${FINAL_LIBS} ${INSTALLDIR}/
 fi
diff --git a/configure.ac b/configure.ac
index 9b65d96b396..9d318b3fc20 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1159,6 +1159,21 @@ AC_ARG_WITH([xorg-driver-dir],
     [XORG_DRIVER_INSTALL_DIR="${libdir}/xorg/modules/drivers"])
 AC_SUBST([XORG_DRIVER_INSTALL_DIR])
 
+AC_ARG_WITH([max-width],
+    [AS_HELP_STRING([--with-max-width=N],
+                    [Maximum framebuffer width (4096)])],
+    [DEFINES="${DEFINES} -DMAX_WIDTH=${withval}";
+     AS_IF([test "${withval}" -gt "4096"],
+           [AC_MSG_WARN([Large framebuffer: see s_tritemp.h comments.])])]
+)
+AC_ARG_WITH([max-height],
+    [AS_HELP_STRING([--with-max-height=N],
+                    [Maximum framebuffer height (4096)])],
+    [DEFINES="${DEFINES} -DMAX_HEIGHT=${withval}";
+     AS_IF([test "${withval}" -gt "4096"],
+           [AC_MSG_WARN([Large framebuffer: see s_tritemp.h comments.])])]
+)
+
 dnl
 dnl Gallium Intel configuration
 dnl
diff --git a/docs/dispatch.html b/docs/dispatch.html
index b9ea8822e60..bcab74c7070 100644
--- a/docs/dispatch.html
+++ b/docs/dispatch.html
@@ -198,7 +198,7 @@ few preprocessor defines.</p>
 <ul>
 <li>If <tt>GLX_USE_TLS</tt> is defined, method #4 is used.</li>
 <li>If <tt>PTHREADS</tt> is defined, method #3 is used.</li>
-<li>If any of <tt>PTHREADS</tt>, <tt>USE_XTHREADS</tt>,
+<li>If any of <tt>PTHREADS</tt>,
 <tt>SOLARIS_THREADS</tt>, <tt>WIN32_THREADS</tt>, or <tt>BEOS_THREADS</tt>
 is defined, method #2 is used.</li>
 <li>If none of the preceeding are defined, method #1 is used.</li>
diff --git a/docs/faq.html b/docs/faq.html
index 11b5d432558..65e279aac57 100644
--- a/docs/faq.html
+++ b/docs/faq.html
@@ -316,6 +316,19 @@ Basically, applying a translation of (0.375, 0.375, 0.0) to your coordinates
 will fix the problem.
 </p>
 
+<h2>3.6 How can I change the maximum framebuffer size in Mesa's
+<tt>swrast</tt> backend?</h2>
+<p>
+These can be overridden by using the <tt>--with-max-width</tt> and
+<tt>--with-max-height</tt> options.  The two need not be equal.
+</p><p>
+Do note that Mesa uses these values to size some internal buffers,
+so increasing these sizes will cause Mesa to require additional
+memory.  Furthermore, increasing these limits beyond <tt>4096</tt>
+may introduce rasterization artifacts; see the leading comments in
+<tt>src/mesa/swrast/s_tritemp.h</tt>.
+</p>
+
 <br>
 <br>
 
diff --git a/docs/relnotes-7.5.1.html b/docs/relnotes-7.5.1.html
new file mode 100644
index 00000000000..1da086de3b5
--- /dev/null
+++ b/docs/relnotes-7.5.1.html
@@ -0,0 +1,61 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.5.1 Release Notes / (date TBD)</H1>
+
+<p>
+Mesa 7.5.1 is a bug-fix release fixing issues found since the 7.5 release.
+</p>
+<p>
+The main new feature of Mesa 7.5.x is the
+<a href="http://wiki.freedesktop.org/wiki/Software/gallium"
+target="_parent">Gallium3D</a> infrastructure.
+</p>
+<p>
+Mesa 7.5.1 implements the OpenGL 2.1 API, but the version reported by
+glGetString(GL_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 2.1.
+</p>
+<p>
+See the <a href="install.html">Compiling/Installing page</a> for prerequisites
+for DRI hardware acceleration.
+</p>
+
+
+<h2>MD5 checksums</h2>
+<pre>
+tbd
+</pre>
+
+
+<h2>New features</h2>
+<ul>
+<li>Added configure --with-max-width=W, --with-max-height=H options to specify
+    max framebuffer, viewport size.
+</ul>
+
+
+<h2>Bug fixes</h2>
+<ul>
+<li>Added missing GLEW library to MesaDemos tarballs.
+<li>Fixed swapbuffers jerkiness in Doom3/etc in Intel drivers.
+<li>Fixed front buffer rendering bug in Intel drivers.
+<li>Fixed minor GLX memory leaks.
+<li>Fixed some texture env / fragment program state bugs.
+<li>Fixed some Gallium glBlitFramebuffer() bugs
+</ul>
+
+
+<h2>Changes</h2>
+<ul>
+</ul>
+
+</body>
+</html>
diff --git a/docs/relnotes-7.5.html b/docs/relnotes-7.5.html
index 8bcacd62e1e..56deca6a86c 100644
--- a/docs/relnotes-7.5.html
+++ b/docs/relnotes-7.5.html
@@ -40,7 +40,15 @@ If you're especially concerned with stability you should probably look for
 
 <h2>MD5 checksums</h2>
 <pre>
-tbd
+553fd956e544727f30fbe249619b6286  MesaLib-7.5.tar.gz
+459f332551f6ebb86f384d21dd15e1f0  MesaLib-7.5.tar.bz2
+8c02c0e17a9025250d20424ae32f5163  MesaLib-7.5.zip
+a188da2886fa5496ea0c2cda602b2eeb  MesaDemos-7.5.tar.gz
+398ee8801814a00e47f6c2314e3dfddc  MesaDemos-7.5.tar.bz2
+15a0c8ae013c54335a26335e1a98d609  MesaDemos-7.5.zip
+81010147def5a644ba14f9bbb7a49a2a  MesaGLUT-7.5.tar.gz
+baa7a1e850b6e39bae58868fd0684004  MesaGLUT-7.5.tar.bz2
+265228418e4423fa328f2f5b7970cf08  MesaGLUT-7.5.zip
 </pre>
 
 
diff --git a/docs/relnotes-7.6.html b/docs/relnotes-7.6.html
index 1e7ccf88bf2..691c0f04cae 100644
--- a/docs/relnotes-7.6.html
+++ b/docs/relnotes-7.6.html
@@ -46,6 +46,10 @@ This was written by Zack Rusin at Tungsten Graphics.
 <li>radeon/r200/r300 GL_EXT_framebuffer_object support when used with
     kernel memory manager
 <li>r300 driver support for GL_EXT_vertex_array_bgra, GL_EXT_texture_sRGB
+<li>i915/945 driver support for GL_ARB_point_sprite, GL_EXT_stencil_two_side
+    and GL_ATI_separate_stencil extensions
+<li>Added configure --with-max-width=W, --with-max-height=H options to specify
+    max framebuffer, viewport size.
 </ul>
 
 
diff --git a/docs/relnotes.html b/docs/relnotes.html
index 35a4c7385f7..ddd2e1cdedb 100644
--- a/docs/relnotes.html
+++ b/docs/relnotes.html
@@ -14,6 +14,7 @@ The release notes summarize what's new or changed in each Mesa release.
 
 <UL>
 <LI><A HREF="relnotes-7.6.html">7.6 release notes</A>
+<LI><A HREF="relnotes-7.5.1.html">7.5.1 release notes</A>
 <LI><A HREF="relnotes-7.5.html">7.5 release notes</A>
 <LI><A HREF="relnotes-7.4.4.html">7.4.4 release notes</A>
 <LI><A HREF="relnotes-7.4.3.html">7.4.3 release notes</A>
diff --git a/include/GL/glext.h b/include/GL/glext.h
index 5429287d3fb..24e6c5dc80b 100644
--- a/include/GL/glext.h
+++ b/include/GL/glext.h
@@ -29,9 +29,9 @@ extern "C" {
 */
 
 /* Header file version number, required by OpenGL ABI for Linux */
-/* glext.h last updated $Date: 2009-05-05 15:27:42 -0700 (Tue, 05 May 2009) $ */
+/* glext.h last updated $Date: 2009-08-03 02:13:51 -0700 (Mon, 03 Aug 2009) $ */
 /* Current version at http://www.opengl.org/registry/ */
-#define GL_GLEXT_VERSION 52
+#define GL_GLEXT_VERSION 54
 
 /* Function declaration macros - to move into glplatform.h */
 
@@ -510,6 +510,8 @@ extern "C" {
 #define GL_CLIP_DISTANCE3                 0x3003
 #define GL_CLIP_DISTANCE4                 0x3004
 #define GL_CLIP_DISTANCE5                 0x3005
+#define GL_CLIP_DISTANCE6                 0x3006
+#define GL_CLIP_DISTANCE7                 0x3007
 #define GL_MAX_CLIP_DISTANCES             0x0D32
 #define GL_MAJOR_VERSION                  0x821B
 #define GL_MINOR_VERSION                  0x821C
@@ -600,6 +602,9 @@ extern "C" {
 #define GL_QUERY_NO_WAIT                  0x8E14
 #define GL_QUERY_BY_REGION_WAIT           0x8E15
 #define GL_QUERY_BY_REGION_NO_WAIT        0x8E16
+#define GL_BUFFER_ACCESS_FLAGS            0x911F
+#define GL_BUFFER_MAP_LENGTH              0x9120
+#define GL_BUFFER_MAP_OFFSET              0x9121
 /* Reuse tokens from ARB_depth_buffer_float */
 /* reuse GL_DEPTH_COMPONENT32F */
 /* reuse GL_DEPTH32F_STENCIL8 */
@@ -804,6 +809,84 @@ extern "C" {
 /* reuse GL_INVALID_INDEX */
 #endif
 
+#ifndef GL_VERSION_3_2
+#define GL_CONTEXT_CORE_PROFILE_BIT       0x00000001
+#define GL_CONTEXT_COMPATIBILITY_PROFILE_BIT 0x00000002
+#define GL_LINES_ADJACENCY                0x000A
+#define GL_LINE_STRIP_ADJACENCY           0x000B
+#define GL_TRIANGLES_ADJACENCY            0x000C
+#define GL_TRIANGLE_STRIP_ADJACENCY       0x000D
+#define GL_PROGRAM_POINT_SIZE             0x8642
+#define GL_MAX_GEOMETRY_TEXTURE_IMAGE_UNITS 0x8C29
+#define GL_FRAMEBUFFER_ATTACHMENT_LAYERED 0x8DA7
+#define GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS 0x8DA8
+#define GL_GEOMETRY_SHADER                0x8DD9
+#define GL_GEOMETRY_VERTICES_OUT          0x8916
+#define GL_GEOMETRY_INPUT_TYPE            0x8917
+#define GL_GEOMETRY_OUTPUT_TYPE           0x8918
+#define GL_MAX_GEOMETRY_UNIFORM_COMPONENTS 0x8DDF
+#define GL_MAX_GEOMETRY_OUTPUT_VERTICES   0x8DE0
+#define GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS 0x8DE1
+#define GL_MAX_VERTEX_OUTPUT_COMPONENTS   0x9122
+#define GL_MAX_GEOMETRY_INPUT_COMPONENTS  0x9123
+#define GL_MAX_GEOMETRY_OUTPUT_COMPONENTS 0x9124
+#define GL_MAX_FRAGMENT_INPUT_COMPONENTS  0x9125
+#define GL_CONTEXT_PROFILE_MASK           0x9126
+/* reuse GL_MAX_VARYING_COMPONENTS */
+/* reuse GL_FRAMEBUFFER_ATTACHMENT_TEXTURE_LAYER */
+/* Reuse tokens from ARB_depth_clamp */
+/* reuse GL_DEPTH_CLAMP */
+/* Would reuse tokens from ARB_draw_elements_base_vertex, but it has none */
+/* Would reuse tokens from ARB_fragment_coord_conventions, but it has none */
+/* Reuse tokens from ARB_provoking_vertex */
+/* reuse GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION */
+/* reuse GL_FIRST_VERTEX_CONVENTION */
+/* reuse GL_LAST_VERTEX_CONVENTION */
+/* reuse GL_PROVOKING_VERTEX */
+/* Reuse tokens from ARB_seamless_cube_map */
+/* reuse GL_TEXTURE_CUBE_MAP_SEAMLESS */
+/* Reuse tokens from ARB_sync */
+/* reuse GL_MAX_SERVER_WAIT_TIMEOUT */
+/* reuse GL_OBJECT_TYPE */
+/* reuse GL_SYNC_CONDITION */
+/* reuse GL_SYNC_STATUS */
+/* reuse GL_SYNC_FLAGS */
+/* reuse GL_SYNC_FENCE */
+/* reuse GL_SYNC_GPU_COMMANDS_COMPLETE */
+/* reuse GL_UNSIGNALED */
+/* reuse GL_SIGNALED */
+/* reuse GL_ALREADY_SIGNALED */
+/* reuse GL_TIMEOUT_EXPIRED */
+/* reuse GL_CONDITION_SATISFIED */
+/* reuse GL_WAIT_FAILED */
+/* reuse GL_TIMEOUT_IGNORED */
+/* reuse GL_SYNC_FLUSH_COMMANDS_BIT */
+/* reuse GL_TIMEOUT_IGNORED */
+/* Reuse tokens from ARB_texture_multisample */
+/* reuse GL_SAMPLE_POSITION */
+/* reuse GL_SAMPLE_MASK */
+/* reuse GL_SAMPLE_MASK_VALUE */
+/* reuse GL_MAX_SAMPLE_MASK_WORDS */
+/* reuse GL_TEXTURE_2D_MULTISAMPLE */
+/* reuse GL_PROXY_TEXTURE_2D_MULTISAMPLE */
+/* reuse GL_TEXTURE_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_PROXY_TEXTURE_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_TEXTURE_BINDING_2D_MULTISAMPLE */
+/* reuse GL_TEXTURE_BINDING_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_TEXTURE_SAMPLES */
+/* reuse GL_TEXTURE_FIXED_SAMPLE_LOCATIONS */
+/* reuse GL_SAMPLER_2D_MULTISAMPLE */
+/* reuse GL_INT_SAMPLER_2D_MULTISAMPLE */
+/* reuse GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE */
+/* reuse GL_SAMPLER_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_INT_SAMPLER_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE_ARRAY */
+/* reuse GL_MAX_COLOR_TEXTURE_SAMPLES */
+/* reuse GL_MAX_DEPTH_TEXTURE_SAMPLES */
+/* reuse GL_MAX_INTEGER_SAMPLES */
+/* Don't need to reuse tokens from ARB_vertex_array_bgra since they're already in 1.2 core */
+#endif
+
 #ifndef GL_ARB_multitexture
 #define GL_TEXTURE0_ARB                   0x84C0
 #define GL_TEXTURE1_ARB                   0x84C1
@@ -1530,6 +1613,100 @@ extern "C" {
 #ifndef GL_ARB_shader_texture_lod
 #endif
 
+#ifndef GL_ARB_depth_clamp
+#define GL_DEPTH_CLAMP                    0x864F
+#endif
+
+#ifndef GL_ARB_draw_elements_base_vertex
+#endif
+
+#ifndef GL_ARB_fragment_coord_conventions
+#endif
+
+#ifndef GL_ARB_provoking_vertex
+#define GL_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION 0x8E4C
+#define GL_FIRST_VERTEX_CONVENTION        0x8E4D
+#define GL_LAST_VERTEX_CONVENTION         0x8E4E
+#define GL_PROVOKING_VERTEX               0x8E4F
+#endif
+
+#ifndef GL_ARB_seamless_cube_map
+#define GL_TEXTURE_CUBE_MAP_SEAMLESS      0x884F
+#endif
+
+#ifndef GL_ARB_sync
+#define GL_MAX_SERVER_WAIT_TIMEOUT        0x9111
+#define GL_OBJECT_TYPE                    0x9112
+#define GL_SYNC_CONDITION                 0x9113
+#define GL_SYNC_STATUS                    0x9114
+#define GL_SYNC_FLAGS                     0x9115
+#define GL_SYNC_FENCE                     0x9116
+#define GL_SYNC_GPU_COMMANDS_COMPLETE     0x9117
+#define GL_UNSIGNALED                     0x9118
+#define GL_SIGNALED                       0x9119
+#define GL_ALREADY_SIGNALED               0x911A
+#define GL_TIMEOUT_EXPIRED                0x911B
+#define GL_CONDITION_SATISFIED            0x911C
+#define GL_WAIT_FAILED                    0x911D
+#define GL_SYNC_FLUSH_COMMANDS_BIT        0x00000001
+#define GL_TIMEOUT_IGNORED                0xFFFFFFFFFFFFFFFFull
+#endif
+
+#ifndef GL_ARB_texture_multisample
+#define GL_SAMPLE_POSITION                0x8E50
+#define GL_SAMPLE_MASK                    0x8E51
+#define GL_SAMPLE_MASK_VALUE              0x8E52
+#define GL_MAX_SAMPLE_MASK_WORDS          0x8E59
+#define GL_TEXTURE_2D_MULTISAMPLE         0x9100
+#define GL_PROXY_TEXTURE_2D_MULTISAMPLE   0x9101
+#define GL_TEXTURE_2D_MULTISAMPLE_ARRAY   0x9102
+#define GL_PROXY_TEXTURE_2D_MULTISAMPLE_ARRAY 0x9103
+#define GL_TEXTURE_BINDING_2D_MULTISAMPLE 0x9104
+#define GL_TEXTURE_BINDING_2D_MULTISAMPLE_ARRAY 0x9105
+#define GL_TEXTURE_SAMPLES                0x9106
+#define GL_TEXTURE_FIXED_SAMPLE_LOCATIONS 0x9107
+#define GL_SAMPLER_2D_MULTISAMPLE         0x9108
+#define GL_INT_SAMPLER_2D_MULTISAMPLE     0x9109
+#define GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE 0x910A
+#define GL_SAMPLER_2D_MULTISAMPLE_ARRAY   0x910B
+#define GL_INT_SAMPLER_2D_MULTISAMPLE_ARRAY 0x910C
+#define GL_UNSIGNED_INT_SAMPLER_2D_MULTISAMPLE_ARRAY 0x910D
+#define GL_MAX_COLOR_TEXTURE_SAMPLES      0x910E
+#define GL_MAX_DEPTH_TEXTURE_SAMPLES      0x910F
+#define GL_MAX_INTEGER_SAMPLES            0x9110
+#endif
+
+#ifndef GL_ARB_vertex_array_bgra
+/* reuse GL_BGRA */
+#endif
+
+#ifndef GL_ARB_draw_buffers_blend
+#endif
+
+#ifndef GL_ARB_sample_shading
+#define GL_SAMPLE_SHADING                 0x8C36
+#define GL_MIN_SAMPLE_SHADING_VALUE       0x8C37
+#endif
+
+#ifndef GL_ARB_texture_cube_map_array
+#define GL_TEXTURE_CUBE_MAP_ARRAY         0x9009
+#define GL_TEXTURE_BINDING_CUBE_MAP_ARRAY 0x900A
+#define GL_PROXY_TEXTURE_CUBE_MAP_ARRAY   0x900B
+#define GL_SAMPLER_CUBE_MAP_ARRAY         0x900C
+#define GL_SAMPLER_CUBE_MAP_ARRAY_SHADOW  0x900D
+#define GL_INT_SAMPLER_CUBE_MAP_ARRAY     0x900E
+#define GL_UNSIGNED_INT_SAMPLER_CUBE_MAP_ARRAY 0x900F
+#endif
+
+#ifndef GL_ARB_texture_gather
+#define GL_MIN_PROGRAM_TEXTURE_GATHER_OFFSET 0x8E5E
+#define GL_MAX_PROGRAM_TEXTURE_GATHER_OFFSET 0x8E5F
+#define GL_MAX_PROGRAM_TEXTURE_GATHER_COMPONENTS 0x8F9F
+#endif
+
+#ifndef GL_ARB_texture_query_lod
+#endif
+
 #ifndef GL_EXT_abgr
 #define GL_ABGR_EXT                       0x8000
 #endif
@@ -3996,11 +4173,11 @@ extern "C" {
 #define GL_SAMPLE_MASK_VALUE_NV           0x8E52
 #define GL_TEXTURE_BINDING_RENDERBUFFER_NV 0x8E53
 #define GL_TEXTURE_RENDERBUFFER_DATA_STORE_BINDING_NV 0x8E54
-#define GL_MAX_SAMPLE_MASK_WORDS_NV       0x8E59
 #define GL_TEXTURE_RENDERBUFFER_NV        0x8E55
 #define GL_SAMPLER_RENDERBUFFER_NV        0x8E56
 #define GL_INT_SAMPLER_RENDERBUFFER_NV    0x8E57
 #define GL_UNSIGNED_INT_SAMPLER_RENDERBUFFER_NV 0x8E58
+#define GL_MAX_SAMPLE_MASK_WORDS_NV       0x8E59
 #endif
 
 #ifndef GL_NV_transform_feedback2
@@ -4046,6 +4223,94 @@ extern "C" {
 #define GL_PROVOKING_VERTEX_EXT           0x8E4F
 #endif
 
+#ifndef GL_EXT_texture_snorm
+#define GL_ALPHA_SNORM                    0x9010
+#define GL_LUMINANCE_SNORM                0x9011
+#define GL_LUMINANCE_ALPHA_SNORM          0x9012
+#define GL_INTENSITY_SNORM                0x9013
+#define GL_ALPHA8_SNORM                   0x9014
+#define GL_LUMINANCE8_SNORM               0x9015
+#define GL_LUMINANCE8_ALPHA8_SNORM        0x9016
+#define GL_INTENSITY8_SNORM               0x9017
+#define GL_ALPHA16_SNORM                  0x9018
+#define GL_LUMINANCE16_SNORM              0x9019
+#define GL_LUMINANCE16_ALPHA16_SNORM      0x901A
+#define GL_INTENSITY16_SNORM              0x901B
+/* reuse GL_R_SNORM */
+/* reuse GL_RG_SNORM */
+/* reuse GL_RGB_SNORM */
+/* reuse GL_RGBA_SNORM */
+/* reuse GL_R8_SNORM */
+/* reuse GL_RG8_SNORM */
+/* reuse GL_RGB8_SNORM */
+/* reuse GL_RGBA8_SNORM */
+/* reuse GL_R16_SNORM */
+/* reuse GL_RG16_SNORM */
+/* reuse GL_RGB16_SNORM */
+/* reuse GL_RGBA16_SNORM */
+/* reuse GL_SIGNED_NORMALIZED */
+#endif
+
+#ifndef GL_AMD_draw_buffers_blend
+#endif
+
+#ifndef GL_APPLE_texture_range
+#define GL_TEXTURE_RANGE_LENGTH_APPLE     0x85B7
+#define GL_TEXTURE_RANGE_POINTER_APPLE    0x85B8
+#define GL_TEXTURE_STORAGE_HINT_APPLE     0x85BC
+#define GL_STORAGE_PRIVATE_APPLE          0x85BD
+/* reuse GL_STORAGE_CACHED_APPLE */
+/* reuse GL_STORAGE_SHARED_APPLE */
+#endif
+
+#ifndef GL_APPLE_float_pixels
+#define GL_HALF_APPLE                     0x140B
+#define GL_RGBA_FLOAT32_APPLE             0x8814
+#define GL_RGB_FLOAT32_APPLE              0x8815
+#define GL_ALPHA_FLOAT32_APPLE            0x8816
+#define GL_INTENSITY_FLOAT32_APPLE        0x8817
+#define GL_LUMINANCE_FLOAT32_APPLE        0x8818
+#define GL_LUMINANCE_ALPHA_FLOAT32_APPLE  0x8819
+#define GL_RGBA_FLOAT16_APPLE             0x881A
+#define GL_RGB_FLOAT16_APPLE              0x881B
+#define GL_ALPHA_FLOAT16_APPLE            0x881C
+#define GL_INTENSITY_FLOAT16_APPLE        0x881D
+#define GL_LUMINANCE_FLOAT16_APPLE        0x881E
+#define GL_LUMINANCE_ALPHA_FLOAT16_APPLE  0x881F
+#define GL_COLOR_FLOAT_APPLE              0x8A0F
+#endif
+
+#ifndef GL_APPLE_vertex_program_evaluators
+#define GL_VERTEX_ATTRIB_MAP1_APPLE       0x8A00
+#define GL_VERTEX_ATTRIB_MAP2_APPLE       0x8A01
+#define GL_VERTEX_ATTRIB_MAP1_SIZE_APPLE  0x8A02
+#define GL_VERTEX_ATTRIB_MAP1_COEFF_APPLE 0x8A03
+#define GL_VERTEX_ATTRIB_MAP1_ORDER_APPLE 0x8A04
+#define GL_VERTEX_ATTRIB_MAP1_DOMAIN_APPLE 0x8A05
+#define GL_VERTEX_ATTRIB_MAP2_SIZE_APPLE  0x8A06
+#define GL_VERTEX_ATTRIB_MAP2_COEFF_APPLE 0x8A07
+#define GL_VERTEX_ATTRIB_MAP2_ORDER_APPLE 0x8A08
+#define GL_VERTEX_ATTRIB_MAP2_DOMAIN_APPLE 0x8A09
+#endif
+
+#ifndef GL_APPLE_aux_depth_stencil
+#define GL_AUX_DEPTH_STENCIL_APPLE        0x8A14
+#endif
+
+#ifndef GL_APPLE_object_purgeable
+#define GL_BUFFER_OBJECT_APPLE            0x85B3
+#define GL_RELEASED_APPLE                 0x8A19
+#define GL_VOLATILE_APPLE                 0x8A1A
+#define GL_RETAINED_APPLE                 0x8A1B
+#define GL_UNDEFINED_APPLE                0x8A1C
+#define GL_PURGEABLE_APPLE                0x8A1D
+#endif
+
+#ifndef GL_APPLE_row_bytes
+#define GL_PACK_ROW_BYTES_APPLE           0x8A15
+#define GL_UNPACK_ROW_BYTES_APPLE         0x8A16
+#endif
+
 
 /*************************************************************/
 
@@ -4125,6 +4390,12 @@ typedef int64_t GLint64EXT;
 typedef uint64_t GLuint64EXT;
 #endif
 
+#ifndef ARB_sync
+typedef int64_t GLint64;
+typedef uint64_t GLuint64;
+typedef struct __GLsync *GLsync;
+#endif
+
 #ifndef GL_VERSION_1_2
 #define GL_VERSION_1_2 1
 #ifdef GL_GLEXT_PROTOTYPES
@@ -4698,6 +4969,26 @@ GLAPI void APIENTRY glEndConditionalRender (void);
 GLAPI void APIENTRY glVertexAttribIPointer (GLuint, GLint, GLenum, GLsizei, const GLvoid *);
 GLAPI void APIENTRY glGetVertexAttribIiv (GLuint, GLenum, GLint *);
 GLAPI void APIENTRY glGetVertexAttribIuiv (GLuint, GLenum, GLuint *);
+GLAPI void APIENTRY glVertexAttribI1i (GLuint, GLint);
+GLAPI void APIENTRY glVertexAttribI2i (GLuint, GLint, GLint);
+GLAPI void APIENTRY glVertexAttribI3i (GLuint, GLint, GLint, GLint);
+GLAPI void APIENTRY glVertexAttribI4i (GLuint, GLint, GLint, GLint, GLint);
+GLAPI void APIENTRY glVertexAttribI1ui (GLuint, GLuint);
+GLAPI void APIENTRY glVertexAttribI2ui (GLuint, GLuint, GLuint);
+GLAPI void APIENTRY glVertexAttribI3ui (GLuint, GLuint, GLuint, GLuint);
+GLAPI void APIENTRY glVertexAttribI4ui (GLuint, GLuint, GLuint, GLuint, GLuint);
+GLAPI void APIENTRY glVertexAttribI1iv (GLuint, const GLint *);
+GLAPI void APIENTRY glVertexAttribI2iv (GLuint, const GLint *);
+GLAPI void APIENTRY glVertexAttribI3iv (GLuint, const GLint *);
+GLAPI void APIENTRY glVertexAttribI4iv (GLuint, const GLint *);
+GLAPI void APIENTRY glVertexAttribI1uiv (GLuint, const GLuint *);
+GLAPI void APIENTRY glVertexAttribI2uiv (GLuint, const GLuint *);
+GLAPI void APIENTRY glVertexAttribI3uiv (GLuint, const GLuint *);
+GLAPI void APIENTRY glVertexAttribI4uiv (GLuint, const GLuint *);
+GLAPI void APIENTRY glVertexAttribI4bv (GLuint, const GLbyte *);
+GLAPI void APIENTRY glVertexAttribI4sv (GLuint, const GLshort *);
+GLAPI void APIENTRY glVertexAttribI4ubv (GLuint, const GLubyte *);
+GLAPI void APIENTRY glVertexAttribI4usv (GLuint, const GLushort *);
 GLAPI void APIENTRY glGetUniformuiv (GLuint, GLint, GLuint *);
 GLAPI void APIENTRY glBindFragDataLocation (GLuint, GLuint, const GLchar *);
 GLAPI GLint APIENTRY glGetFragDataLocation (GLuint, const GLchar *);
@@ -4737,52 +5028,6 @@ typedef void (APIENTRYP PFNGLENDCONDITIONALRENDERPROC) (void);
 typedef void (APIENTRYP PFNGLVERTEXATTRIBIPOINTERPROC) (GLuint index, GLint size, GLenum type, GLsizei stride, const GLvoid *pointer);
 typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIIVPROC) (GLuint index, GLenum pname, GLint *params);
 typedef void (APIENTRYP PFNGLGETVERTEXATTRIBIUIVPROC) (GLuint index, GLenum pname, GLuint *params);
-typedef void (APIENTRYP PFNGLGETUNIFORMUIVPROC) (GLuint program, GLint location, GLuint *params);
-typedef void (APIENTRYP PFNGLBINDFRAGDATALOCATIONPROC) (GLuint program, GLuint color, const GLchar *name);
-typedef GLint (APIENTRYP PFNGLGETFRAGDATALOCATIONPROC) (GLuint program, const GLchar *name);
-typedef void (APIENTRYP PFNGLUNIFORM1UIPROC) (GLint location, GLuint v0);
-typedef void (APIENTRYP PFNGLUNIFORM2UIPROC) (GLint location, GLuint v0, GLuint v1);
-typedef void (APIENTRYP PFNGLUNIFORM3UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2);
-typedef void (APIENTRYP PFNGLUNIFORM4UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3);
-typedef void (APIENTRYP PFNGLUNIFORM1UIVPROC) (GLint location, GLsizei count, const GLuint *value);
-typedef void (APIENTRYP PFNGLUNIFORM2UIVPROC) (GLint location, GLsizei count, const GLuint *value);
-typedef void (APIENTRYP PFNGLUNIFORM3UIVPROC) (GLint location, GLsizei count, const GLuint *value);
-typedef void (APIENTRYP PFNGLUNIFORM4UIVPROC) (GLint location, GLsizei count, const GLuint *value);
-typedef void (APIENTRYP PFNGLTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, const GLint *params);
-typedef void (APIENTRYP PFNGLTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, const GLuint *params);
-typedef void (APIENTRYP PFNGLGETTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, GLint *params);
-typedef void (APIENTRYP PFNGLGETTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, GLuint *params);
-typedef void (APIENTRYP PFNGLCLEARBUFFERIVPROC) (GLenum buffer, GLint drawbuffer, const GLint *value);
-typedef void (APIENTRYP PFNGLCLEARBUFFERUIVPROC) (GLenum buffer, GLint drawbuffer, const GLuint *value);
-typedef void (APIENTRYP PFNGLCLEARBUFFERFVPROC) (GLenum buffer, GLint drawbuffer, const GLfloat *value);
-typedef void (APIENTRYP PFNGLCLEARBUFFERFIPROC) (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);
-typedef const GLubyte * (APIENTRYP PFNGLGETSTRINGIPROC) (GLenum name, GLuint index);
-#endif
-
-#ifndef GL_VERSION_3_0_DEPRECATED
-#define GL_VERSION_3_0_DEPRECATED 1
-#ifdef GL_GLEXT_PROTOTYPES
-GLAPI void APIENTRY glVertexAttribI1i (GLuint, GLint);
-GLAPI void APIENTRY glVertexAttribI2i (GLuint, GLint, GLint);
-GLAPI void APIENTRY glVertexAttribI3i (GLuint, GLint, GLint, GLint);
-GLAPI void APIENTRY glVertexAttribI4i (GLuint, GLint, GLint, GLint, GLint);
-GLAPI void APIENTRY glVertexAttribI1ui (GLuint, GLuint);
-GLAPI void APIENTRY glVertexAttribI2ui (GLuint, GLuint, GLuint);
-GLAPI void APIENTRY glVertexAttribI3ui (GLuint, GLuint, GLuint, GLuint);
-GLAPI void APIENTRY glVertexAttribI4ui (GLuint, GLuint, GLuint, GLuint, GLuint);
-GLAPI void APIENTRY glVertexAttribI1iv (GLuint, const GLint *);
-GLAPI void APIENTRY glVertexAttribI2iv (GLuint, const GLint *);
-GLAPI void APIENTRY glVertexAttribI3iv (GLuint, const GLint *);
-GLAPI void APIENTRY glVertexAttribI4iv (GLuint, const GLint *);
-GLAPI void APIENTRY glVertexAttribI1uiv (GLuint, const GLuint *);
-GLAPI void APIENTRY glVertexAttribI2uiv (GLuint, const GLuint *);
-GLAPI void APIENTRY glVertexAttribI3uiv (GLuint, const GLuint *);
-GLAPI void APIENTRY glVertexAttribI4uiv (GLuint, const GLuint *);
-GLAPI void APIENTRY glVertexAttribI4bv (GLuint, const GLbyte *);
-GLAPI void APIENTRY glVertexAttribI4sv (GLuint, const GLshort *);
-GLAPI void APIENTRY glVertexAttribI4ubv (GLuint, const GLubyte *);
-GLAPI void APIENTRY glVertexAttribI4usv (GLuint, const GLushort *);
-#endif /* GL_GLEXT_PROTOTYPES */
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI1IPROC) (GLuint index, GLint x);
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI2IPROC) (GLuint index, GLint x, GLint y);
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI3IPROC) (GLuint index, GLint x, GLint y, GLint z);
@@ -4803,6 +5048,26 @@ typedef void (APIENTRYP PFNGLVERTEXATTRIBI4BVPROC) (GLuint index, const GLbyte *
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI4SVPROC) (GLuint index, const GLshort *v);
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI4UBVPROC) (GLuint index, const GLubyte *v);
 typedef void (APIENTRYP PFNGLVERTEXATTRIBI4USVPROC) (GLuint index, const GLushort *v);
+typedef void (APIENTRYP PFNGLGETUNIFORMUIVPROC) (GLuint program, GLint location, GLuint *params);
+typedef void (APIENTRYP PFNGLBINDFRAGDATALOCATIONPROC) (GLuint program, GLuint color, const GLchar *name);
+typedef GLint (APIENTRYP PFNGLGETFRAGDATALOCATIONPROC) (GLuint program, const GLchar *name);
+typedef void (APIENTRYP PFNGLUNIFORM1UIPROC) (GLint location, GLuint v0);
+typedef void (APIENTRYP PFNGLUNIFORM2UIPROC) (GLint location, GLuint v0, GLuint v1);
+typedef void (APIENTRYP PFNGLUNIFORM3UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2);
+typedef void (APIENTRYP PFNGLUNIFORM4UIPROC) (GLint location, GLuint v0, GLuint v1, GLuint v2, GLuint v3);
+typedef void (APIENTRYP PFNGLUNIFORM1UIVPROC) (GLint location, GLsizei count, const GLuint *value);
+typedef void (APIENTRYP PFNGLUNIFORM2UIVPROC) (GLint location, GLsizei count, const GLuint *value);
+typedef void (APIENTRYP PFNGLUNIFORM3UIVPROC) (GLint location, GLsizei count, const GLuint *value);
+typedef void (APIENTRYP PFNGLUNIFORM4UIVPROC) (GLint location, GLsizei count, const GLuint *value);
+typedef void (APIENTRYP PFNGLTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, const GLint *params);
+typedef void (APIENTRYP PFNGLTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, const GLuint *params);
+typedef void (APIENTRYP PFNGLGETTEXPARAMETERIIVPROC) (GLenum target, GLenum pname, GLint *params);
+typedef void (APIENTRYP PFNGLGETTEXPARAMETERIUIVPROC) (GLenum target, GLenum pname, GLuint *params);
+typedef void (APIENTRYP PFNGLCLEARBUFFERIVPROC) (GLenum buffer, GLint drawbuffer, const GLint *value);
+typedef void (APIENTRYP PFNGLCLEARBUFFERUIVPROC) (GLenum buffer, GLint drawbuffer, const GLuint *value);
+typedef void (APIENTRYP PFNGLCLEARBUFFERFVPROC) (GLenum buffer, GLint drawbuffer, const GLfloat *value);
+typedef void (APIENTRYP PFNGLCLEARBUFFERFIPROC) (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil);
+typedef const GLubyte * (APIENTRYP PFNGLGETSTRINGIPROC) (GLenum name, GLuint index);
 #endif
 
 #ifndef GL_VERSION_3_1
@@ -4822,6 +5087,27 @@ typedef void (APIENTRYP PFNGLTEXBUFFERPROC) (GLenum target, GLenum internalforma
 typedef void (APIENTRYP PFNGLPRIMITIVERESTARTINDEXPROC) (GLuint index);
 #endif
 
+#ifndef GL_VERSION_3_2
+#define GL_VERSION_3_2 1
+/* OpenGL 3.2 also reuses entry points from these extensions: */
+/* ARB_draw_elements_base_vertex */
+/* ARB_provoking_vertex */
+/* ARB_sync */
+/* ARB_texture_multisample */
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glGetInteger64i_v (GLenum, GLuint, GLint64 *);
+GLAPI void APIENTRY glGetBufferParameteri64v (GLenum, GLenum, GLint64 *);
+GLAPI void APIENTRY glProgramParameteri (GLuint, GLenum, GLint);
+GLAPI void APIENTRY glFramebufferTexture (GLenum, GLenum, GLuint, GLint);
+GLAPI void APIENTRY glFramebufferTextureFace (GLenum, GLenum, GLuint, GLint, GLenum);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLGETINTEGER64I_VPROC) (GLenum target, GLuint index, GLint64 *data);
+typedef void (APIENTRYP PFNGLGETBUFFERPARAMETERI64VPROC) (GLenum target, GLenum pname, GLint64 *params);
+typedef void (APIENTRYP PFNGLPROGRAMPARAMETERIPROC) (GLuint program, GLenum pname, GLint value);
+typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level);
+typedef void (APIENTRYP PFNGLFRAMEBUFFERTEXTUREFACEPROC) (GLenum target, GLenum attachment, GLuint texture, GLint level, GLenum face);
+#endif
+
 #ifndef GL_ARB_multitexture
 #define GL_ARB_multitexture 1
 #ifdef GL_GLEXT_PROTOTYPES
@@ -5567,6 +5853,112 @@ typedef void (APIENTRYP PFNGLCOPYBUFFERSUBDATAPROC) (GLenum readTarget, GLenum w
 #define GL_ARB_shader_texture_lod 1
 #endif
 
+#ifndef GL_ARB_depth_clamp
+#define GL_ARB_depth_clamp 1
+#endif
+
+#ifndef GL_ARB_draw_elements_base_vertex
+#define GL_ARB_draw_elements_base_vertex 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glDrawElementsBaseVertex (GLenum, GLsizei, GLenum, const GLvoid *, GLint);
+GLAPI void APIENTRY glDrawRangeElementsBaseVertex (GLenum, GLuint, GLuint, GLsizei, GLenum, const GLvoid *, GLint);
+GLAPI void APIENTRY glDrawElementsInstancedBaseVertex (GLenum, GLsizei, GLenum, const GLvoid *, GLsizei, GLint);
+GLAPI void APIENTRY glMultiDrawElementsBaseVertex (GLenum, const GLsizei *, GLenum, const GLvoid* *, GLsizei, const GLint *);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLDRAWELEMENTSBASEVERTEXPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLint basevertex);
+typedef void (APIENTRYP PFNGLDRAWRANGEELEMENTSBASEVERTEXPROC) (GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const GLvoid *indices, GLint basevertex);
+typedef void (APIENTRYP PFNGLDRAWELEMENTSINSTANCEDBASEVERTEXPROC) (GLenum mode, GLsizei count, GLenum type, const GLvoid *indices, GLsizei primcount, GLint basevertex);
+typedef void (APIENTRYP PFNGLMULTIDRAWELEMENTSBASEVERTEXPROC) (GLenum mode, const GLsizei *count, GLenum type, const GLvoid* *indices, GLsizei primcount, const GLint *basevertex);
+#endif
+
+#ifndef GL_ARB_fragment_coord_conventions
+#define GL_ARB_fragment_coord_conventions 1
+#endif
+
+#ifndef GL_ARB_provoking_vertex
+#define GL_ARB_provoking_vertex 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glProvokingVertex (GLenum);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLPROVOKINGVERTEXPROC) (GLenum mode);
+#endif
+
+#ifndef GL_ARB_seamless_cube_map
+#define GL_ARB_seamless_cube_map 1
+#endif
+
+#ifndef GL_ARB_sync
+#define GL_ARB_sync 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI GLsync APIENTRY glFenceSync (GLenum, GLbitfield);
+GLAPI GLboolean APIENTRY glIsSync (GLsync);
+GLAPI void APIENTRY glDeleteSync (GLsync);
+GLAPI GLenum APIENTRY glClientWaitSync (GLsync, GLbitfield, GLuint64);
+GLAPI void APIENTRY glWaitSync (GLsync, GLbitfield, GLuint64);
+GLAPI void APIENTRY glGetInteger64v (GLenum, GLint64 *);
+GLAPI void APIENTRY glGetSynciv (GLsync, GLenum, GLsizei, GLsizei *, GLint *);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef GLsync (APIENTRYP PFNGLFENCESYNCPROC) (GLenum condition, GLbitfield flags);
+typedef GLboolean (APIENTRYP PFNGLISSYNCPROC) (GLsync sync);
+typedef void (APIENTRYP PFNGLDELETESYNCPROC) (GLsync sync);
+typedef GLenum (APIENTRYP PFNGLCLIENTWAITSYNCPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);
+typedef void (APIENTRYP PFNGLWAITSYNCPROC) (GLsync sync, GLbitfield flags, GLuint64 timeout);
+typedef void (APIENTRYP PFNGLGETINTEGER64VPROC) (GLenum pname, GLint64 *params);
+typedef void (APIENTRYP PFNGLGETSYNCIVPROC) (GLsync sync, GLenum pname, GLsizei bufSize, GLsizei *length, GLint *values);
+#endif
+
+#ifndef GL_ARB_texture_multisample
+#define GL_ARB_texture_multisample 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glTexImage2DMultisample (GLenum, GLsizei, GLint, GLsizei, GLsizei, GLboolean);
+GLAPI void APIENTRY glTexImage3DMultisample (GLenum, GLsizei, GLint, GLsizei, GLsizei, GLsizei, GLboolean);
+GLAPI void APIENTRY glGetMultisamplefv (GLenum, GLuint, GLfloat *);
+GLAPI void APIENTRY glSampleMaski (GLuint, GLbitfield);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLTEXIMAGE2DMULTISAMPLEPROC) (GLenum target, GLsizei samples, GLint internalformat, GLsizei width, GLsizei height, GLboolean fixedsamplelocations);
+typedef void (APIENTRYP PFNGLTEXIMAGE3DMULTISAMPLEPROC) (GLenum target, GLsizei samples, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations);
+typedef void (APIENTRYP PFNGLGETMULTISAMPLEFVPROC) (GLenum pname, GLuint index, GLfloat *val);
+typedef void (APIENTRYP PFNGLSAMPLEMASKIPROC) (GLuint index, GLbitfield mask);
+#endif
+
+#ifndef GL_ARB_vertex_array_bgra
+#define GL_ARB_vertex_array_bgra 1
+#endif
+
+#ifndef GL_ARB_draw_buffers_blend
+#define GL_ARB_draw_buffers_blend 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glBlendEquationi (GLuint, GLenum);
+GLAPI void APIENTRY glBlendEquationSeparatei (GLuint, GLenum, GLenum);
+GLAPI void APIENTRY glBlendFunci (GLuint, GLenum, GLenum);
+GLAPI void APIENTRY glBlendFuncSeparatei (GLuint, GLenum, GLenum, GLenum, GLenum);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLBLENDEQUATIONIPROC) (GLuint buf, GLenum mode);
+typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEIPROC) (GLuint buf, GLenum modeRGB, GLenum modeAlpha);
+typedef void (APIENTRYP PFNGLBLENDFUNCIPROC) (GLuint buf, GLenum src, GLenum dst);
+typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEIPROC) (GLuint buf, GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);
+#endif
+
+#ifndef GL_ARB_sample_shading
+#define GL_ARB_sample_shading 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glMinSampleShading (GLclampf);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLMINSAMPLESHADINGPROC) (GLclampf value);
+#endif
+
+#ifndef GL_ARB_texture_cube_map_array
+#define GL_ARB_texture_cube_map_array 1
+#endif
+
+#ifndef GL_ARB_texture_gather
+#define GL_ARB_texture_gather 1
+#endif
+
+#ifndef GL_ARB_texture_query_lod
+#define GL_ARB_texture_query_lod 1
+#endif
+
 #ifndef GL_EXT_abgr
 #define GL_EXT_abgr 1
 #endif
@@ -8204,7 +8596,7 @@ GLAPI void APIENTRY glTransformFeedbackAttribsNV (GLuint, const GLint *, GLenum)
 GLAPI void APIENTRY glBindBufferRangeNV (GLenum, GLuint, GLuint, GLintptr, GLsizeiptr);
 GLAPI void APIENTRY glBindBufferOffsetNV (GLenum, GLuint, GLuint, GLintptr);
 GLAPI void APIENTRY glBindBufferBaseNV (GLenum, GLuint, GLuint);
-GLAPI void APIENTRY glTransformFeedbackVaryingsNV (GLuint, GLsizei, const GLint *, GLenum);
+GLAPI void APIENTRY glTransformFeedbackVaryingsNV (GLuint, GLsizei, const GLchar* *, GLenum);
 GLAPI void APIENTRY glActiveVaryingNV (GLuint, const GLchar *);
 GLAPI GLint APIENTRY glGetVaryingLocationNV (GLuint, const GLchar *);
 GLAPI void APIENTRY glGetActiveVaryingNV (GLuint, GLuint, GLsizei, GLsizei *, GLsizei *, GLenum *, GLchar *);
@@ -8216,7 +8608,7 @@ typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKATTRIBSNVPROC) (GLuint count, cons
 typedef void (APIENTRYP PFNGLBINDBUFFERRANGENVPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size);
 typedef void (APIENTRYP PFNGLBINDBUFFEROFFSETNVPROC) (GLenum target, GLuint index, GLuint buffer, GLintptr offset);
 typedef void (APIENTRYP PFNGLBINDBUFFERBASENVPROC) (GLenum target, GLuint index, GLuint buffer);
-typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKVARYINGSNVPROC) (GLuint program, GLsizei count, const GLint *locations, GLenum bufferMode);
+typedef void (APIENTRYP PFNGLTRANSFORMFEEDBACKVARYINGSNVPROC) (GLuint program, GLsizei count, const GLchar* *varyings, GLenum bufferMode);
 typedef void (APIENTRYP PFNGLACTIVEVARYINGNVPROC) (GLuint program, const GLchar *name);
 typedef GLint (APIENTRYP PFNGLGETVARYINGLOCATIONNVPROC) (GLuint program, const GLchar *name);
 typedef void (APIENTRYP PFNGLGETACTIVEVARYINGNVPROC) (GLuint program, GLuint index, GLsizei bufSize, GLsizei *length, GLsizei *size, GLenum *type, GLchar *name);
@@ -8781,6 +9173,78 @@ GLAPI void APIENTRY glProvokingVertexEXT (GLenum);
 typedef void (APIENTRYP PFNGLPROVOKINGVERTEXEXTPROC) (GLenum mode);
 #endif
 
+#ifndef GL_EXT_texture_snorm
+#define GL_EXT_texture_snorm 1
+#endif
+
+#ifndef GL_AMD_draw_buffers_blend
+#define GL_AMD_draw_buffers_blend 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glBlendFuncIndexedAMD (GLuint, GLenum, GLenum);
+GLAPI void APIENTRY glBlendFuncSeparateIndexedAMD (GLuint, GLenum, GLenum, GLenum, GLenum);
+GLAPI void APIENTRY glBlendEquationIndexedAMD (GLuint, GLenum);
+GLAPI void APIENTRY glBlendEquationSeparateIndexedAMD (GLuint, GLenum, GLenum);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLBLENDFUNCINDEXEDAMDPROC) (GLuint buf, GLenum src, GLenum dst);
+typedef void (APIENTRYP PFNGLBLENDFUNCSEPARATEINDEXEDAMDPROC) (GLuint buf, GLenum srcRGB, GLenum dstRGB, GLenum srcAlpha, GLenum dstAlpha);
+typedef void (APIENTRYP PFNGLBLENDEQUATIONINDEXEDAMDPROC) (GLuint buf, GLenum mode);
+typedef void (APIENTRYP PFNGLBLENDEQUATIONSEPARATEINDEXEDAMDPROC) (GLuint buf, GLenum modeRGB, GLenum modeAlpha);
+#endif
+
+#ifndef GL_APPLE_texture_range
+#define GL_APPLE_texture_range 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glTextureRangeAPPLE (GLenum, GLsizei, const GLvoid *);
+GLAPI void APIENTRY glGetTexParameterPointervAPPLE (GLenum, GLenum, GLvoid* *);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLTEXTURERANGEAPPLEPROC) (GLenum target, GLsizei length, const GLvoid *pointer);
+typedef void (APIENTRYP PFNGLGETTEXPARAMETERPOINTERVAPPLEPROC) (GLenum target, GLenum pname, GLvoid* *params);
+#endif
+
+#ifndef GL_APPLE_float_pixels
+#define GL_APPLE_float_pixels 1
+#endif
+
+#ifndef GL_APPLE_vertex_program_evaluators
+#define GL_APPLE_vertex_program_evaluators 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI void APIENTRY glEnableVertexAttribAPPLE (GLuint, GLenum);
+GLAPI void APIENTRY glDisableVertexAttribAPPLE (GLuint, GLenum);
+GLAPI GLboolean APIENTRY glIsVertexAttribEnabledAPPLE (GLuint, GLenum);
+GLAPI void APIENTRY glMapVertexAttrib1dAPPLE (GLuint, GLuint, GLdouble, GLdouble, GLint, GLint, const GLdouble *);
+GLAPI void APIENTRY glMapVertexAttrib1fAPPLE (GLuint, GLuint, GLfloat, GLfloat, GLint, GLint, const GLfloat *);
+GLAPI void APIENTRY glMapVertexAttrib2dAPPLE (GLuint, GLuint, GLdouble, GLdouble, GLint, GLint, GLdouble, GLdouble, GLint, GLint, const GLdouble *);
+GLAPI void APIENTRY glMapVertexAttrib2fAPPLE (GLuint, GLuint, GLfloat, GLfloat, GLint, GLint, GLfloat, GLfloat, GLint, GLint, const GLfloat *);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef void (APIENTRYP PFNGLENABLEVERTEXATTRIBAPPLEPROC) (GLuint index, GLenum pname);
+typedef void (APIENTRYP PFNGLDISABLEVERTEXATTRIBAPPLEPROC) (GLuint index, GLenum pname);
+typedef GLboolean (APIENTRYP PFNGLISVERTEXATTRIBENABLEDAPPLEPROC) (GLuint index, GLenum pname);
+typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB1DAPPLEPROC) (GLuint index, GLuint size, GLdouble u1, GLdouble u2, GLint stride, GLint order, const GLdouble *points);
+typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB1FAPPLEPROC) (GLuint index, GLuint size, GLfloat u1, GLfloat u2, GLint stride, GLint order, const GLfloat *points);
+typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB2DAPPLEPROC) (GLuint index, GLuint size, GLdouble u1, GLdouble u2, GLint ustride, GLint uorder, GLdouble v1, GLdouble v2, GLint vstride, GLint vorder, const GLdouble *points);
+typedef void (APIENTRYP PFNGLMAPVERTEXATTRIB2FAPPLEPROC) (GLuint index, GLuint size, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, GLfloat v1, GLfloat v2, GLint vstride, GLint vorder, const GLfloat *points);
+#endif
+
+#ifndef GL_APPLE_aux_depth_stencil
+#define GL_APPLE_aux_depth_stencil 1
+#endif
+
+#ifndef GL_APPLE_object_purgeable
+#define GL_APPLE_object_purgeable 1
+#ifdef GL_GLEXT_PROTOTYPES
+GLAPI GLenum APIENTRY glObjectPurgeableAPPLE (GLenum, GLuint, GLenum);
+GLAPI GLenum APIENTRY glObjectUnpurgeableAPPLE (GLenum, GLuint, GLenum);
+GLAPI void APIENTRY glGetObjectParameterivAPPLE (GLenum, GLuint, GLenum, GLint *);
+#endif /* GL_GLEXT_PROTOTYPES */
+typedef GLenum (APIENTRYP PFNGLOBJECTPURGEABLEAPPLEPROC) (GLenum objectType, GLuint name, GLenum option);
+typedef GLenum (APIENTRYP PFNGLOBJECTUNPURGEABLEAPPLEPROC) (GLenum objectType, GLuint name, GLenum option);
+typedef void (APIENTRYP PFNGLGETOBJECTPARAMETERIVAPPLEPROC) (GLenum objectType, GLuint name, GLenum pname, GLint *params);
+#endif
+
+#ifndef GL_APPLE_row_bytes
+#define GL_APPLE_row_bytes 1
+#endif
+
 
 #ifdef __cplusplus
 }
diff --git a/include/GL/glxext.h b/include/GL/glxext.h
index 536fb25a76e..eac09f94fa3 100644
--- a/include/GL/glxext.h
+++ b/include/GL/glxext.h
@@ -6,7 +6,7 @@ extern "C" {
 #endif
 
 /*
-** Copyright (c) 2007 The Khronos Group Inc.
+** Copyright (c) 2007-2009 The Khronos Group Inc.
 ** 
 ** Permission is hereby granted, free of charge, to any person obtaining a
 ** copy of this software and/or associated documentation files (the
@@ -28,6 +28,8 @@ extern "C" {
 ** MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
 */
 
+/* Function declaration macros - to move into glplatform.h */
+
 #if defined(_WIN32) && !defined(APIENTRY) && !defined(__CYGWIN__) && !defined(__SCITECH_SNAP__)
 #define WIN32_LEAN_AND_MEAN 1
 #include <windows.h>
@@ -46,9 +48,9 @@ extern "C" {
 /*************************************************************/
 
 /* Header file version number, required by OpenGL ABI for Linux */
-/* glxext.h last updated 2008/10/22 */
+/* glxext.h last updated 2009/08/03 */
 /* Current version at http://www.opengl.org/registry/ */
-#define GLX_GLXEXT_VERSION 21
+#define GLX_GLXEXT_VERSION 23
 
 #ifndef GLX_VERSION_1_3
 #define GLX_WINDOW_BIT                     0x00000001
@@ -135,6 +137,12 @@ extern "C" {
 #define GLX_CONTEXT_FLAGS_ARB              0x2094
 #endif
 
+#ifndef GLX_ARB_create_context_profile
+#define GLX_CONTEXT_CORE_PROFILE_BIT_ARB   0x00000001
+#define GLX_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB 0x00000002
+#define GLX_CONTEXT_PROFILE_MASK_ARB       0x9126
+#endif
+
 #ifndef GLX_SGIS_multisample
 #define GLX_SAMPLE_BUFFERS_SGIS            100000
 #define GLX_SAMPLES_SGIS                   100001
@@ -518,6 +526,10 @@ extern GLXContext glXCreateContextAttribsARB (Display *, GLXFBConfig, GLXContext
 typedef GLXContext ( * PFNGLXCREATECONTEXTATTRIBSARBPROC) (Display *dpy, GLXFBConfig config, GLXContext share_context, Bool direct, const int *attrib_list);
 #endif
 
+#ifndef GLX_ARB_create_context_profile
+#define GLX_ARB_create_context_profile 1
+#endif
+
 #ifndef GLX_SGIS_multisample
 #define GLX_SGIS_multisample 1
 #endif
diff --git a/include/GL/wglext.h b/include/GL/wglext.h
index 3c492dd6348..1f096fb9193 100644
--- a/include/GL/wglext.h
+++ b/include/GL/wglext.h
@@ -48,9 +48,9 @@ extern "C" {
 /*************************************************************/
 
 /* Header file version number */
-/* wglext.h last updated 2009/03/03 */
+/* wglext.h last updated 2009/08/03 */
 /* Current version at http://www.opengl.org/registry/ */
-#define WGL_WGLEXT_VERSION 12
+#define WGL_WGLEXT_VERSION 15
 
 #ifndef WGL_ARB_buffer_region
 #define WGL_FRONT_COLOR_BUFFER_BIT_ARB 0x00000001
@@ -176,8 +176,8 @@ extern "C" {
 #endif
 
 #ifndef WGL_ARB_create_context
-#define WGL_CONTEXT_DEBUG_BIT_ARB      0x0001
-#define WGL_CONTEXT_FORWARD_COMPATIBLE_BIT_ARB 0x0002
+#define WGL_CONTEXT_DEBUG_BIT_ARB      0x00000001
+#define WGL_CONTEXT_FORWARD_COMPATIBLE_BIT_ARB 0x00000002
 #define WGL_CONTEXT_MAJOR_VERSION_ARB  0x2091
 #define WGL_CONTEXT_MINOR_VERSION_ARB  0x2092
 #define WGL_CONTEXT_LAYER_PLANE_ARB    0x2093
@@ -185,6 +185,13 @@ extern "C" {
 #define ERROR_INVALID_VERSION_ARB      0x2095
 #endif
 
+#ifndef WGL_ARB_create_context_profile
+#define WGL_CONTEXT_PROFILE_MASK_ARB   0x9126
+#define WGL_CONTEXT_CORE_PROFILE_BIT_ARB 0x00000001
+#define WGL_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB 0x00000002
+#define ERROR_INVALID_PROFILE_ARB      0x2096
+#endif
+
 #ifndef WGL_EXT_make_current_read
 #define ERROR_INVALID_PIXEL_TYPE_EXT   0x2043
 #endif
@@ -497,6 +504,10 @@ extern HGLRC WINAPI wglCreateContextAttribsARB (HDC, HGLRC, const int *);
 typedef HGLRC (WINAPI * PFNWGLCREATECONTEXTATTRIBSARBPROC) (HDC hDC, HGLRC hShareContext, const int *attribList);
 #endif
 
+#ifndef WGL_ARB_create_context_profile
+#define WGL_ARB_create_context_profile 1
+#endif
+
 #ifndef WGL_EXT_display_color_table
 #define WGL_EXT_display_color_table 1
 #ifdef WGL_WGLEXT_PROTOTYPES
diff --git a/progs/demos/cubemap.c b/progs/demos/cubemap.c
index 1f9f2905759..0df5ff09c33 100644
--- a/progs/demos/cubemap.c
+++ b/progs/demos/cubemap.c
@@ -43,6 +43,9 @@
 #include "GL/glut.h"
 #include "readtex.h"
 
+#ifndef GL_TEXTURE_CUBE_MAP_SEAMLESS
+#define GL_TEXTURE_CUBE_MAP_SEAMLESS 0x884F
+#endif
 
 static GLfloat Xrot = 0, Yrot = 0;
 static GLfloat EyeDist = 10;
@@ -53,6 +56,8 @@ static GLint FrameParity = 0;
 static GLenum FilterIndex = 0;
 static GLint ClampIndex = 0;
 static GLboolean supportFBO = GL_FALSE;
+static GLboolean supportSeamless = GL_FALSE;
+static GLboolean seamless = GL_FALSE;
 
 
 static struct {
@@ -91,7 +96,9 @@ static struct {
 
 
 
-#define eps1 0.99
+/* The effects of GL_ARB_seamless_cube_map don't show up unless eps1 is 1.0.
+ */
+#define eps1 1.0 /*0.99*/
 #define br   20.0  /* box radius */
 
 static const GLfloat tex_coords[] = {
@@ -231,6 +238,13 @@ static void draw( void )
    glTexParameteri(GL_TEXTURE_CUBE_MAP_ARB, GL_TEXTURE_MAG_FILTER,
                    FilterModes[FilterIndex].mag_mode);
 
+   if (supportSeamless) {
+      if (seamless) {
+	 glEnable(GL_TEXTURE_CUBE_MAP_SEAMLESS);
+      } else {
+	 glDisable(GL_TEXTURE_CUBE_MAP_SEAMLESS);
+      }
+   }
    wrap = ClampModes[ClampIndex].mode;
    glTexParameteri(GL_TEXTURE_CUBE_MAP_ARB, GL_TEXTURE_WRAP_S, wrap);
    glTexParameteri(GL_TEXTURE_CUBE_MAP_ARB, GL_TEXTURE_WRAP_T, wrap);
@@ -321,6 +335,11 @@ static void key(unsigned char k, int x, int y)
          mode = !mode;
          set_mode(mode);
          break;
+      case 's':
+	 seamless = ! seamless;
+	 printf("Seamless cube map filtering is %sabled\n",
+		(seamless) ? "en" : "dis" );
+	 break;
       case 'v':
          use_vertex_arrays = ! use_vertex_arrays;
          printf( "Vertex arrays are %sabled\n",
@@ -502,23 +521,26 @@ static void load_envmaps(void)
 static void init( GLboolean useImageFiles )
 {
    /* check for extensions */
-   {
-      char *exten = (char *) glGetString(GL_EXTENSIONS);
-      if (!strstr(exten, "GL_ARB_texture_cube_map")) {
-         printf("Sorry, this demo requires GL_ARB_texture_cube_map\n");
-         exit(0);
-      }
+   if (!GLEW_ARB_texture_cube_map) {
+      printf("Sorry, this demo requires GL_ARB_texture_cube_map\n");
+      exit(0);
+   }
 
-      /* Needed for glGenerateMipmapEXT / auto mipmapping
-       */
-      if (strstr(exten, "GL_EXT_framebuffer_object")) {
-         supportFBO = GL_TRUE;
-      }
-      else if (!strstr(exten, "GL_SGIS_generate_mipmap")) {
-         printf("Sorry, this demo requires GL_EXT_framebuffer_object or GL_SGIS_generate_mipmap\n");
-         exit(0);
-      }
+   /* Needed for glGenerateMipmapEXT / auto mipmapping
+    */
+   supportFBO = GLEW_EXT_framebuffer_object;
+
+   if (!supportFBO && !GLEW_SGIS_generate_mipmap) {
+      printf("Sorry, this demo requires GL_EXT_framebuffer_object or "
+	     "GL_SGIS_generate_mipmap\n");
+      exit(0);
    }
+
+   /* GLEW doesn't know about this extension yet, so use the old GLUT function
+    * to check for availability.
+    */
+   supportSeamless = glutExtensionSupported("GL_ARB_seamless_cube_map");
+
    printf("GL_RENDERER: %s\n", (char *) glGetString(GL_RENDERER));
 
    if (useImageFiles) {
diff --git a/progs/egl/Makefile b/progs/egl/Makefile
index d3c32d46f78..fd72f540cae 100644
--- a/progs/egl/Makefile
+++ b/progs/egl/Makefile
@@ -17,6 +17,7 @@ PROGRAMS = \
 	eglgears \
 	eglscreen \
 	peglgears \
+	xeglbindtex \
 	xeglgears \
 	xeglthreads \
 	xegl_tri
@@ -84,6 +85,12 @@ peglgears.o: peglgears.c $(HEADERS)
 	$(CC) -c $(CFLAGS) -I$(TOP)/include peglgears.c
 
 
+xeglbindtex: xeglbindtex.o $(TOP)/$(LIB_DIR)/libEGL.so
+	$(CC) $(CFLAGS) xeglbindtex.o -L$(TOP)/$(LIB_DIR) -lEGL -lGL $(LIBDRM_LIB) $(APP_LIB_DEPS) -o $@
+
+xeglbindtex.o: xeglbindtex.c $(HEADERS)
+	$(CC) -c $(CFLAGS) -I$(TOP)/include xeglbindtex.c
+
 xeglgears: xeglgears.o $(TOP)/$(LIB_DIR)/libEGL.so
 	$(CC) $(CFLAGS) xeglgears.o -L$(TOP)/$(LIB_DIR) -lEGL -lGL $(LIBDRM_LIB) $(APP_LIB_DEPS) -o $@
 
diff --git a/progs/egl/xeglbindtex.c b/progs/egl/xeglbindtex.c
new file mode 100644
index 00000000000..fdd9fe2b87b
--- /dev/null
+++ b/progs/egl/xeglbindtex.c
@@ -0,0 +1,474 @@
+/*
+ * Simple demo for eglBindTexImage.  Based on xegl_tri.c by
+ *
+ * Copyright (C) 2008  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * The spec says that eglBindTexImage supports only OpenGL ES context, but this
+ * demo uses OpenGL context.  Keep in mind that this is non-standard.
+ */
+
+#include <math.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <X11/Xlib.h>
+#include <X11/Xutil.h>
+#include <X11/keysym.h>
+#include <GL/gl.h>
+#include <EGL/egl.h>
+
+static EGLDisplay dpy;
+static EGLContext ctx_win, ctx_pbuf;
+static EGLSurface surf_win, surf_pbuf;
+static GLuint tex_pbuf;
+
+static GLfloat view_rotx = 0.0, view_roty = 0.0, view_rotz = 0.0;
+static GLboolean blend = GL_TRUE;
+static GLuint color_flow;
+
+static void
+make_pbuffer(int width, int height)
+{
+   static const EGLint config_attribs[] = {
+      EGL_RED_SIZE, 8,
+      EGL_GREEN_SIZE, 8,
+      EGL_BLUE_SIZE, 8,
+      EGL_BIND_TO_TEXTURE_RGB, EGL_TRUE,
+      EGL_NONE
+   };
+   EGLint pbuf_attribs[] = {
+      EGL_WIDTH, width,
+      EGL_HEIGHT, height,
+      EGL_TEXTURE_FORMAT, EGL_TEXTURE_RGB,
+      EGL_TEXTURE_TARGET, EGL_TEXTURE_2D,
+      EGL_NONE
+   };
+   EGLConfig config;
+   EGLint num_configs;
+
+   if (!eglChooseConfig(dpy, config_attribs, &config, 1, &num_configs)) {
+      printf("Error: couldn't get an EGL visual config for pbuffer\n");
+      exit(1);
+   }
+
+   eglBindAPI(EGL_OPENGL_API);
+   ctx_pbuf = eglCreateContext(dpy, config, EGL_NO_CONTEXT, NULL );
+   surf_pbuf = eglCreatePbufferSurface(dpy, config, pbuf_attribs);
+   if (surf_pbuf == EGL_NO_SURFACE) {
+      printf("failed to allocate pbuffer\n");
+      exit(1);
+   }
+
+   glGenTextures(1, &tex_pbuf);
+}
+
+static void
+use_pbuffer(void)
+{
+   static int initialized;
+
+   eglMakeCurrent(dpy, surf_pbuf, surf_pbuf, ctx_pbuf);
+   if (!initialized) {
+      EGLint width, height;
+      GLfloat ar;
+
+      initialized = 1;
+
+      eglQuerySurface(dpy, surf_pbuf, EGL_WIDTH, &width);
+      eglQuerySurface(dpy, surf_pbuf, EGL_WIDTH, &height);
+      ar = (GLfloat) width / (GLfloat) height;
+
+      glViewport(0, 0, (GLint) width, (GLint) height);
+
+      glMatrixMode(GL_PROJECTION);
+      glLoadIdentity();
+      glFrustum(-ar, ar, -1, 1, 1.0, 10.0);
+
+      glMatrixMode(GL_MODELVIEW);
+      glLoadIdentity();
+
+      /* y-inverted */
+      glScalef(1.0, -1.0, 1.0);
+
+      glTranslatef(0.0, 0.0, -5.0);
+
+      glClearColor(0.2, 0.2, 0.2, 0.0);
+   }
+}
+
+static void
+make_window(Display *x_dpy, const char *name,
+            int x, int y, int width, int height,
+            Window *winRet)
+{
+   static const EGLint attribs[] = {
+      EGL_RED_SIZE, 8,
+      EGL_GREEN_SIZE, 8,
+      EGL_BLUE_SIZE, 8,
+      EGL_ALPHA_SIZE, 8,
+      EGL_DEPTH_SIZE, 8,
+      EGL_NONE
+   };
+
+   int scrnum;
+   XSetWindowAttributes attr;
+   unsigned long mask;
+   Window root;
+   Window win;
+   XVisualInfo *visInfo, visTemplate;
+   int num_visuals;
+   EGLConfig config;
+   EGLint num_configs, vid;
+
+   scrnum = DefaultScreen( x_dpy );
+   root = RootWindow( x_dpy, scrnum );
+
+   if (!eglChooseConfig(dpy, attribs, &config, 1, &num_configs)) {
+      printf("Error: couldn't get an EGL visual config\n");
+      exit(1);
+   }
+
+   if (!eglGetConfigAttrib(dpy, config, EGL_NATIVE_VISUAL_ID, &vid)) {
+      printf("Error: eglGetConfigAttrib() failed\n");
+      exit(1);
+   }
+
+   /* The X window visual must match the EGL config */
+   visTemplate.visualid = vid;
+   visInfo = XGetVisualInfo(x_dpy, VisualIDMask, &visTemplate, &num_visuals);
+   if (!visInfo) {
+      printf("Error: couldn't get X visual\n");
+      exit(1);
+   }
+
+   /* window attributes */
+   attr.background_pixel = 0;
+   attr.border_pixel = 0;
+   attr.colormap = XCreateColormap( x_dpy, root, visInfo->visual, AllocNone);
+   attr.event_mask = StructureNotifyMask | ExposureMask | KeyPressMask;
+   attr.override_redirect = 0;
+   mask = CWBackPixel | CWBorderPixel | CWColormap | CWEventMask | CWOverrideRedirect;
+
+   win = XCreateWindow( x_dpy, root, 0, 0, width, height,
+		        0, visInfo->depth, InputOutput,
+		        visInfo->visual, mask, &attr );
+
+   /* set hints and properties */
+   {
+      XSizeHints sizehints;
+      sizehints.x = x;
+      sizehints.y = y;
+      sizehints.width  = width;
+      sizehints.height = height;
+      sizehints.flags = USSize | USPosition;
+      XSetNormalHints(x_dpy, win, &sizehints);
+      XSetStandardProperties(x_dpy, win, name, name,
+                              None, (char **)NULL, 0, &sizehints);
+   }
+
+   eglBindAPI(EGL_OPENGL_API);
+   ctx_win = eglCreateContext(dpy, config, EGL_NO_CONTEXT, NULL );
+   if (!ctx_win) {
+      printf("Error: glXCreateContext failed\n");
+      exit(1);
+   }
+
+   surf_win = eglCreateWindowSurface(dpy, config, win, NULL);
+
+   XFree(visInfo);
+
+   *winRet = win;
+}
+
+static void
+use_window(void)
+{
+   static int initialized;
+
+   eglMakeCurrent(dpy, surf_win, surf_win, ctx_win);
+   if (!initialized) {
+      initialized = 1;
+      glEnable(GL_TEXTURE_2D);
+      glBindTexture(GL_TEXTURE_2D, tex_pbuf);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+
+      glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
+   }
+}
+
+static void
+draw_triangle(void)
+{
+   static const GLfloat verts[3][2] = {
+      { -3, -3 },
+      {  3, -3 },
+      {  0,  3 }
+   };
+   GLfloat colors[3][3] = {
+      { 1, 0, 0 },
+      { 0, 1, 0 },
+      { 0, 0, 1 }
+   };
+   GLint i;
+
+   /* flow the color */
+   for (i = 0; i < 3; i++) {
+      GLint first = (i + color_flow / 256) % 3;
+      GLint second = (first + 1) % 3;
+      GLint third = (second + 1) % 3;
+      GLfloat c = (color_flow % 256) / 256.0f;
+
+      c = c * c * c;
+      colors[i][first] = 1.0f - c;
+      colors[i][second] = c;
+      colors[i][third] = 0.0f;
+   }
+
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   glVertexPointer(2, GL_FLOAT, 0, verts);
+   glColorPointer(3, GL_FLOAT, 0, colors);
+   glEnableClientState(GL_VERTEX_ARRAY);
+   glEnableClientState(GL_COLOR_ARRAY);
+
+   glDrawArrays(GL_TRIANGLES, 0, 3);
+
+   glDisableClientState(GL_VERTEX_ARRAY);
+   glDisableClientState(GL_COLOR_ARRAY);
+}
+
+static void
+draw_textured_cube(void)
+{
+   static const GLfloat verts[][2] = {
+      { -4, -4 },
+      {  4, -4 },
+      {  4,  4 },
+      { -4,  4 }
+   };
+   static const GLfloat colors[][4] = {
+      { 1, 1, 1, 0.5 },
+      { 1, 1, 1, 0.5 },
+      { 1, 1, 1, 0.5 },
+      { 1, 1, 1, 0.5 }
+   };
+   static const GLfloat texs[][2] = {
+      { 0, 0 },
+      { 1, 0 },
+      { 1, 1 },
+      { 0, 1 }
+   };
+   static const GLfloat xforms[6][4] = {
+      {   0, 0, 1, 0 },
+      {  90, 0, 1, 0 },
+      { 180, 0, 1, 0 },
+      { 270, 0, 1, 0 },
+      {  90, 1, 0, 0 },
+      { -90, 1, 0, 0 }
+   };
+   GLint i;
+
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   if (blend) {
+      glDisable(GL_DEPTH_TEST);
+      glEnable(GL_BLEND);
+   } else {
+      glEnable(GL_DEPTH_TEST);
+      glDisable(GL_BLEND);
+   }
+
+   glVertexPointer(2, GL_FLOAT, 0, verts);
+   glColorPointer(4, GL_FLOAT, 0, colors);
+   glTexCoordPointer(2, GL_FLOAT, 0, texs);
+
+   glEnableClientState(GL_VERTEX_ARRAY);
+   glEnableClientState(GL_COLOR_ARRAY);
+   glEnableClientState(GL_TEXTURE_COORD_ARRAY);
+
+   for (i = 0; i < 6; i++) {
+      glPushMatrix();
+      glRotatef(xforms[i][0], xforms[i][1], xforms[i][2], xforms[i][3]);
+      glTranslatef(0, 0, 4.1);
+      glDrawArrays(GL_TRIANGLE_FAN, 0, 4);
+      glPopMatrix();
+   }
+
+   glDisableClientState(GL_VERTEX_ARRAY);
+   glDisableClientState(GL_COLOR_ARRAY);
+   glDisableClientState(GL_TEXTURE_COORD_ARRAY);
+}
+
+static void
+draw(void)
+{
+   use_pbuffer();
+   draw_triangle();
+
+   use_window();
+
+   eglBindTexImage(dpy, surf_pbuf, EGL_BACK_BUFFER);
+
+   glPushMatrix();
+   glRotatef(view_rotx, 1, 0, 0);
+   glRotatef(view_roty, 0, 1, 0);
+   glRotatef(view_rotz, 0, 0, 1);
+
+   draw_textured_cube();
+
+   glPopMatrix();
+
+   eglReleaseTexImage(dpy, surf_pbuf, EGL_BACK_BUFFER);
+}
+
+/* new window size or exposure */
+static void
+reshape(int width, int height)
+{
+   GLfloat ar = (GLfloat) width / (GLfloat) height;
+
+   use_window();
+
+   glViewport(0, 0, (GLint) width, (GLint) height);
+
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glFrustum(-ar, ar, -1, 1, 5.0, 60.0);
+
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+   glTranslatef(0.0, 0.0, -40.0);
+}
+
+static void
+event_loop(Display *x_dpy, Window win)
+{
+   while (1) {
+      int redraw = 1;
+
+      if (XPending(x_dpy) > 0) {
+         XEvent event;
+         XNextEvent(x_dpy, &event);
+
+         switch (event.type) {
+         case Expose:
+            redraw = 1;
+            break;
+         case ConfigureNotify:
+            reshape(event.xconfigure.width, event.xconfigure.height);
+            break;
+         case KeyPress:
+            {
+               char buffer[10];
+               int r, code;
+               code = XLookupKeysym(&event.xkey, 0);
+               if (code == XK_Left) {
+                  view_roty += 5.0;
+               }
+               else if (code == XK_Right) {
+                  view_roty -= 5.0;
+               }
+               else if (code == XK_Up) {
+                  view_rotx += 5.0;
+               }
+               else if (code == XK_Down) {
+                  view_rotx -= 5.0;
+               }
+               else if (code == XK_b) {
+                  blend = !blend;
+               }
+               else {
+                  r = XLookupString(&event.xkey, buffer, sizeof(buffer),
+                        NULL, NULL);
+                  if (buffer[0] == 27) {
+                     /* escape */
+                     return;
+                  }
+               }
+            }
+            redraw = 1;
+            break;
+         default:
+            ; /*no-op*/
+         }
+      }
+
+      if (redraw) {
+         view_rotx += 1.0;
+         view_roty += 2.0;
+         view_rotz += 1.5;
+         color_flow += 20;
+         draw();
+         eglSwapBuffers(dpy, surf_win);
+      }
+   }
+}
+
+int
+main(int argc, char *argv[])
+{
+   const int winWidth = 300, winHeight = 300;
+   Display *x_dpy;
+   Window win;
+   char *dpyName = NULL;
+   EGLint egl_major, egl_minor;
+   const char *s;
+
+   x_dpy = XOpenDisplay(dpyName);
+   if (!x_dpy) {
+      printf("Error: couldn't open display %s\n",
+	     dpyName ? dpyName : getenv("DISPLAY"));
+      return -1;
+   }
+
+   dpy = eglGetDisplay(x_dpy);
+   if (!dpy) {
+      printf("Error: eglGetDisplay() failed\n");
+      return -1;
+   }
+
+   if (!eglInitialize(dpy, &egl_major, &egl_minor)) {
+      printf("Error: eglInitialize() failed\n");
+      return -1;
+   }
+
+   s = eglQueryString(dpy, EGL_VERSION);
+   printf("EGL_VERSION = %s\n", s);
+
+   make_window(x_dpy, "color flow", 0, 0, winWidth, winHeight, &win);
+   make_pbuffer(winWidth, winHeight);
+
+   XMapWindow(x_dpy, win);
+
+   reshape(winWidth, winHeight);
+   event_loop(x_dpy, win);
+
+   glDeleteTextures(1, &tex_pbuf);
+
+   eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, EGL_NO_CONTEXT);
+   eglTerminate(dpy);
+
+   XDestroyWindow(x_dpy, win);
+   XCloseDisplay(x_dpy);
+
+   return 0;
+}
diff --git a/progs/glsl/.gitignore b/progs/glsl/.gitignore
index 39d90c23ac5..986775bac22 100644
--- a/progs/glsl/.gitignore
+++ b/progs/glsl/.gitignore
@@ -22,6 +22,7 @@ samplers_array
 shaderutil.c
 shaderutil.h
 shadow_sampler
+shtest
 skinning
 texaaline
 texdemo1
diff --git a/progs/glsl/Makefile b/progs/glsl/Makefile
index eedd866c957..0875bdd7950 100644
--- a/progs/glsl/Makefile
+++ b/progs/glsl/Makefile
@@ -14,6 +14,11 @@ LIBS = -L$(TOP)/$(LIB_DIR) -l$(GLUT_LIB) -l$(GLEW_LIB) -l$(GLU_LIB) -l$(GL_LIB)
 
 INCLUDE_DIRS = -I$(TOP)/progs/util
 
+# using : to avoid APP_CC pointing to CC loop
+CC:=$(APP_CC)
+CFLAGS += -I$(INCDIR)
+LDLIBS=$(LIBS)
+
 DEMO_SOURCES = \
 	array.c \
 	bitmap.c \
@@ -33,6 +38,7 @@ DEMO_SOURCES = \
 	points.c \
 	samplers.c \
 	shadow_sampler.c \
+	shtest.c \
 	skinning.c \
 	texaaline.c \
 	texdemo1.c \
@@ -52,28 +58,16 @@ UTIL_SOURCES = \
 	readtex.c
 
 UTIL_OBJS = $(UTIL_SOURCES:.c=.o)
-
-
+PROG_OBJS = $(DEMO_SOURCES:.c=.o)
 PROGS = $(DEMO_SOURCES:%.c=%)
 
-
-
-##### RULES #####
-
-# make .o file from .c file:
-.c.o:
-	$(APP_CC) -c -I$(INCDIR) $(CFLAGS) $< -o $@
-
-
-# make executable from .o files
-.o:
-	$(APP_CC) $(INCLUDES) $(CFLAGS) $(LDFLAGS) $< $(UTIL_OBJS) $(LIBS) -o $@
-
-
 ##### TARGETS #####
 
 default: $(PROGS)
 
+$(PROG_OBJS): $(UTIL_HEADERS)
+
+$(PROGS): $(UTIL_OBJS)
 
 clean:
 	-rm -f $(PROGS)
@@ -83,7 +77,6 @@ clean:
 	-rm -f readtex.*
 
 
-
 ##### Extra dependencies
 
 extfuncs.h: $(TOP)/progs/util/extfuncs.h
@@ -101,138 +94,3 @@ shaderutil.c: $(TOP)/progs/util/shaderutil.c
 shaderutil.h: $(TOP)/progs/util/shaderutil.h
 	cp $< .
 
-
-
-array.o: $(UTIL_HEADERS)
-
-array: array.o  $(UTIL_OBJS)
-
-
-bitmap.o: $(UTIL_HEADERS)
-
-bitmap: bitmap.o $(UTIL_OBJS)
-
-
-brick.o: $(UTIL_HEADERS)
-
-brick: brick.o $(UTIL_OBJS)
-
-
-bump.o: $(UTIL_HEADERS)
-
-bump: bump.o $(UTIL_OBJS)
-
-
-convolutions.o: $(UTIL_HEADERS)
-
-convolutions: convolutions.o $(UTIL_OBJS)
-
-
-deriv.o: deriv.c $(UTIL_HEADERS)
-
-deriv: deriv.o $(UTIL_OBJS)
-
-
-identity.o: $(UTIL_HEADERS)
-
-identity: identity.o $(UTIL_OBJS)
-
-
-fragcoord.o: $(UTIL_HEADERS)
-
-fragcoord: fragcoord.o $(UTIL_OBJS)
-
-
-linktest.o: $(UTIL_HEADERS)
-
-linktest: linktest.o $(UTIL_OBJS)
-
-
-mandelbrot.o: $(UTIL_HEADERS)
-
-mandelbrot: mandelbrot.o $(UTIL_OBJS)
-
-
-multinoise.o: $(UTIL_HEADERS)
-
-multinoise: multinoise.o $(UTIL_OBJS)
-
-
-multitex.o: $(UTIL_HEADERS)
-
-multitex: multitex.o $(UTIL_OBJS)
-
-
-noise.o: $(UTIL_HEADERS)
-
-noise: noise.o $(UTIL_OBJS)
-
-
-noise2.o: $(UTIL_HEADERS)
-
-noise2: noise2.o $(UTIL_OBJS)
-
-
-points.o: $(UTIL_HEADERS)
-
-points: points.o $(UTIL_OBJS)
-
-
-pointcoord.o: $(UTIL_HEADERS)
-
-pointcoord: pointcoord.o $(UTIL_OBJS)
-
-
-samplers.o: $(UTIL_HEADERS)
-
-samplers: samplers.o $(UTIL_OBJS)
-
-
-samplers_array.o: $(UTIL_HEADERS)
-
-samplers_array: samplers_array.o $(UTIL_OBJS)
-
-
-shadow_sampler.o: $(UTIL_HEADERS)
-
-shadow_sampler: shadow_sampler.o $(UTIL_OBJS)
-
-
-skinning.o: $(UTIL_HEADERS)
-
-skinning: skinning.o $(UTIL_OBJS)
-
-
-texaaline.o: $(UTIL_HEADERS)
-
-texaaline: texaaline.o $(UTIL_OBJS)
-
-
-texdemo1.o: $(UTIL_HEADERS)
-
-texdemo1: texdemo1.o $(UTIL_OBJS)
-
-
-toyball.o: $(UTIL_HEADERS)
-
-toyball: toyball.o $(UTIL_OBJS)
-
-
-twoside.o: $(UTIL_HEADERS)
-
-twoside: twoside.o $(UTIL_OBJS)
-
-
-trirast.o: $(UTIL_HEADERS)
-
-trirast: trirast.o $(UTIL_OBJS)
-
-
-vert-or-frag-only.o: $(UTIL_HEADERS)
-
-vert-or-frag-only: vert-or-frag-only.o $(UTIL_OBJS)
-
-
-vert-tex.o: $(UTIL_HEADERS)
-
-vert-tex: vert-tex.o $(UTIL_OBJS)
diff --git a/progs/glsl/brick.c b/progs/glsl/brick.c
index 1d08b231e7e..20417aa4626 100644
--- a/progs/glsl/brick.c
+++ b/progs/glsl/brick.c
@@ -24,12 +24,12 @@ static GLuint program;
 
 static struct uniform_info Uniforms[] = {
    /* vert */
-   { "LightPosition",     3, GL_FLOAT, { 0.1, 0.1, 9.0, 0}, -1 },
+   { "LightPosition",     1, GL_FLOAT_VEC3, { 0.1, 0.1, 9.0, 0}, -1 },
    /* frag */
-   { "BrickColor",        3, GL_FLOAT, { 0.8, 0.2, 0.2, 0 }, -1 },
-   { "MortarColor",       3, GL_FLOAT, { 0.6, 0.6, 0.6, 0 }, -1 },
-   { "BrickSize",         2, GL_FLOAT, { 1.0, 0.3, 0, 0 }, -1 },
-   { "BrickPct",          2, GL_FLOAT, { 0.9, 0.8, 0, 0 }, -1 },
+   { "BrickColor",        1, GL_FLOAT_VEC3, { 0.8, 0.2, 0.2, 0 }, -1 },
+   { "MortarColor",       1, GL_FLOAT_VEC3, { 0.6, 0.6, 0.6, 0 }, -1 },
+   { "BrickSize",         1, GL_FLOAT_VEC2, { 1.0, 0.3, 0, 0 }, -1 },
+   { "BrickPct",          1, GL_FLOAT_VEC2, { 0.9, 0.8, 0, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -148,7 +148,8 @@ Init(void)
 
    glUseProgram(program);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
+   PrintUniforms(Uniforms);
 
    assert(glGetError() == 0);
 
@@ -183,7 +184,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/brick.shtest b/progs/glsl/brick.shtest
new file mode 100644
index 00000000000..8a2152692eb
--- /dev/null
+++ b/progs/glsl/brick.shtest
@@ -0,0 +1,8 @@
+vs CH06-brick.vert
+fs CH06-brick.frag
+uniform LightPosition GL_FLOAT_VEC3 0.1 0.1 9.0
+uniform BrickColor GL_FLOAT_VEC3 0.8 0.2 0.2
+uniform MortarColor GL_FLOAT_VEC3 0.6 0.6 0.6
+uniform BrickSize  GL_FLOAT_VEC2 1.0 0.3
+uniform BrickPct GL_FLOAT_VEC2 0.9 0.8
+
diff --git a/progs/glsl/bump.c b/progs/glsl/bump.c
index ddb986abcb5..87669aec736 100644
--- a/progs/glsl/bump.c
+++ b/progs/glsl/bump.c
@@ -24,11 +24,11 @@ static GLuint program;
 
 
 static struct uniform_info Uniforms[] = {
-   { "LightPosition",    3, GL_FLOAT, { 0.57737, 0.57735, 0.57735, 0.0 }, -1 },
-   { "SurfaceColor",     3, GL_FLOAT, { 0.8, 0.8, 0.2, 0 }, -1 },
-   { "BumpDensity",      1, GL_FLOAT, { 10.0, 0, 0, 0 }, -1 },
-   { "BumpSize",         1, GL_FLOAT, { 0.125, 0, 0, 0 }, -1 },
-   { "SpecularFactor",   1, GL_FLOAT, { 0.5, 0, 0, 0 }, -1 },
+   { "LightPosition",  1, GL_FLOAT_VEC3, { 0.57737, 0.57735, 0.57735, 0.0 }, -1 },
+   { "SurfaceColor",   1, GL_FLOAT_VEC3, { 0.8, 0.8, 0.2, 0 }, -1 },
+   { "BumpDensity",    1, GL_FLOAT, { 10.0, 0, 0, 0 }, -1 },
+   { "BumpSize",       1, GL_FLOAT, { 0.125, 0, 0, 0 }, -1 },
+   { "SpecularFactor", 1, GL_FLOAT, { 0.5, 0, 0, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -242,7 +242,8 @@ Init(void)
 
    CheckError(__LINE__);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
+   PrintUniforms(Uniforms);
 
    CheckError(__LINE__);
 
@@ -280,7 +281,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/convolutions.c b/progs/glsl/convolutions.c
index 22ce7edcdc5..c2fb76e1aa5 100644
--- a/progs/glsl/convolutions.c
+++ b/progs/glsl/convolutions.c
@@ -448,7 +448,6 @@ int main(int argc, char **argv)
 {
    glutInit(&argc, argv);
 
-   glutInitWindowPosition(0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_ALPHA | GLUT_DOUBLE);
 
diff --git a/progs/glsl/deriv.c b/progs/glsl/deriv.c
index 9cf1e40e3e6..265a5157154 100644
--- a/progs/glsl/deriv.c
+++ b/progs/glsl/deriv.c
@@ -220,7 +220,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(200, 200);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/fragcoord.c b/progs/glsl/fragcoord.c
index 9f56a038c96..3dfcec87a56 100644
--- a/progs/glsl/fragcoord.c
+++ b/progs/glsl/fragcoord.c
@@ -166,7 +166,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(WinWidth, WinHeight);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/identity.c b/progs/glsl/identity.c
index a772ccd716c..526e9b82c10 100644
--- a/progs/glsl/identity.c
+++ b/progs/glsl/identity.c
@@ -187,7 +187,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(200, 200);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/mandelbrot.c b/progs/glsl/mandelbrot.c
index 38dffc3e741..b05ef37fae1 100644
--- a/progs/glsl/mandelbrot.c
+++ b/progs/glsl/mandelbrot.c
@@ -25,7 +25,7 @@ static GLuint program;
 
 static struct uniform_info Uniforms[] = {
    /* vert */
-   { "LightPosition",        3, GL_FLOAT, { 0.1, 0.1, 9.0, 0}, -1 },
+   { "LightPosition",        1, GL_FLOAT_VEC3, { 0.1, 0.1, 9.0, 0}, -1 },
    { "SpecularContribution", 1, GL_FLOAT, { 0.5, 0, 0, 0 }, -1 },
    { "DiffuseContribution",  1, GL_FLOAT, { 0.5, 0, 0, 0 }, -1 },
    { "Shininess",            1, GL_FLOAT, { 20.0, 0, 0, 0 }, -1 },
@@ -34,9 +34,9 @@ static struct uniform_info Uniforms[] = {
    { "Zoom",                 1, GL_FLOAT, { 0.125, 0, 0, 0 }, -1 },
    { "Xcenter",              1, GL_FLOAT, { -1.5, 0, 0, 0 }, -1 },
    { "Ycenter",              1, GL_FLOAT, { .005, 0, 0, 0 }, -1 },
-   { "InnerColor",           3, GL_FLOAT, { 1, 0, 0, 0 }, -1 },
-   { "OuterColor1",          3, GL_FLOAT, { 0, 1, 0, 0 }, -1 },
-   { "OuterColor2",          3, GL_FLOAT, { 0, 0, 1, 0 }, -1 },
+   { "InnerColor",           1, GL_FLOAT_VEC3, { 1, 0, 0, 0 }, -1 },
+   { "OuterColor1",          1, GL_FLOAT_VEC3, { 0, 1, 0, 0 }, -1 },
+   { "OuterColor2",          1, GL_FLOAT_VEC3, { 0, 0, 1, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -159,7 +159,8 @@ Init(void)
 
    glUseProgram(program);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
+   PrintUniforms(Uniforms);
 
    uZoom = glGetUniformLocation(program, "Zoom");
    uXcenter = glGetUniformLocation(program, "Xcenter");
@@ -198,7 +199,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/mandelbrot.shtest b/progs/glsl/mandelbrot.shtest
new file mode 100644
index 00000000000..4f4e5c747ea
--- /dev/null
+++ b/progs/glsl/mandelbrot.shtest
@@ -0,0 +1,13 @@
+vs CH18-mandel.vert
+fs CH18-mandel.frag
+uniform LightPosition        GL_FLOAT_VEC3 0.1 0.1 9.0
+uniform SpecularContribution GL_FLOAT 0.5
+uniform DiffuseContribution  GL_FLOAT 0.5
+uniform Shininess            GL_FLOAT 20.0
+uniform Iterations           GL_FLOAT 12
+uniform Zoom                 GL_FLOAT 0.125
+uniform Xcenter              GL_FLOAT -1.5
+uniform Ycenter              GL_FLOAT .005
+uniform InnerColor           GL_FLOAT_VEC3 1 0 0
+uniform OuterColor1          GL_FLOAT_VEC3 0 1 0
+uniform OuterColor2          GL_FLOAT_VEC3 0 0 1
diff --git a/progs/glsl/multinoise.c b/progs/glsl/multinoise.c
index 0afe2308012..0d4026e29cf 100644
--- a/progs/glsl/multinoise.c
+++ b/progs/glsl/multinoise.c
@@ -262,7 +262,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/multitex.c b/progs/glsl/multitex.c
index 5e971716add..6ec9c833e67 100644
--- a/progs/glsl/multitex.c
+++ b/progs/glsl/multitex.c
@@ -59,8 +59,8 @@ static GLint VertCoord_attr = -1, TexCoord0_attr = -1, TexCoord1_attr = -1;
 
 /* value[0] = tex unit */
 static struct uniform_info Uniforms[] = {
-   { "tex1",  1, GL_INT, { 0, 0, 0, 0 }, -1 },
-   { "tex2",  1, GL_INT, { 1, 0, 0, 0 }, -1 },
+   { "tex1",  1, GL_SAMPLER_2D, { 0, 0, 0, 0 }, -1 },
+   { "tex2",  1, GL_SAMPLER_2D, { 1, 0, 0, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -134,7 +134,7 @@ DrawPolygonArray(void)
 
    if (VertCoord_attr >= 0) {
       glVertexAttribPointer(VertCoord_attr, 2, GL_FLOAT, GL_FALSE,
-                                 0, VertCoords);
+                                 0, vertPtr);
       glEnableVertexAttribArray(VertCoord_attr);
    }
    else {
@@ -143,11 +143,11 @@ DrawPolygonArray(void)
    }
 
    glVertexAttribPointer(TexCoord0_attr, 2, GL_FLOAT, GL_FALSE,
-                              0, Tex0Coords);
+                              0, tex0Ptr);
    glEnableVertexAttribArray(TexCoord0_attr);
 
    glVertexAttribPointer(TexCoord1_attr, 2, GL_FLOAT, GL_FALSE,
-                              0, Tex1Coords);
+                              0, tex1Ptr);
    glEnableVertexAttribArray(TexCoord1_attr);
 
    glDrawArrays(GL_TRIANGLE_FAN, 0, 4);
@@ -328,7 +328,8 @@ CreateProgram(const char *vertProgFile, const char *fragProgFile,
 
    glUseProgram(program);
 
-   InitUniforms(program, uniforms);
+   SetUniformValues(program, uniforms);
+   PrintUniforms(Uniforms);
 
    VertCoord_attr = glGetAttribLocation(program, "VertCoord");
    if (VertCoord_attr > 0) {
diff --git a/progs/glsl/multitex.shtest b/progs/glsl/multitex.shtest
new file mode 100644
index 00000000000..5be45f6c7cb
--- /dev/null
+++ b/progs/glsl/multitex.shtest
@@ -0,0 +1,6 @@
+vs multitex.vert
+fs multitex.frag
+texture 0 ../images/tile.rgb
+texture 1 ../images/tree2.rgba
+uniform tex1  GL_SAMPLER_2D  0
+uniform tex2  GL_SAMPLER_2D  1
diff --git a/progs/glsl/noise.c b/progs/glsl/noise.c
index 59f594e78bf..fdab263ea6a 100644
--- a/progs/glsl/noise.c
+++ b/progs/glsl/noise.c
@@ -35,8 +35,8 @@ static const char *FragShaderText =
 
 
 static struct uniform_info Uniforms[] = {
-   { "Scale",    4, GL_FLOAT, { 0.5, 0.4, 0.0, 0}, -1 },
-   { "Bias",     4, GL_FLOAT, { 0.5, 0.3, 0.0, 0}, -1 },
+   { "Scale",    1, GL_FLOAT_VEC4, { 0.5, 0.4, 0.0, 0}, -1 },
+   { "Bias",     1, GL_FLOAT_VEC4, { 0.5, 0.3, 0.0, 0}, -1 },
    { "Slice",    1, GL_FLOAT, { 0.5, 0, 0, 0}, -1 },
    END_OF_UNIFORMS
 };
@@ -179,7 +179,8 @@ Init(void)
 
    glUseProgram(program);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
+   PrintUniforms(Uniforms);
 
    assert(glGetError() == 0);
 
@@ -199,7 +200,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/noise2.c b/progs/glsl/noise2.c
index e972b62673f..7a28f09947c 100644
--- a/progs/glsl/noise2.c
+++ b/progs/glsl/noise2.c
@@ -186,7 +186,6 @@ static void Init (void)
 int main (int argc, char *argv[])
 {
 	glutInit (&argc, argv);
-	glutInitWindowPosition ( 0, 0);
 	glutInitWindowSize (200, 200);
 	glutInitDisplayMode (GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
 	glutCreateWindow (argv[0]);
diff --git a/progs/glsl/pointcoord.c b/progs/glsl/pointcoord.c
index 27b73a05dee..5dced6fac3a 100644
--- a/progs/glsl/pointcoord.c
+++ b/progs/glsl/pointcoord.c
@@ -187,7 +187,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(WinWidth, WinHeight);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/shadow_sampler.c b/progs/glsl/shadow_sampler.c
index 0a4d04dd8cf..0adc9d88ba4 100644
--- a/progs/glsl/shadow_sampler.c
+++ b/progs/glsl/shadow_sampler.c
@@ -321,7 +321,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 300);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/shtest.c b/progs/glsl/shtest.c
new file mode 100644
index 00000000000..2622af13138
--- /dev/null
+++ b/progs/glsl/shtest.c
@@ -0,0 +1,629 @@
+/*
+ * Simple shader test harness.
+ * Brian Paul
+ * 13 Aug 2009
+ *
+ * Usage:
+ *   shtest --vs vertShaderFile --fs fragShaderFile
+ *
+ *   In this case the given vertex/frag shaders are read and compiled.
+ *   Random values are assigned to the uniforms.
+ *
+ * or:
+ *   shtest configFile
+ *
+ *   In this case a config file is read that specifies the file names
+ *   of the shaders plus initial values for uniforms.
+ *
+ * Example config file:
+ *
+ * vs shader.vert
+ * fs shader.frag
+ * uniform pi 3.14159
+ * uniform v1 1.0 0.5 0.2 0.3
+ *
+ */
+
+
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+#include <GL/glew.h>
+#include <GL/glu.h>
+#include <GL/glut.h>
+#include "shaderutil.h"
+#include "readtex.h"
+
+
+typedef enum
+{
+   SPHERE,
+   CUBE,
+   NUM_SHAPES
+} shape;
+
+
+static char *FragShaderFile = NULL;
+static char *VertShaderFile = NULL;
+static char *ConfigFile = NULL;
+
+/* program/shader objects */
+static GLuint fragShader;
+static GLuint vertShader;
+static GLuint Program;
+
+
+#define MAX_UNIFORMS 100
+static struct uniform_info Uniforms[MAX_UNIFORMS];
+static GLuint NumUniforms = 0;
+
+
+#define MAX_ATTRIBS 100
+static struct attrib_info Attribs[MAX_ATTRIBS];
+static GLuint NumAttribs = 0;
+
+
+/**
+ * Config file info.
+ */
+struct config_file
+{
+   struct name_value
+   {
+      char name[100];
+      float value[4];
+      int type;
+   } uniforms[100];
+
+   int num_uniforms;
+};
+
+
+static GLint win = 0;
+static GLboolean Anim = GL_FALSE;
+static GLfloat TexRot = 0.0;
+static GLfloat xRot = 0.0f, yRot = 0.0f, zRot = 0.0f;
+static shape Object = SPHERE;
+
+
+static float
+RandomFloat(float min, float max)
+{
+   int k = rand() % 10000;
+   float x = min + (max - min) * k / 10000.0;
+   return x;
+}
+
+
+/** Set new random values for uniforms */
+static void
+RandomUniformValues(void)
+{
+   GLuint i;
+   for (i = 0; i < NumUniforms; i++) {
+      if (Uniforms[i].type == GL_FLOAT) {
+         Uniforms[i].value[0] = RandomFloat(0.0, 1.0);
+      }
+      else {
+         Uniforms[i].value[0] = RandomFloat(-1.0, 2.0);
+         Uniforms[i].value[1] = RandomFloat(-1.0, 2.0);
+         Uniforms[i].value[2] = RandomFloat(-1.0, 2.0);
+         Uniforms[i].value[3] = RandomFloat(-1.0, 2.0);
+      }
+   }
+}
+
+
+static void
+Idle(void)
+{
+   yRot += 2.0;
+   if (yRot > 360.0)
+      yRot -= 360.0;
+   glutPostRedisplay();
+}
+
+
+
+static void
+SquareVertex(GLfloat s, GLfloat t, GLfloat size)
+{
+   GLfloat x = -size + s * 2.0 * size;
+   GLfloat y = -size + t * 2.0 * size;
+   GLuint i;
+
+   glMultiTexCoord2f(GL_TEXTURE0, s, t);
+   glMultiTexCoord2f(GL_TEXTURE1, s, t);
+   glMultiTexCoord2f(GL_TEXTURE2, s, t);
+   glMultiTexCoord2f(GL_TEXTURE3, s, t);
+
+   /* assign (s,t) to the generic attributes */
+   for (i = 0; i < NumAttribs; i++) {
+      if (Attribs[i].location >= 0) {
+         glVertexAttrib2f(Attribs[i].location, s, t);
+      }
+   }
+
+   glVertex2f(x, y);
+}
+
+
+/*
+ * Draw a square, specifying normal and tangent vectors.
+ */
+static void
+Square(GLfloat size)
+{
+   GLint tangentAttrib = 1;
+   glNormal3f(0, 0, 1);
+   glVertexAttrib3f(tangentAttrib, 1, 0, 0);
+   glBegin(GL_POLYGON);
+#if 1
+   SquareVertex(0, 0, size);
+   SquareVertex(1, 0, size);
+   SquareVertex(1, 1, size);
+   SquareVertex(0, 1, size);
+#else
+   glTexCoord2f(0, 0);  glVertex2f(-size, -size);
+   glTexCoord2f(1, 0);  glVertex2f( size, -size);
+   glTexCoord2f(1, 1);  glVertex2f( size,  size);
+   glTexCoord2f(0, 1);  glVertex2f(-size,  size);
+#endif
+   glEnd();
+}
+
+
+static void
+Cube(GLfloat size)
+{
+   /* +X */
+   glPushMatrix();
+   glRotatef(90, 0, 1, 0);
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+
+   /* -X */
+   glPushMatrix();
+   glRotatef(-90, 0, 1, 0);
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+
+   /* +Y */
+   glPushMatrix();
+   glRotatef(90, 1, 0, 0);
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+
+   /* -Y */
+   glPushMatrix();
+   glRotatef(-90, 1, 0, 0);
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+
+
+   /* +Z */
+   glPushMatrix();
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+
+   /* -Z */
+   glPushMatrix();
+   glRotatef(180, 0, 1, 0);
+   glTranslatef(0, 0, size);
+   Square(size);
+   glPopMatrix();
+}
+
+
+static void
+Sphere(GLfloat radius, GLint slices, GLint stacks)
+{
+   static GLUquadricObj *q = NULL;
+
+   if (!q) {
+      q = gluNewQuadric();
+      gluQuadricDrawStyle(q, GLU_FILL);
+      gluQuadricNormals(q, GLU_SMOOTH);
+      gluQuadricTexture(q, GL_TRUE);
+   }
+
+   gluSphere(q, radius, slices, stacks);
+}
+
+
+static void
+Redisplay(void)
+{
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+   
+   glPushMatrix();
+   glRotatef(xRot, 1.0f, 0.0f, 0.0f);
+   glRotatef(yRot, 0.0f, 1.0f, 0.0f);
+   glRotatef(zRot, 0.0f, 0.0f, 1.0f);
+
+   glMatrixMode(GL_TEXTURE);
+   glLoadIdentity();
+   glRotatef(TexRot, 0.0f, 1.0f, 0.0f);
+   glMatrixMode(GL_MODELVIEW);
+
+   if (Object == SPHERE) {
+      Sphere(2.0, 20, 10);
+   }
+   else if (Object == CUBE) {
+      Cube(2.0);
+   }
+
+   glPopMatrix();
+
+   glutSwapBuffers();
+}
+
+
+static void
+Reshape(int width, int height)
+{
+   glViewport(0, 0, width, height);
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glFrustum(-1.0, 1.0, -1.0, 1.0, 5.0, 25.0);
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+   glTranslatef(0.0f, 0.0f, -15.0f);
+}
+
+
+static void
+CleanUp(void)
+{
+   glDeleteShader(fragShader);
+   glDeleteShader(vertShader);
+   glDeleteProgram(Program);
+   glutDestroyWindow(win);
+}
+
+
+static void
+Key(unsigned char key, int x, int y)
+{
+   const GLfloat step = 2.0;
+  (void) x;
+  (void) y;
+
+   switch(key) {
+   case 'a':
+      Anim = !Anim;
+      if (Anim)
+         glutIdleFunc(Idle);
+      else
+         glutIdleFunc(NULL);
+      break;
+   case 'z':
+      zRot += step;
+      break;
+   case 'Z':
+      zRot -= step;
+      break;
+   case 'o':
+      Object = (Object + 1) % NUM_SHAPES;
+      break;
+   case 'r':
+      RandomUniformValues();
+      SetUniformValues(Program, Uniforms);
+      PrintUniforms(Uniforms);
+      break;
+   case 27:
+      CleanUp();
+      exit(0);
+      break;
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+SpecialKey(int key, int x, int y)
+{
+   const GLfloat step = 2.0;
+
+  (void) x;
+  (void) y;
+
+   switch(key) {
+   case GLUT_KEY_UP:
+      xRot += step;
+      break;
+   case GLUT_KEY_DOWN:
+      xRot -= step;
+      break;
+   case GLUT_KEY_LEFT:
+      yRot -= step;
+      break;
+   case GLUT_KEY_RIGHT:
+      yRot += step;
+      break;
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+InitUniforms(const struct config_file *conf,
+             struct uniform_info uniforms[])
+{
+   int i;
+
+   for (i = 0; i < conf->num_uniforms; i++) {
+      int j;
+      for (j = 0; uniforms[j].name; j++) {
+         if (strcmp(uniforms[j].name, conf->uniforms[i].name) == 0) {
+            uniforms[j].type = conf->uniforms[i].type;
+            uniforms[j].value[0] = conf->uniforms[i].value[0];
+            uniforms[j].value[1] = conf->uniforms[i].value[1];
+            uniforms[j].value[2] = conf->uniforms[i].value[2];
+            uniforms[j].value[3] = conf->uniforms[i].value[3];
+         }
+      }
+   }
+}
+
+
+static void
+LoadTexture(GLint unit, const char *texFileName)
+{
+   GLint imgWidth, imgHeight;
+   GLenum imgFormat;
+   GLubyte *image = NULL;
+   GLuint tex;
+   GLenum filter = GL_LINEAR;
+
+   image = LoadRGBImage(texFileName, &imgWidth, &imgHeight, &imgFormat);
+   if (!image) {
+      printf("Couldn't read %s\n", texFileName);
+      exit(1);
+   }
+
+   printf("Load Texture: unit %d: %s %d x %d\n",
+          unit, texFileName, imgWidth, imgHeight);
+
+   glActiveTexture(GL_TEXTURE0 + unit);
+   glGenTextures(1, &tex);
+   glBindTexture(GL_TEXTURE_2D, tex);
+
+   gluBuild2DMipmaps(GL_TEXTURE_2D, 4, imgWidth, imgHeight,
+                     imgFormat, GL_UNSIGNED_BYTE, image);
+   free(image);
+      
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter);
+   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter);
+}
+
+
+static GLenum
+TypeFromName(const char *n)
+{
+   static const struct {
+      const char *name;
+      GLenum type;
+   } types[] = {
+      { "GL_FLOAT", GL_FLOAT },
+      { "GL_FLOAT_VEC2", GL_FLOAT_VEC2 },
+      { "GL_FLOAT_VEC3", GL_FLOAT_VEC3 },
+      { "GL_FLOAT_VEC4", GL_FLOAT_VEC4 },
+      { "GL_INT", GL_INT },
+      { "GL_INT_VEC2", GL_INT_VEC2 },
+      { "GL_INT_VEC3", GL_INT_VEC3 },
+      { "GL_INT_VEC4", GL_INT_VEC4 },
+      { "GL_SAMPLER_2D", GL_SAMPLER_2D },
+      { NULL, 0 }
+   };
+   GLuint i;
+
+   for (i = 0; types[i].name; i++) {
+      if (strcmp(types[i].name, n) == 0)
+         return types[i].type;
+   }
+   abort();
+   return GL_NONE;
+}
+
+
+
+/**
+ * Read a config file.
+ */
+static void
+ReadConfigFile(const char *filename, struct config_file *conf)
+{
+   char line[1000];
+   FILE *f;
+
+   f = fopen(filename, "r");
+   if (!f) {
+      fprintf(stderr, "Unable to open config file %s\n", filename);
+      exit(1);
+   }
+
+   conf->num_uniforms = 0;
+
+   /* ugly but functional parser */
+   while (!feof(f)) {
+      fgets(line, sizeof(line), f);
+      if (!feof(f) && line[0]) {
+         if (strncmp(line, "vs ", 3) == 0) {
+            VertShaderFile = strdup(line + 3);
+            VertShaderFile[strlen(VertShaderFile) - 1] = 0;
+         }
+         else if (strncmp(line, "fs ", 3) == 0) {
+            FragShaderFile = strdup(line + 3);
+            FragShaderFile[strlen(FragShaderFile) - 1] = 0;
+         }
+         else if (strncmp(line, "texture ", 8) == 0) {
+            char texFileName[100];
+            int unit, k;
+            k = sscanf(line + 8, "%d %s", &unit, texFileName);
+            assert(k == 2);
+            LoadTexture(unit, texFileName);
+         }
+         else if (strncmp(line, "uniform ", 8) == 0) {
+            char name[1000], typeName[100];
+            int k;
+            float v1 = 0.0F, v2 = 0.0F, v3 = 0.0F, v4 = 0.0F;
+            GLenum type;
+
+            k = sscanf(line + 8, "%s %s %f %f %f %f", name, typeName,
+                       &v1, &v2, &v3, &v4);
+
+            type = TypeFromName(typeName);
+
+            strcpy(conf->uniforms[conf->num_uniforms].name, name);
+            conf->uniforms[conf->num_uniforms].value[0] = v1;
+            conf->uniforms[conf->num_uniforms].value[1] = v2;
+            conf->uniforms[conf->num_uniforms].value[2] = v3;
+            conf->uniforms[conf->num_uniforms].value[3] = v4;
+            conf->uniforms[conf->num_uniforms].type = type;
+            conf->num_uniforms++;
+         }
+         else {
+            if (strlen(line) > 1) {
+               fprintf(stderr, "syntax error in: %s\n", line);
+               break;
+            }
+         }
+      }
+   }
+
+   fclose(f);
+}
+
+
+static void
+Init(void)
+{
+   struct config_file config;
+   memset(&config, 0, sizeof(config));
+
+   if (ConfigFile)
+      ReadConfigFile(ConfigFile, &config);
+
+   if (!VertShaderFile) {
+      fprintf(stderr, "Error: no vertex shader\n");
+      exit(1);
+   }
+
+   if (!FragShaderFile) {
+      fprintf(stderr, "Error: no fragment shader\n");
+      exit(1);
+   }
+
+   if (!ShadersSupported())
+      exit(1);
+
+   vertShader = CompileShaderFile(GL_VERTEX_SHADER, VertShaderFile);
+   fragShader = CompileShaderFile(GL_FRAGMENT_SHADER, FragShaderFile);
+   Program = LinkShaders(vertShader, fragShader);
+
+   glUseProgram(Program);
+
+   NumUniforms = GetUniforms(Program, Uniforms);
+   if (config.num_uniforms) {
+      InitUniforms(&config, Uniforms);
+   }
+   else {
+      RandomUniformValues();
+   }
+   SetUniformValues(Program, Uniforms);
+   PrintUniforms(Uniforms);
+
+   NumAttribs = GetAttribs(Program, Attribs);
+   PrintAttribs(Attribs);
+
+   //assert(glGetError() == 0);
+
+   glClearColor(0.4f, 0.4f, 0.8f, 0.0f);
+
+   glEnable(GL_DEPTH_TEST);
+
+   glColor3f(1, 0, 0);
+}
+
+
+static void
+Keys(void)
+{
+   printf("Keyboard:\n");
+   printf("       a  Animation toggle\n");
+   printf("       r  Randomize uniform values\n");
+   printf("       o  Change object\n");
+   printf("  arrows  Rotate object\n");
+   printf("     ESC  Exit\n");
+}
+
+
+static void
+Usage(void)
+{
+   printf("Usage:\n");
+   printf("   shtest config.shtest\n");
+   printf("       Run w/ given config file.\n");
+   printf("   shtest --vs vertShader --fs fragShader\n");
+   printf("       Load/compile given shaders.\n");
+}
+
+
+static void
+ParseOptions(int argc, char *argv[])
+{
+   int i;
+
+   if (argc == 1) {
+      Usage();
+      exit(1);
+   }
+
+   for (i = 1; i < argc; i++) {
+      if (strcmp(argv[i], "--fs") == 0) {
+         FragShaderFile = argv[i+1];
+         i++;
+      }
+      else if (strcmp(argv[i], "--vs") == 0) {
+         VertShaderFile = argv[i+1];
+         i++;
+      }
+      else {
+         /* assume the arg is a config file */
+         ConfigFile = argv[i];
+         break;
+      }
+   }
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   glutInit(&argc, argv);
+   glutInitWindowSize(400, 400);
+   glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
+   win = glutCreateWindow(argv[0]);
+   glewInit();
+   glutReshapeFunc(Reshape);
+   glutKeyboardFunc(Key);
+   glutSpecialFunc(SpecialKey);
+   glutDisplayFunc(Redisplay);
+   ParseOptions(argc, argv);
+   Init();
+   Keys();
+   glutMainLoop();
+   return 0;
+}
+
diff --git a/progs/glsl/texaaline.c b/progs/glsl/texaaline.c
index 1f566c86a6a..2481c0f36e0 100644
--- a/progs/glsl/texaaline.c
+++ b/progs/glsl/texaaline.c
@@ -351,7 +351,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(WinWidth, WinHeight);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/texdemo1.c b/progs/glsl/texdemo1.c
index d55f9e7dd97..5b1913a722b 100644
--- a/progs/glsl/texdemo1.c
+++ b/progs/glsl/texdemo1.c
@@ -53,14 +53,14 @@ static int win = 0;
 
 
 static struct uniform_info ReflectUniforms[] = {
-   { "cubeTex",  1, GL_INT, { 0, 0, 0, 0 }, -1 },
-   { "lightPos", 3, GL_FLOAT, { 10, 10, 20, 0 }, -1 },
+   { "cubeTex",  1, GL_SAMPLER_CUBE, { 0, 0, 0, 0 }, -1 },
+   { "lightPos", 1, GL_FLOAT_VEC3, { 10, 10, 20, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
 static struct uniform_info SimpleUniforms[] = {
-   { "tex2d",    1, GL_INT,   { 1, 0, 0, 0 }, -1 },
-   { "lightPos", 3, GL_FLOAT, { 10, 10, 20, 0 }, -1 },
+   { "tex2d",    1, GL_SAMPLER_2D, { 1, 0, 0, 0 }, -1 },
+   { "lightPos", 1, GL_FLOAT_VEC3, { 10, 10, 20, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -382,7 +382,8 @@ CreateProgram(const char *vertProgFile, const char *fragProgFile,
 
    glUseProgram(program);
 
-   InitUniforms(program, uniforms);
+   SetUniformValues(program, uniforms);
+   PrintUniforms(uniforms);
 
    return program;
 }
diff --git a/progs/glsl/toyball.c b/progs/glsl/toyball.c
index 7fe27aebfe6..c502f24077e 100644
--- a/progs/glsl/toyball.c
+++ b/progs/glsl/toyball.c
@@ -24,18 +24,18 @@ static GLuint program;
 
 
 static struct uniform_info Uniforms[] = {
-   { "LightDir",       4, GL_FLOAT, { 0.57737, 0.57735, 0.57735, 0.0 }, -1 },
-   { "HVector",        4, GL_FLOAT, { 0.32506, 0.32506, 0.88808, 0.0 }, -1 },
-   { "BallCenter",     4, GL_FLOAT, { 0.0, 0.0, 0.0, 1.0 }, -1 },
-   { "SpecularColor",  4, GL_FLOAT, { 0.4, 0.4, 0.4, 60.0 }, -1 },
-   { "Red",         4, GL_FLOAT, { 0.6, 0.0, 0.0, 1.0 }, -1 },
-   { "Blue",        4, GL_FLOAT, { 0.0, 0.3, 0.6, 1.0 }, -1 },
-   { "Yellow",      4, GL_FLOAT, { 0.6, 0.5, 0.0, 1.0 }, -1 },
-   { "HalfSpace0",  4, GL_FLOAT, { 1.0, 0.0, 0.0, 0.2 }, -1 },
-   { "HalfSpace1",  4, GL_FLOAT, { 0.309016994, 0.951056516, 0.0, 0.2 }, -1 },
-   { "HalfSpace2",  4, GL_FLOAT, { -0.809016994, 0.587785252, 0.0, 0.2 }, -1 },
-   { "HalfSpace3",  4, GL_FLOAT, { -0.809016994, -0.587785252, 0.0, 0.2 }, -1 },
-   { "HalfSpace4",  4, GL_FLOAT, { 0.309116994, -0.951056516, 0.0, 0.2 }, -1 },
+   { "LightDir",    1, GL_FLOAT_VEC4, { 0.57737, 0.57735, 0.57735, 0.0 }, -1 },
+   { "HVector",     1, GL_FLOAT_VEC4, { 0.32506, 0.32506, 0.88808, 0.0 }, -1 },
+   { "BallCenter",  1, GL_FLOAT_VEC4, { 0.0, 0.0, 0.0, 1.0 }, -1 },
+   { "SpecularColor", 1, GL_FLOAT_VEC4, { 0.4, 0.4, 0.4, 60.0 }, -1 },
+   { "Red",         1, GL_FLOAT_VEC4, { 0.6, 0.0, 0.0, 1.0 }, -1 },
+   { "Blue",        1, GL_FLOAT_VEC4, { 0.0, 0.3, 0.6, 1.0 }, -1 },
+   { "Yellow",      1, GL_FLOAT_VEC4, { 0.6, 0.5, 0.0, 1.0 }, -1 },
+   { "HalfSpace0",  1, GL_FLOAT_VEC4, { 1.0, 0.0, 0.0, 0.2 }, -1 },
+   { "HalfSpace1",  1, GL_FLOAT_VEC4, { 0.309016994, 0.951056516, 0.0, 0.2 }, -1 },
+   { "HalfSpace2",  1, GL_FLOAT_VEC4, { -0.809016994, 0.587785252, 0.0, 0.2 }, -1 },
+   { "HalfSpace3",  1, GL_FLOAT_VEC4, { -0.809016994, -0.587785252, 0.0, 0.2 }, -1 },
+   { "HalfSpace4",  1, GL_FLOAT_VEC4, { 0.309116994, -0.951056516, 0.0, 0.2 }, -1 },
    { "InOrOutInit", 1, GL_FLOAT, { -3.0, 0, 0, 0 }, -1 },
    { "StripeWidth", 1, GL_FLOAT, {  0.3, 0, 0, 0 }, -1 },
    { "FWidth",      1, GL_FLOAT, { 0.005, 0, 0, 0 }, -1 },
@@ -173,7 +173,8 @@ Init(void)
 
    glUseProgram(program);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
+   PrintUniforms(Uniforms);
 
    assert(glGetError() == 0);
 
@@ -204,7 +205,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 400);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/toyball.shtest b/progs/glsl/toyball.shtest
new file mode 100644
index 00000000000..887663abd32
--- /dev/null
+++ b/progs/glsl/toyball.shtest
@@ -0,0 +1,17 @@
+vs CH11-toyball.vert
+fs CH11-toyball.frag
+uniform LightDir      GL_FLOAT_VEC4 0.57737 0.57735 0.57735 0.0
+uniform HVector       GL_FLOAT_VEC4 0.32506 0.32506 0.88808 0.0
+uniform BallCenter    GL_FLOAT_VEC4 0.0 0.0 0.0 1.0
+uniform SpecularColor GL_FLOAT_VEC4 0.4 0.4 0.4 60.0
+uniform Red           GL_FLOAT_VEC4 0.6 0.0 0.0 1.0
+uniform Blue          GL_FLOAT_VEC4 0.0 0.3 0.6 1.0
+uniform Yellow        GL_FLOAT_VEC4 0.6 0.5 0.0 1.0
+uniform HalfSpace0    GL_FLOAT_VEC4 1.0 0.0 0.0 0.2
+uniform HalfSpace1    GL_FLOAT_VEC4 .309016994 0.951056516 0.0 0.2
+uniform HalfSpace2    GL_FLOAT_VEC4 -0.809016994 0.587785252 0.0 0.2
+uniform HalfSpace3    GL_FLOAT_VEC4 -0.809016994 -0.587785252 0.0 0.2
+uniform HalfSpace4    GL_FLOAT_VEC4 .309116994 -0.951056516 0.0 0.2
+uniform InOrOutInit   GL_FLOAT -3.0
+uniform StripeWidth   GL_FLOAT 0.3
+uniform FWidth        GL_FLOAT .005
diff --git a/progs/glsl/trirast.c b/progs/glsl/trirast.c
index f7546f25a2e..53bd91ef976 100644
--- a/progs/glsl/trirast.c
+++ b/progs/glsl/trirast.c
@@ -239,7 +239,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(WinWidth, WinHeight);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/twoside.c b/progs/glsl/twoside.c
index b6c1b477dde..a57484f96cc 100644
--- a/progs/glsl/twoside.c
+++ b/progs/glsl/twoside.c
@@ -285,7 +285,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(WinWidth, WinHeight);
    glutInitDisplayMode(GLUT_RGB | GLUT_DEPTH | GLUT_DOUBLE);
    win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/vert-or-frag-only.c b/progs/glsl/vert-or-frag-only.c
index 81fcab8c5b8..148991ca83e 100644
--- a/progs/glsl/vert-or-frag-only.c
+++ b/progs/glsl/vert-or-frag-only.c
@@ -173,7 +173,6 @@ int
 main(int argc, char *argv[])
 {
    glutInit(&argc, argv);
-   glutInitWindowPosition( 0, 0);
    glutInitWindowSize(400, 200);
    glutInitDisplayMode(GLUT_RGB | GLUT_DOUBLE | GLUT_DEPTH);
    Win = glutCreateWindow(argv[0]);
diff --git a/progs/glsl/vert-tex.c b/progs/glsl/vert-tex.c
index e791a5759a7..4c8bfa587aa 100644
--- a/progs/glsl/vert-tex.c
+++ b/progs/glsl/vert-tex.c
@@ -43,7 +43,7 @@ static GLfloat xRot = -70.0f, yRot = 0.0f, zRot = 0.0f;
 
 /* value[0] = tex unit */
 static struct uniform_info Uniforms[] = {
-   { "tex1",  1, GL_INT, { 0, 0, 0, 0 }, -1 },
+   { "tex1",  1, GL_SAMPLER_2D, { 0, 0, 0, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
diff --git a/progs/tests/SConscript b/progs/tests/SConscript
index b17fa905936..bb6a1d2b8a9 100644
--- a/progs/tests/SConscript
+++ b/progs/tests/SConscript
@@ -72,6 +72,7 @@ progs = [
     'fogcoord',
     'fptest1',
     'fptexture',
+    'getteximage',
     'glutfx',
     'interleave',
     'invert',
diff --git a/progs/tests/floattex.c b/progs/tests/floattex.c
index ad14cacdcbb..39302ce3aff 100644
--- a/progs/tests/floattex.c
+++ b/progs/tests/floattex.c
@@ -33,7 +33,7 @@ static const char *VertShaderText =
    "} \n";
 
 static struct uniform_info Uniforms[] = {
-   { "tex1",  1, GL_INT, { 0, 0, 0, 0 }, -1 },
+   { "tex1",  1, GL_SAMPLER_2D, { 0, 0, 0, 0 }, -1 },
    END_OF_UNIFORMS
 };
 
@@ -189,7 +189,7 @@ CreateProgram(void)
 
    glUseProgram_func(program);
 
-   InitUniforms(program, Uniforms);
+   SetUniformValues(program, Uniforms);
 
    return program;
 }
diff --git a/progs/tests/getprocaddress.c b/progs/tests/getprocaddress.c
index ca66025d2dc..a09ea58e1da 100644
--- a/progs/tests/getprocaddress.c
+++ b/progs/tests/getprocaddress.c
@@ -39,13 +39,2600 @@ typedef void (*generic_func)();
 
 #define EQUAL(X, Y)  (fabs((X) - (Y)) < 0.001)
 
-/**
+/* This macro simplifies the task of querying an extension function
+ * pointer and checking to see whether it resolved.
+ */
+#define DECLARE_GLFUNC_PTR(name,type) \
+   type name = (type) glXGetProcAddressARB((const GLubyte *) "gl" #name)
+
+/********************************************************************
+ * Generic helper functions used by the test functions.
+ */
+
+static void CheckGLError(int line, const char *file, const char *function)
+{
+    int errorCode;
+    glFinish();
+    errorCode  = glGetError();
+    if (errorCode == GL_NO_ERROR) return;
+    while (errorCode != GL_NO_ERROR) {
+	fprintf(stderr, "OpenGL error 0x%x (%s) at line %d of file %s in function %s()\n",
+	    errorCode,
+	    errorCode == GL_INVALID_VALUE? "GL_INVALID_VALUE":
+	    errorCode == GL_INVALID_ENUM? "GL_INVALID_ENUM":
+	    errorCode == GL_INVALID_OPERATION? "GL_INVALID_OPERATION":
+	    errorCode == GL_STACK_OVERFLOW? "GL_STACK_OVERFLOW":
+	    errorCode == GL_STACK_UNDERFLOW? "GL_STACK_UNDERFLOW":
+	    errorCode == GL_OUT_OF_MEMORY? "GL_OUT_OF_MEMORY":
+	    "unknown",
+	    line, file, function);
+	errorCode = glGetError();
+    }
+    fflush(stderr);
+}
+
+static GLboolean 
+compare_bytes(const char *errorLabel, GLuint expectedSize, 
+   const GLubyte *expectedData, GLuint actualSize, const GLubyte *actualData)
+{
+   int i;
+
+   if (expectedSize == actualSize &&
+      memcmp(expectedData, actualData, actualSize) == 0) {
+      /* All is well */
+      return GL_TRUE;
+   }
+
+   /* Trouble; we don't match.  Print out why. */
+   fprintf(stderr, "%s: actual data is not as expected\n", errorLabel);
+   for (i = 0; i <= 1; i++) {
+      const GLubyte *ptr;
+      int size;
+      char *label;
+      int j;
+
+      switch(i) {
+         case 0:
+            label = "expected";
+            size = expectedSize;
+            ptr = expectedData;
+            break;
+         case 1:
+            label = "  actual";
+            size = actualSize;
+            ptr = actualData;
+            break;
+      }
+      
+      fprintf(stderr, "    %s: size %d: {", label, size);
+      for (j = 0; j < size; j++) {
+         fprintf(stderr, "%s0x%02x", j > 0 ? ", " : "", ptr[j]);
+      }
+      fprintf(stderr, "}\n");
+   }
+
+   /* We fail if the data is unexpected. */
+   return GL_FALSE;
+}
+
+
+static GLboolean 
+compare_ints(const char *errorLabel, GLuint expectedSize, 
+   const GLint *expectedData, GLuint actualSize, const GLint *actualData)
+{
+   int i;
+
+   if (expectedSize == actualSize &&
+      memcmp(expectedData, actualData, actualSize*sizeof(*expectedData)) == 0) {
+      /* All is well */
+      return GL_TRUE;
+   }
+
+   /* Trouble; we don't match.  Print out why. */
+   fprintf(stderr, "%s: actual data is not as expected\n", errorLabel);
+   for (i = 0; i <= 1; i++) {
+      const GLint *ptr;
+      int size;
+      char *label;
+      int j;
+
+      switch(i) {
+         case 0:
+            label = "expected";
+            size = expectedSize;
+            ptr = expectedData;
+            break;
+         case 1:
+            label = "  actual";
+            size = actualSize;
+            ptr = actualData;
+            break;
+      }
+      
+      fprintf(stderr, "    %s: size %d: {", label, size);
+      for (j = 0; j < size; j++) {
+         fprintf(stderr, "%s%d", j > 0 ? ", " : "", ptr[j]);
+      }
+      fprintf(stderr, "}\n");
+   }
+
+   /* We fail if the data is unexpected. */
+   return GL_FALSE;
+}
+
+#define MAX_CONVERTED_VALUES 4
+static GLboolean 
+compare_shorts_to_ints(const char *errorLabel, GLuint expectedSize, 
+   const GLshort *expectedData, GLuint actualSize, const GLint *actualData)
+{
+   int i;
+   GLint convertedValues[MAX_CONVERTED_VALUES];
+
+   if (expectedSize > MAX_CONVERTED_VALUES) {
+      fprintf(stderr, "%s: too much data [need %d values, have %d values]\n",
+         errorLabel, expectedSize, MAX_CONVERTED_VALUES);
+      return GL_FALSE;
+   }
+
+   for (i = 0; i < expectedSize; i++) {
+      convertedValues[i] = (GLint) expectedData[i];
+   }
+
+   return compare_ints(errorLabel, expectedSize, convertedValues, 
+      actualSize, actualData);
+}
+
+static GLboolean 
+compare_floats(const char *errorLabel, GLuint expectedSize, 
+   const GLfloat *expectedData, GLuint actualSize, const GLfloat *actualData)
+{
+   int i;
+
+   if (expectedSize == actualSize &&
+      memcmp(expectedData, actualData, actualSize*sizeof(*expectedData)) == 0) {
+      /* All is well */
+      return GL_TRUE;
+   }
+
+   /* Trouble; we don't match.  Print out why. */
+   fprintf(stderr, "%s: actual data is not as expected\n", errorLabel);
+   for (i = 0; i <= 1; i++) {
+      const GLfloat *ptr;
+      int size;
+      char *label;
+      int j;
+
+      switch(i) {
+         case 0:
+            label = "expected";
+            size = expectedSize;
+            ptr = expectedData;
+            break;
+         case 1:
+            label = "  actual";
+            size = actualSize;
+            ptr = actualData;
+            break;
+      }
+      
+      fprintf(stderr, "    %s: size %d: {", label, size);
+      for (j = 0; j < size; j++) {
+         fprintf(stderr, "%s%f", j > 0 ? ", " : "", ptr[j]);
+      }
+      fprintf(stderr, "}\n");
+   }
+
+   /* We fail if the data is unexpected. */
+   return GL_FALSE;
+}
+
+static GLboolean 
+compare_doubles(const char *errorLabel, GLuint expectedSize, 
+   const GLdouble *expectedData, GLuint actualSize, const GLdouble *actualData)
+{
+   int i;
+
+   if (expectedSize == actualSize || 
+      memcmp(expectedData, actualData, actualSize*sizeof(*expectedData)) == 0) {
+      /* All is well */
+      return GL_TRUE;
+   }
+
+   /* Trouble; we don't match.  Print out why. */
+   fprintf(stderr, "%s: actual data is not as expected\n", errorLabel);
+   for (i = 0; i <= 1; i++) {
+      const GLdouble *ptr;
+      int size;
+      char *label;
+      int j;
+
+      switch(i) {
+         case 0:
+            label = "expected";
+            size = expectedSize;
+            ptr = expectedData;
+            break;
+         case 1:
+            label = "  actual";
+            size = actualSize;
+            ptr = actualData;
+            break;
+      }
+      
+      fprintf(stderr, "    %s: size %d: {", label, size);
+      for (j = 0; j < size; j++) {
+         fprintf(stderr, "%s%f", j > 0 ? ", " : "", ptr[j]);
+      }
+      fprintf(stderr, "}\n");
+   }
+
+   /* We fail if the data is unexpected. */
+   return GL_FALSE;
+}
+
+/********************************************************************
+ * Functions to assist with GL_ARB_texture_compressiong testing
+ */
+
+static GLboolean
+check_texture_format_supported(GLenum format)
+{
+   GLint numFormats;
+   GLint *formats;
+   register int i;
+
+   glGetIntegerv(GL_NUM_COMPRESSED_TEXTURE_FORMATS_ARB, &numFormats);
+   formats = malloc(numFormats * sizeof(GLint));
+   if (formats == NULL) {
+      fprintf(stderr, "check_texture_format_supported: could not allocate memory for %d GLints\n", 
+         numFormats);
+      return GL_FALSE;
+   }
+   
+   memset(formats, 0, numFormats * sizeof(GLint));
+   glGetIntegerv(GL_COMPRESSED_TEXTURE_FORMATS_ARB, formats);
+
+   for (i = 0; i < numFormats; i++) {
+      if (formats[i] == format) {
+         free(formats);
+         return GL_TRUE;
+      }
+   }
+
+   /* We didn't find the format we were looking for.  Give an error. */
+#define FORMAT_NAME(x) (\
+   x == GL_COMPRESSED_RGB_FXT1_3DFX ? "GL_COMPRESSED_RGB_FXT1_3DFX" : \
+   x == GL_COMPRESSED_RGBA_FXT1_3DFX ? "GL_COMPRESSED_RGBA_FXT1_3DFX" : \
+   x == GL_COMPRESSED_RGB_S3TC_DXT1_EXT ? "GL_COMPRESSED_RGB_S3TC_DXT1_EXT" : \
+   x == GL_COMPRESSED_RGBA_S3TC_DXT1_EXT ? "GL_COMPRESSED_RGBA_S3TC_DXT1_EXT" : \
+   x == GL_COMPRESSED_RGBA_S3TC_DXT3_EXT ? "GL_COMPRESSED_RGBA_S3TC_DXT3_EXT" : \
+   x == GL_COMPRESSED_RGBA_S3TC_DXT5_EXT ? "GL_COMPRESSED_RGBA_S3TC_DXT5_EXT" : \
+   x == GL_RGB_S3TC ? "GL_RGB_S3TC" : \
+   x == GL_RGB4_S3TC ? "GL_RGB4_S3TC" : \
+   x == GL_RGBA_S3TC ? "GL_RGBA_S3TC" : \
+   x == GL_RGBA4_S3TC ? "GL_RGBA4_S3TC" : \
+   "unknown")
+   fprintf(stderr, "check_texture_format_supported: unsupported format 0x%04x [%s]\n",
+      format, FORMAT_NAME(format));
+   fprintf(stderr, "supported formats:");
+   for (i = 0; i < numFormats; i++) {
+      fprintf(stderr, " 0x%04x [%s]", formats[i], FORMAT_NAME(formats[i]));
+   }
+   fprintf(stderr, "\n");
+   return GL_FALSE;
+}
+
+/* This helper function compresses an RGBA texture and compares it
+ * against the expected compressed data.  It returns GL_TRUE if all
+ * went as expected, or GL_FALSE in the case of error.
+ */
+static GLboolean
+check_texture_compression(const char *message, GLenum dimension,
+   GLint width, GLint height, GLint depth, const GLubyte *texture, 
+   int expectedCompressedSize, const GLubyte *expectedCompressedData)
+{
+   /* These are the data we query about the texture. */
+   GLint isCompressed;
+   GLenum compressedFormat;
+   GLint compressedSize;
+   GLubyte *compressedData;
+
+   /* We need this function pointer to operate. */
+   DECLARE_GLFUNC_PTR(GetCompressedTexImageARB, PFNGLGETCOMPRESSEDTEXIMAGEARBPROC);
+   if (GetCompressedTexImageARB == NULL) {
+      fprintf(stderr, 
+         "%s: could not query GetCompressedTexImageARB function pointer\n",
+         message);
+      return GL_FALSE;
+   }
+
+   /* Verify that we actually have the GL_COMPRESSED_RGBA_S3TC_DXT3_EXT format available. */
+   if (!check_texture_format_supported(GL_COMPRESSED_RGBA_S3TC_DXT3_EXT)) {
+      return GL_FALSE;
+   }
+
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   /* Set up the base image, requesting that the GL library compress it. */
+   switch(dimension) {
+      case GL_TEXTURE_1D:
+         glTexImage1D(GL_TEXTURE_1D, 0, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            width, 0, 
+            GL_RGBA, GL_UNSIGNED_BYTE, texture);
+         break;
+      case GL_TEXTURE_2D:
+         glTexImage2D(GL_TEXTURE_2D, 0, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            width, height, 0, 
+            GL_RGBA, GL_UNSIGNED_BYTE, texture);
+         break;
+      case GL_TEXTURE_3D:
+         glTexImage3D(GL_TEXTURE_3D, 0, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            width, height, depth, 0, 
+            GL_RGBA, GL_UNSIGNED_BYTE, texture);
+         break;
+      default:
+         fprintf(stderr, "%s: unknown dimension 0x%04x.\n", message, dimension);
+         return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Make sure the texture is compressed, and pull it out if it is. */
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_COMPRESSED_ARB, 
+      &isCompressed);
+   if (!isCompressed) {
+      fprintf(stderr, "%s: could not compress GL_COMPRESSED_RGBA_S3TC_DXT3_EXT texture\n",
+         message);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_INTERNAL_FORMAT, 
+      (GLint *)&compressedFormat);
+   if (compressedFormat != GL_COMPRESSED_RGBA_S3TC_DXT3_EXT) {
+      fprintf(stderr, "%s: got internal format 0x%04x, expected GL_COMPRESSED_RGBA_S3TC_DXT3_EXT [0x%04x]\n",
+         __FUNCTION__, compressedFormat, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB, &compressedSize);
+   compressedData = malloc(compressedSize);
+   if (compressedData == NULL) {
+      fprintf(stderr, "%s: could not malloc %d bytes for compressed texture\n",
+         message, compressedSize);
+      return GL_FALSE;
+   }
+   memset(compressedData, 0, compressedSize);
+   (*GetCompressedTexImageARB)(dimension, 0, compressedData);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Compare it to the expected compressed data. The compare_bytes()
+    * call will print out diagnostics in the case of failure.
+    */
+   if (!compare_bytes(message, 
+      expectedCompressedSize, expectedCompressedData,
+      compressedSize, compressedData)) {
+
+      free(compressedData);
+      return GL_FALSE;
+   }
+
+   /* All done.  Free our allocated data and return success. */
+   free(compressedData);
+   return GL_TRUE;
+}
+
+/* We'll use one function to exercise 1D, 2D, and 3D textures. */
+
+/* The test function for compressed 3D texture images requires several
+ * different function pointers that have to be queried.  This function
+ * gets all the function pointers it needs itself, and so is suitable for 
+ * use to test any and all of the incorporated functions.
+ */
+
+static GLboolean
+exercise_CompressedTextures(GLenum dimension)
+{
+   /* Set up a basic (uncompressed) texture.  We're doing a blue/yellow
+    * checkerboard.  The 8x4/32-pixel board is well-suited to S3TC
+    * compression, which works on 4x4 blocks of pixels.
+    */
+#define B 0,0,255,255
+#define Y 255,255,0,255
+#define TEXTURE_WIDTH 16 
+#define TEXTURE_HEIGHT 4
+#define TEXTURE_DEPTH 1
+   static GLubyte texture[TEXTURE_WIDTH*TEXTURE_HEIGHT*TEXTURE_DEPTH*4] = {
+      B, B, Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, Y, Y,
+      B, B, Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, Y, Y,
+      Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, 
+      Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, Y, Y, B, B, 
+   };
+#undef B
+#undef Y
+   GLubyte uncompressedTexture[TEXTURE_WIDTH*TEXTURE_HEIGHT*TEXTURE_DEPTH*4];
+
+   /* We'll use this as a texture subimage. */
+#define R 255,0,0,255
+#define G 0,255,0,255
+#define SUBTEXTURE_WIDTH 4
+#define SUBTEXTURE_HEIGHT 4
+#define SUBTEXTURE_DEPTH 1
+   static GLubyte subtexture[SUBTEXTURE_WIDTH*SUBTEXTURE_HEIGHT*SUBTEXTURE_DEPTH*4] = {
+      G, G, R, R,
+      G, G, R, R,
+      R, R, G, G,
+      R, R, G, G,
+   };
+#undef R
+#undef G
+
+   /* These are the expected compressed textures.  (In the case of
+    * a failed comparison, the test program will print out the
+    * actual compressed data in a format that can be directly used
+    * here, if desired.)  The brave of heart can calculate the compression 
+    * themselves based on the formulae described at:
+    *   http://en.wikipedia.org/wiki/S3_Texture_Compression
+    * In a nutshell, each group of 16 bytes encodes a 4x4 texture block.
+    * The first eight bytes of each group are 4-bit alpha values
+    * for each of the 16 pixels in the texture block.
+    * The next four bytes in each group are LSB-first RGB565 colors; the
+    * first two bytes are identified as the color C0, and the next two
+    * are the color C1.  (Two more colors C2 and C3 will be calculated 
+    * from these, but do not appear in the compression data.)  The
+    * last 4 bytes of the group are sixteen 2-bit indices that, for
+    * each of the 16 pixels in the texture block, select one of the
+    * colors C0, C1, C2, or C3.
+    *
+    * For example, our blue/yellow checkerboard is made up of
+    * four identical 4x4 blocks.  Each of those blocks will
+    * be encoded as: eight bytes of 0xff (16 alpha values, each 0xf),
+    * C0 as the RGB565 color yellow (0xffe0), encoded LSB-first;
+    * C1 as the RGB565 color blue (0x001f), encoded LSB-first;
+    * and 4 bytes of 16 2-bit color indices reflecting the
+    * choice of color for each of the 16 pixels:
+    *     00, 00, 01, 01, = 0x05
+    *     00, 00, 01, 01, = 0x05
+    *     01, 01, 00, 00, = 0x50
+    *     01, 01, 00, 00, = 0x50
+    */
+   static GLubyte compressedTexture[] = {
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50
+   };
+
+   /* The similar calculations for the 4x4 subtexture are left
+    * as an exercise for the reader.
+    */
+   static GLubyte compressedSubTexture[] = {
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0x00, 0xf8, 0xe0, 0x07, 0x05, 0x05, 0x50, 0x50,
+   };
+
+   /* The combined texture replaces the initial blue/yellow
+    * block with the green/red block.  (I'd wanted to do
+    * the more interesting exercise of putting the
+    * green/red block in the middle of the blue/yellow
+    * texture, which is a non-trivial replacement, but
+    * the attempt produces GL_INVALID_OPERATION, showing
+    * that you can only replace whole blocks of 
+    * subimages with S3TC.)  The combined texture looks
+    * like:
+    *      G G R R  B B Y Y  B B Y Y  B B Y Y
+    *      G G R R  B B Y Y  B B Y Y  B B Y Y
+    *      R R G G  Y Y B B  Y Y B B  Y Y B B 
+    *      R R G G  Y Y B B  Y Y B B  Y Y B B 
+    * which encodes just like the green/red block followed
+    * by 3 copies of the yellow/blue block.
+    */
+   static GLubyte compressedCombinedTexture[] = {
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0x00, 0xf8, 0xe0, 0x07, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50,
+      0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+      0xe0, 0xff, 0x1f, 0x00, 0x05, 0x05, 0x50, 0x50
+   };
+
+   /* These are the data we query about the texture. */
+   GLint queryIsCompressed;
+   GLenum queryCompressedFormat;
+   GLint queryCompressedSize;
+   GLubyte queryCompressedData[sizeof(compressedTexture)];
+
+   /* Query the function pointers we need.  We actually won't need most
+    * of these (the "dimension" parameter dictates whether we're testing
+    * 1D, 2D, or 3D textures), but we'll have them all ready just in case.
+    */
+   DECLARE_GLFUNC_PTR(GetCompressedTexImageARB, PFNGLGETCOMPRESSEDTEXIMAGEARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexImage3DARB, PFNGLCOMPRESSEDTEXIMAGE3DARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexSubImage3DARB, PFNGLCOMPRESSEDTEXSUBIMAGE3DARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexImage2DARB, PFNGLCOMPRESSEDTEXIMAGE2DARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexSubImage2DARB, PFNGLCOMPRESSEDTEXSUBIMAGE2DARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexImage1DARB, PFNGLCOMPRESSEDTEXIMAGE1DARBPROC);
+   DECLARE_GLFUNC_PTR(CompressedTexSubImage1DARB, PFNGLCOMPRESSEDTEXSUBIMAGE1DARBPROC);
+
+   /* If the necessary functions are missing, we can't continue */
+   if (GetCompressedTexImageARB == NULL) {
+      fprintf(stderr, "%s: GetCompressedTexImageARB function is missing\n",
+         __FUNCTION__);
+      return GL_FALSE;
+   }
+   switch (dimension) {
+      case GL_TEXTURE_1D:
+         if (CompressedTexImage1DARB == NULL || CompressedTexSubImage1DARB == NULL) {
+            fprintf(stderr, "%s: 1D compressed texture functions are missing\n",
+               __FUNCTION__);
+            return GL_FALSE;
+         };
+         break;
+      case GL_TEXTURE_2D:
+         if (CompressedTexImage2DARB == NULL || CompressedTexSubImage2DARB == NULL) {
+            fprintf(stderr, "%s: 2D compressed texture functions are missing\n",
+               __FUNCTION__);
+            return GL_FALSE;
+         };
+         break;
+      case GL_TEXTURE_3D:
+         if (CompressedTexImage3DARB == NULL || CompressedTexSubImage3DARB == NULL) {
+            fprintf(stderr, "%s: 3D compressed texture functions are missing\n",
+               __FUNCTION__);
+            return GL_FALSE;
+         };
+         break;
+      default:
+         fprintf(stderr, "%s: unknown texture dimension 0x%04x passed.\n",
+            __FUNCTION__, dimension);
+         return GL_FALSE;
+   }
+   
+   /* Check the compression of our base texture image. */
+   if (!check_texture_compression("texture compression", dimension,
+         TEXTURE_WIDTH, TEXTURE_HEIGHT, TEXTURE_DEPTH, texture,
+         sizeof(compressedTexture), compressedTexture)) {
+
+      /* Something's wrong with texture compression.  The function
+       * above will have printed an appropriate error.
+       */
+      return GL_FALSE;
+   }
+
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Do the same for our texture subimage */
+   if (!check_texture_compression("subtexture compression", dimension,
+         SUBTEXTURE_WIDTH, SUBTEXTURE_HEIGHT, SUBTEXTURE_DEPTH, subtexture,
+         sizeof(compressedSubTexture), compressedSubTexture)) {
+
+      /* Something's wrong with texture compression.  The function
+       * above will have printed an appropriate error.
+       */
+      return GL_FALSE;
+   }
+
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Send the base compressed texture down to the hardware. */
+   switch(dimension) {
+      case GL_TEXTURE_3D:
+         (*CompressedTexImage3DARB)(GL_TEXTURE_3D, 0, 
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            TEXTURE_WIDTH, TEXTURE_HEIGHT, TEXTURE_DEPTH, 0, 
+            sizeof(compressedTexture), compressedTexture);
+         break;
+
+      case GL_TEXTURE_2D:
+         (*CompressedTexImage2DARB)(GL_TEXTURE_2D, 0, 
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            TEXTURE_WIDTH, TEXTURE_HEIGHT, 0, 
+            sizeof(compressedTexture), compressedTexture);
+         break;
+
+      case GL_TEXTURE_1D:
+         (*CompressedTexImage1DARB)(GL_TEXTURE_1D, 0, 
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            TEXTURE_WIDTH, 0, 
+            sizeof(compressedTexture), compressedTexture);
+         break;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* For grins, query it to make sure it is as expected. */
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_COMPRESSED_ARB, 
+      &queryIsCompressed);
+   if (!queryIsCompressed) {
+      fprintf(stderr, "%s: compressed texture did not come back as compressed\n",
+         __FUNCTION__);
+      return GL_FALSE;
+   }
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_INTERNAL_FORMAT, 
+      (GLint *)&queryCompressedFormat);
+   if (queryCompressedFormat != GL_COMPRESSED_RGBA_S3TC_DXT3_EXT) {
+      fprintf(stderr, "%s: got internal format 0x%04x, expected GL_COMPRESSED_RGBA_S3TC_DXT3_EXT [0x%04x]\n",
+         __FUNCTION__, queryCompressedFormat, GL_COMPRESSED_RGBA_S3TC_DXT3_EXT);
+      return GL_FALSE;
+   }
+   glGetTexLevelParameteriv(dimension, 0, GL_TEXTURE_COMPRESSED_IMAGE_SIZE_ARB, 
+      &queryCompressedSize);
+   if (queryCompressedSize != sizeof(compressedTexture)) {
+      fprintf(stderr, "%s: compressed 3D texture changed size: expected %d, actual %d\n",
+         __FUNCTION__, sizeof(compressedTexture), queryCompressedSize);
+      return GL_FALSE;
+   }
+   (*GetCompressedTexImageARB)(dimension, 0, queryCompressedData);
+   if (!compare_bytes(
+      "exercise_CompressedTextures:doublechecking compressed texture",
+      sizeof(compressedTexture), compressedTexture,
+      queryCompressedSize, queryCompressedData)) {
+      return GL_FALSE;
+   }
+
+   /* Now apply the texture subimage.  The current implementation of
+    * S3TC requires that subimages be only applied to whole blocks.
+    */
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   switch(dimension) {
+      case GL_TEXTURE_3D:
+         (*CompressedTexSubImage3DARB)(GL_TEXTURE_3D, 0, 
+            0, 0, 0, /* offsets */
+            SUBTEXTURE_WIDTH, SUBTEXTURE_HEIGHT, SUBTEXTURE_DEPTH,
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            sizeof(compressedSubTexture), compressedSubTexture);
+         break;
+      case GL_TEXTURE_2D:
+         (*CompressedTexSubImage2DARB)(GL_TEXTURE_2D, 0, 
+            0, 0, /* offsets */
+            SUBTEXTURE_WIDTH, SUBTEXTURE_HEIGHT,
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            sizeof(compressedSubTexture), compressedSubTexture);
+         break;
+      case GL_TEXTURE_1D:
+         (*CompressedTexSubImage2DARB)(GL_TEXTURE_2D, 0, 
+            0, 0, /* offsets */
+            SUBTEXTURE_WIDTH, SUBTEXTURE_HEIGHT,
+            GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, 
+            sizeof(compressedSubTexture), compressedSubTexture);
+         break;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query the compressed texture back now, and see that it
+    * is as expected.
+    */
+   (*GetCompressedTexImageARB)(dimension, 0, queryCompressedData);
+   if (!compare_bytes("exercise_CompressedTextures:combined texture",
+      sizeof(compressedCombinedTexture), compressedCombinedTexture,
+      queryCompressedSize, queryCompressedData)) {
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Just for the exercise, uncompress the texture and pull it out. 
+    * We don't check it because the compression is lossy, so it won't
+    * compare exactly to the source texture; we just 
+    * want to exercise the code paths that convert it.
+    */
+   glGetTexImage(dimension, 0, GL_RGBA, GL_UNSIGNED_BYTE, uncompressedTexture);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* If we survived this far, we pass. */
+   return GL_TRUE;
+}
+
+/**************************************************************************
+ * Functions to assist with GL_EXT_framebuffer_object and
+ * GL_EXT_framebuffer_blit testing.
+ */
+
+#define FB_STATUS_NAME(x) (\
+   x == GL_FRAMEBUFFER_COMPLETE_EXT ? "GL_FRAMEBUFFER_COMPLETE_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER_EXT" : \
+   x == GL_FRAMEBUFFER_UNSUPPORTED_EXT ? "GL_FRAMEBUFFER_UNSUPPORTED_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_FORMATS_EXT" : \
+   x == GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT ? "GL_FRAMEBUFFER_INCOMPLETE_DIMENSIONS_EXT" : \
+   "unknown")
+
+static GLboolean
+exercise_framebuffer(void)
+{
+   GLuint framebufferID = 0;
+   GLuint renderbufferID = 0;
+   
+   /* Dimensions of the framebuffer and renderbuffers are arbitrary.
+    * Since they won't be shown on-screen, we can use whatever we want.
+    */
+   const GLint Width = 100;
+   const GLint Height = 100;
+
+   /* Every function we use will be referenced through function pointers.
+    * This will allow this test program to run on OpenGL implementations
+    * that *don't* implement these extensions (though the implementation
+    * used to compile them must have up-to-date header files).
+    */
+   DECLARE_GLFUNC_PTR(GenFramebuffersEXT, PFNGLGENFRAMEBUFFERSEXTPROC);
+   DECLARE_GLFUNC_PTR(IsFramebufferEXT, PFNGLISFRAMEBUFFEREXTPROC);
+   DECLARE_GLFUNC_PTR(DeleteFramebuffersEXT, PFNGLDELETEFRAMEBUFFERSEXTPROC);
+   DECLARE_GLFUNC_PTR(BindFramebufferEXT, PFNGLBINDFRAMEBUFFEREXTPROC);
+   DECLARE_GLFUNC_PTR(GenRenderbuffersEXT, PFNGLGENRENDERBUFFERSEXTPROC);
+   DECLARE_GLFUNC_PTR(IsRenderbufferEXT, PFNGLISRENDERBUFFEREXTPROC);
+   DECLARE_GLFUNC_PTR(DeleteRenderbuffersEXT, PFNGLDELETERENDERBUFFERSEXTPROC);
+   DECLARE_GLFUNC_PTR(BindRenderbufferEXT, PFNGLBINDRENDERBUFFEREXTPROC);
+   DECLARE_GLFUNC_PTR(FramebufferRenderbufferEXT, PFNGLFRAMEBUFFERRENDERBUFFEREXTPROC);
+   DECLARE_GLFUNC_PTR(RenderbufferStorageEXT, PFNGLRENDERBUFFERSTORAGEEXTPROC);
+   DECLARE_GLFUNC_PTR(CheckFramebufferStatusEXT, PFNGLCHECKFRAMEBUFFERSTATUSEXTPROC);
+
+   /* The BlitFramebuffer function comes from a different extension.
+    * It's possible for an implementation to implement all the above,
+    * but not BlitFramebuffer; so it's okay if this one comes back
+    * NULL, as we can still test the rest.
+    */
+   DECLARE_GLFUNC_PTR(BlitFramebufferEXT, PFNGLBLITFRAMEBUFFEREXTPROC);
+
+   /* We cannot test unless we have all the function pointers. */
+   if (
+      GenFramebuffersEXT == NULL ||
+      IsFramebufferEXT == NULL || 
+      DeleteFramebuffersEXT == NULL ||
+      BindFramebufferEXT == NULL ||
+      GenRenderbuffersEXT == NULL ||
+      IsRenderbufferEXT == NULL ||
+      DeleteRenderbuffersEXT == NULL ||
+      BindRenderbufferEXT == NULL ||
+      FramebufferRenderbufferEXT == NULL ||
+      RenderbufferStorageEXT == NULL ||
+      CheckFramebufferStatusEXT == NULL
+   ) {
+      fprintf(stderr, "%s: could not locate all framebuffer functions\n",
+         __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   /* Generate a framebuffer for us to play with. */
+   (*GenFramebuffersEXT)(1, &framebufferID);
+   if (framebufferID == 0) {
+      fprintf(stderr, "%s: failed to generate a frame buffer ID.\n",
+         __FUNCTION__);
+      return GL_FALSE;
+   }
+   /* The generated name is not a framebuffer object until bound. */
+   (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, framebufferID);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   if (!(*IsFramebufferEXT)(framebufferID)) {
+      fprintf(stderr, "%s: generated a frame buffer ID 0x%x that wasn't a framebuffer\n",
+         __FUNCTION__, framebufferID);
+      (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+      (*DeleteFramebuffersEXT)(1, &framebufferID);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   {
+      GLint queriedFramebufferID;
+      glGetIntegerv(GL_FRAMEBUFFER_BINDING_EXT, &queriedFramebufferID);
+      if (queriedFramebufferID != framebufferID) {
+         fprintf(stderr, "%s: bound frame buffer 0x%x, but queried 0x%x\n",
+            __FUNCTION__, framebufferID, queriedFramebufferID);
+         (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+         (*DeleteFramebuffersEXT)(1, &framebufferID);
+         return GL_FALSE;
+      }
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Create a color buffer to attach to the frame buffer object, so
+    * we can actually operate on it.  We go through the same basic checks
+    * with the renderbuffer that we do with the framebuffer.
+    */
+   (*GenRenderbuffersEXT)(1, &renderbufferID);
+   if (renderbufferID == 0) {
+      fprintf(stderr, "%s: could not generate a renderbuffer ID\n",
+         __FUNCTION__);
+      (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+      (*DeleteFramebuffersEXT)(1, &framebufferID);
+      return GL_FALSE;
+   }
+   (*BindRenderbufferEXT)(GL_RENDERBUFFER_EXT, renderbufferID);
+   if (!(*IsRenderbufferEXT)(renderbufferID)) {
+      fprintf(stderr, "%s: generated renderbuffer 0x%x is not a renderbuffer\n",
+         __FUNCTION__, renderbufferID);
+      (*BindRenderbufferEXT)(GL_RENDERBUFFER_EXT, 0);
+      (*DeleteRenderbuffersEXT)(1, &renderbufferID);
+      (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+      (*DeleteFramebuffersEXT)(1, &framebufferID);
+      return GL_FALSE;
+   }
+   {
+      GLint queriedRenderbufferID = 0;
+      glGetIntegerv(GL_RENDERBUFFER_BINDING_EXT, &queriedRenderbufferID);
+      if (renderbufferID != queriedRenderbufferID) {
+         fprintf(stderr, "%s: bound renderbuffer 0x%x, but got 0x%x\n",
+            __FUNCTION__, renderbufferID, queriedRenderbufferID);
+         (*BindRenderbufferEXT)(GL_RENDERBUFFER_EXT, 0);
+         (*DeleteRenderbuffersEXT)(1, &renderbufferID);
+         (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+         (*DeleteFramebuffersEXT)(1, &framebufferID);
+         return GL_FALSE;
+      }
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Add the renderbuffer as a color attachment to the current
+    * framebuffer (which is our generated framebuffer).
+    */
+   (*FramebufferRenderbufferEXT)(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT1_EXT,
+      GL_RENDERBUFFER_EXT, renderbufferID);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* The renderbuffer will need some dimensions and storage space. */
+   (*RenderbufferStorageEXT)(GL_RENDERBUFFER_EXT, GL_RGB, Width, Height);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* That should be everything we need.  If we set up to draw and to
+    * read from our color attachment, we should be "framebuffer complete",
+    * meaning the framebuffer is ready to go.
+    */
+   glDrawBuffer(GL_COLOR_ATTACHMENT1_EXT);
+   glReadBuffer(GL_COLOR_ATTACHMENT1_EXT);
+   {
+      GLenum status = (*CheckFramebufferStatusEXT)(GL_FRAMEBUFFER_EXT);
+      if (status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+         fprintf(stderr, "%s: framebuffer not complete; status = %s [0x%x]\n",
+            __FUNCTION__, FB_STATUS_NAME(status), status);
+         glReadBuffer(0);
+         glDrawBuffer(0);
+         (*BindRenderbufferEXT)(GL_RENDERBUFFER_EXT, 0);
+         (*DeleteRenderbuffersEXT)(1, &renderbufferID);
+         (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+         (*DeleteFramebuffersEXT)(1, &framebufferID);
+         return GL_FALSE;
+      }
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Define the contents of the frame buffer */
+   glClearColor(0.5, 0.5, 0.5, 0.0);
+   glClear(GL_COLOR_BUFFER_BIT);
+
+   /* If the GL_EXT_framebuffer_blit is supported, attempt a framebuffer
+    * blit from (5,5)-(10,10) to (90,90)-(95,95).  This is *not* an
+    * error if framebuffer_blit is *not* supported (as we can still
+    * effectively test the other functions).
+    */
+   if (BlitFramebufferEXT != NULL) {
+      (*BlitFramebufferEXT)(5, 5, 10, 10, 90, 90, 95, 95,
+         GL_COLOR_BUFFER_BIT, GL_NEAREST);
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* We could now test to see whether the framebuffer had the desired
+    * contents.  As this is just a touch test, we'll leave that for now.
+    * Clean up and go home.
+    */
+   glReadBuffer(0);
+   glDrawBuffer(0);
+   (*BindRenderbufferEXT)(GL_RENDERBUFFER_EXT, 0);
+   (*DeleteRenderbuffersEXT)(1, &renderbufferID);
+   (*BindFramebufferEXT)(GL_FRAMEBUFFER_EXT, 0);
+   (*DeleteFramebuffersEXT)(1, &framebufferID);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   return GL_TRUE;
+}
+
+/**************************************************************************
+ * Functions to assist with GL_ARB_shader_objects testing.
+ */
+
+static void
+print_info_log(const char *message, GLhandleARB object)
+{
+   DECLARE_GLFUNC_PTR(GetObjectParameterivARB, PFNGLGETOBJECTPARAMETERIVARBPROC);
+   DECLARE_GLFUNC_PTR(GetInfoLogARB, PFNGLGETINFOLOGARBPROC);
+   int logLength, queryLength;
+   char *log;
+
+   if (GetObjectParameterivARB == NULL) {
+      fprintf(stderr, "%s: could not get GetObjectParameterivARB address\n",
+         message);
+      return;
+   }
+   if (GetInfoLogARB == NULL) {
+      fprintf(stderr, "%s: could not get GetInfoLogARB address\n",
+         message);
+      return;
+   }
+
+   (*GetObjectParameterivARB)(object, GL_OBJECT_INFO_LOG_LENGTH_ARB, 
+      &logLength);
+   if (logLength == 0) {
+      fprintf(stderr, "%s: info log length is 0\n", message);
+      return;
+   }
+   log = malloc(logLength);
+   if (log == NULL) {
+      fprintf(stderr, "%s: could not malloc %d bytes for info log\n",
+         message, logLength);
+   }
+   else {
+      (*GetInfoLogARB)(object, logLength, &queryLength, log);
+      fprintf(stderr, "%s: info log says '%s'\n",
+         message, log);
+   }
+   free(log);
+}
+
+static GLboolean
+exercise_uniform_start(const char *fragmentShaderText, const char *uniformName,
+   GLhandleARB *returnProgram, GLint *returnUniformLocation)
+{
+   DECLARE_GLFUNC_PTR(CreateShaderObjectARB, PFNGLCREATESHADEROBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(ShaderSourceARB, PFNGLSHADERSOURCEARBPROC);
+   DECLARE_GLFUNC_PTR(CompileShaderARB, PFNGLCOMPILESHADERARBPROC);
+   DECLARE_GLFUNC_PTR(CreateProgramObjectARB, PFNGLCREATEPROGRAMOBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(AttachObjectARB, PFNGLATTACHOBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(LinkProgramARB, PFNGLLINKPROGRAMARBPROC);
+   DECLARE_GLFUNC_PTR(UseProgramObjectARB, PFNGLUSEPROGRAMOBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(ValidateProgramARB, PFNGLVALIDATEPROGRAMARBPROC);
+   DECLARE_GLFUNC_PTR(GetUniformLocationARB, PFNGLGETUNIFORMLOCATIONARBPROC);
+   DECLARE_GLFUNC_PTR(DeleteObjectARB, PFNGLDELETEOBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(GetObjectParameterivARB, PFNGLGETOBJECTPARAMETERIVARBPROC);
+   GLhandleARB fs, program;
+   GLint uniformLocation;
+   GLint shaderCompiled, programValidated;
+
+   if (CreateShaderObjectARB == NULL ||
+       ShaderSourceARB == NULL ||
+       CompileShaderARB == NULL ||
+       CreateProgramObjectARB == NULL ||
+       AttachObjectARB == NULL ||
+       LinkProgramARB == NULL ||
+       UseProgramObjectARB == NULL ||
+       ValidateProgramARB == NULL ||
+       GetUniformLocationARB == NULL ||
+       DeleteObjectARB == NULL ||
+       GetObjectParameterivARB == NULL ||
+       0) {
+      return GL_FALSE;
+   }
+
+   /* Create the trivial fragment shader and program.  For safety
+    * we'll check to make sure they compile and link correctly.
+    */
+   fs = (*CreateShaderObjectARB)(GL_FRAGMENT_SHADER_ARB);
+   (*ShaderSourceARB)(fs, 1, &fragmentShaderText, NULL);
+   (*CompileShaderARB)(fs);
+   (*GetObjectParameterivARB)(fs, GL_OBJECT_COMPILE_STATUS_ARB,
+      &shaderCompiled);
+   if (!shaderCompiled) {
+      print_info_log("shader did not compile", fs);
+      (*DeleteObjectARB)(fs);
+      CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   program = (*CreateProgramObjectARB)();
+   (*AttachObjectARB)(program, fs);
+   (*LinkProgramARB)(program);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Make sure we're going to run successfully */
+   (*ValidateProgramARB)(program);
+   (*GetObjectParameterivARB)(program, GL_OBJECT_VALIDATE_STATUS_ARB, 
+      &programValidated);
+   if (!programValidated) {; 
+      print_info_log("program did not validate", program);
+      (*DeleteObjectARB)(program);
+      (*DeleteObjectARB)(fs);
+      CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   /* Put the program in place.  We're not allowed to assign to uniform
+    * variables used by the program until the program is put into use.
+    */
+   (*UseProgramObjectARB)(program);
+
+   /* Once the shader is in place, we're free to delete it; this
+    * won't affect the copy that's part of the program.
+    */
+   (*DeleteObjectARB)(fs);
+
+   /* Find the location index of the uniform variable we declared;
+    * the caller will ned that to set the value.
+    */
+   uniformLocation = (*GetUniformLocationARB)(program, uniformName);
+   if (uniformLocation == -1) {
+      fprintf(stderr, "%s: could not determine uniform location\n",
+         __FUNCTION__);
+      (*DeleteObjectARB)(program);
+      CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   /* All done with what we're supposed to do - return the program
+    * handle and the uniform location to the caller.
+    */
+   *returnProgram = program;
+   *returnUniformLocation = uniformLocation;
+   return GL_TRUE;
+}
+
+static void
+exercise_uniform_end(GLhandleARB program)
+{
+   DECLARE_GLFUNC_PTR(UseProgramObjectARB, PFNGLUSEPROGRAMOBJECTARBPROC);
+   DECLARE_GLFUNC_PTR(DeleteObjectARB, PFNGLDELETEOBJECTARBPROC);
+   if (UseProgramObjectARB == NULL || DeleteObjectARB == NULL) {
+      return;
+   }
+
+   /* Turn off our program by setting the special value 0, and
+    * then delete the program object.
+    */
+   (*UseProgramObjectARB)(0);
+   (*DeleteObjectARB)(program);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+}
+
+/**************************************************************************
+ * Exercises for fences
+ */
+static GLboolean
+exercise_fences(void)
+{
+   DECLARE_GLFUNC_PTR(DeleteFencesNV, PFNGLDELETEFENCESNVPROC);
+   DECLARE_GLFUNC_PTR(FinishFenceNV, PFNGLFINISHFENCENVPROC);
+   DECLARE_GLFUNC_PTR(GenFencesNV, PFNGLGENFENCESNVPROC);
+   DECLARE_GLFUNC_PTR(GetFenceivNV, PFNGLGETFENCEIVNVPROC);
+   DECLARE_GLFUNC_PTR(IsFenceNV, PFNGLISFENCENVPROC);
+   DECLARE_GLFUNC_PTR(SetFenceNV, PFNGLSETFENCENVPROC);
+   DECLARE_GLFUNC_PTR(TestFenceNV, PFNGLTESTFENCENVPROC);
+   GLuint fence;
+   GLint fenceStatus, fenceCondition;
+   int count;
+
+   /* Make sure we have all the function pointers we need. */
+   if (GenFencesNV == NULL ||
+      SetFenceNV == NULL ||
+      IsFenceNV == NULL ||
+      GetFenceivNV == NULL ||
+      TestFenceNV == NULL ||
+      FinishFenceNV == NULL ||
+      DeleteFencesNV == NULL) {
+      fprintf(stderr, "%s: don't have all the fence functions\n",
+         __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   /* Create and set a simple fence. */
+   (*GenFencesNV)(1, &fence);
+   (*SetFenceNV)(fence, GL_ALL_COMPLETED_NV);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Make sure it reads as a fence. */
+   if (!(*IsFenceNV)(fence)) {
+      fprintf(stderr, "%s: set fence is not a fence\n", __FUNCTION__);
+      (*DeleteFencesNV)(1, &fence);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Try to read back its current status and condition. */
+   (*GetFenceivNV)(fence, GL_FENCE_CONDITION_NV, &fenceCondition);
+   if (fenceCondition != GL_ALL_COMPLETED_NV) {
+      fprintf(stderr, "%s: expected fence condition 0x%x, got 0x%x\n",
+         __FUNCTION__, GL_ALL_COMPLETED_NV, fenceCondition);
+      (*DeleteFencesNV)(1, &fence);
+      return GL_FALSE;
+   }
+   (*GetFenceivNV)(fence, GL_FENCE_STATUS_NV, &fenceStatus);
+   if (fenceStatus != GL_TRUE && fenceStatus != GL_FALSE) {
+      fprintf(stderr,"%s: fence status should be GL_TRUE or GL_FALSE, got 0x%x\n",
+         __FUNCTION__, fenceStatus);
+      (*DeleteFencesNV)(1, &fence);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Set the fence again, query its status, and wait for it to finish
+    * two different ways: once by looping on TestFence(), and a 
+    * second time by a simple call to FinishFence();
+    */
+   (*SetFenceNV)(fence, GL_ALL_COMPLETED_NV);
+   glFlush();
+   count = 1;
+   while (!(*TestFenceNV)(fence)) {
+      count++;
+      if (count == 0) {
+         break;
+      }
+   }
+   if (count == 0) {
+      fprintf(stderr, "%s: fence never returned true\n", __FUNCTION__);
+      (*DeleteFencesNV)(1, &fence);
+      return GL_FALSE;
+   }
+   (*SetFenceNV)(fence, GL_ALL_COMPLETED_NV);
+   (*FinishFenceNV)(fence);
+   if ((*TestFenceNV)(fence) != GL_TRUE) {
+      fprintf(stderr, "%s: finished fence does not have status GL_TRUE\n",
+         __FUNCTION__);
+      (*DeleteFencesNV)(1, &fence);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* All done.  Delete the fence and return. */
+   (*DeleteFencesNV)(1, &fence);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   return GL_TRUE;
+}
+
+/**************************************************************************
+ * Exercises for buffer objects
+ */
+enum Map_Buffer_Usage{ Use_Map_Buffer, Use_Map_Buffer_Range};
+static GLboolean
+exercise_buffer_objects(enum Map_Buffer_Usage usage)
+{
+#define BUFFER_DATA_SIZE 1024
+   GLuint bufferID;
+   GLint bufferMapped;
+   static GLubyte data[BUFFER_DATA_SIZE] = {0};
+   float *dataPtr;
+
+   /* Get the function pointers we need.  These are from
+    * GL_ARB_vertex_buffer_object and are required in all
+    * cases.
+    */
+   DECLARE_GLFUNC_PTR(GenBuffersARB, PFNGLGENBUFFERSARBPROC);
+   DECLARE_GLFUNC_PTR(BindBufferARB, PFNGLBINDBUFFERARBPROC);
+   DECLARE_GLFUNC_PTR(BufferDataARB, PFNGLBUFFERDATAARBPROC);
+   DECLARE_GLFUNC_PTR(MapBufferARB, PFNGLMAPBUFFERARBPROC);
+   DECLARE_GLFUNC_PTR(UnmapBufferARB, PFNGLUNMAPBUFFERARBPROC);
+   DECLARE_GLFUNC_PTR(DeleteBuffersARB, PFNGLDELETEBUFFERSARBPROC);
+   DECLARE_GLFUNC_PTR(GetBufferParameterivARB, PFNGLGETBUFFERPARAMETERIVARBPROC);
+
+   /* These are from GL_ARB_map_buffer_range, and are optional
+    * unless we're given Use_Map_Buffer_Range.  Note that they do *not*
+    * have the standard "ARB" suffixes; this is because the extension
+    * was introduced *after* a superset was standardized in OpenGL 3.0.
+    * (The extension really only exists to allow the functionality on
+    * devices that cannot implement a full OpenGL 3.0 driver.)
+    */
+   DECLARE_GLFUNC_PTR(FlushMappedBufferRange, PFNGLFLUSHMAPPEDBUFFERRANGEPROC);
+   DECLARE_GLFUNC_PTR(MapBufferRange, PFNGLMAPBUFFERRANGEPROC);
+   
+   /* This is from APPLE_flush_buffer_range, and is optional even if
+    * we're given Use_Map_Buffer_Range.  Test it before using it.
+    */
+   DECLARE_GLFUNC_PTR(BufferParameteriAPPLE, PFNGLBUFFERPARAMETERIAPPLEPROC);
+
+   /* Make sure we have all the function pointers we need. */
+   if (GenBuffersARB == NULL ||
+      BindBufferARB == NULL ||
+      BufferDataARB == NULL ||
+      MapBufferARB == NULL ||
+      UnmapBufferARB == NULL ||
+      DeleteBuffersARB == NULL ||
+      GetBufferParameterivARB == NULL) {
+      fprintf(stderr, "%s: missing basic MapBuffer functions\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+   if (usage == Use_Map_Buffer_Range) {
+      if (FlushMappedBufferRange == NULL || MapBufferRange == NULL) {
+         fprintf(stderr, "%s: missing MapBufferRange functions\n", __FUNCTION__);
+         return GL_FALSE;
+      }
+   }
+
+   /* Create and define a buffer */
+   (*GenBuffersARB)(1, &bufferID);
+   (*BindBufferARB)(GL_ARRAY_BUFFER_ARB, bufferID);
+   (*BufferDataARB)(GL_ARRAY_BUFFER_ARB, BUFFER_DATA_SIZE, data, 
+      GL_DYNAMIC_DRAW_ARB);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* If we're using MapBufferRange, and if the BufferParameteriAPPLE
+    * function is present, use it before mapping.  This particular
+    * use is a no-op, intended just to exercise the entry point.
+    */
+   if (usage == Use_Map_Buffer_Range && BufferParameteriAPPLE != NULL) {
+      (*BufferParameteriAPPLE)(GL_ARRAY_BUFFER_ARB, 
+         GL_BUFFER_SERIALIZED_MODIFY_APPLE, GL_TRUE);
+   }
+
+   /* Map it, and make sure it's mapped. */
+   switch(usage) {
+      case Use_Map_Buffer:
+         dataPtr = (float *) (*MapBufferARB)(
+            GL_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
+         break;
+      case Use_Map_Buffer_Range:
+         dataPtr = (float *)(*MapBufferRange)(GL_ARRAY_BUFFER_ARB,
+            4, 16, GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT);
+         break;
+   }
+   if (dataPtr == NULL) {
+      fprintf(stderr, "%s: %s returned NULL\n", __FUNCTION__,
+         usage == Use_Map_Buffer ? "MapBuffer" : "MapBufferRange");
+      (*BindBufferARB)(GL_ARRAY_BUFFER_ARB, 0);
+      (*DeleteBuffersARB)(1, &bufferID);
+      return GL_FALSE;
+   }
+   (*GetBufferParameterivARB)(GL_ARRAY_BUFFER_ARB, GL_BUFFER_MAPPED_ARB, 
+      &bufferMapped);
+   if (!bufferMapped) {
+      fprintf(stderr, "%s: buffer should be mapped but isn't\n", __FUNCTION__);
+      (*BindBufferARB)(GL_ARRAY_BUFFER_ARB, 0);
+      (*DeleteBuffersARB)(1, &bufferID);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Write something to it, just to make sure we don't segfault. */
+   *dataPtr = 1.5;
+
+   /* Unmap to show we're finished with the buffer.  Note that if we're
+    * using MapBufferRange, we first have to flush the range we modified.
+    */
+   if (usage == Use_Map_Buffer_Range) {
+      (*FlushMappedBufferRange)(GL_ARRAY_BUFFER_ARB, 4, 16);
+   }
+   if (!(*UnmapBufferARB)(GL_ARRAY_BUFFER_ARB)) {
+      fprintf(stderr, "%s: UnmapBuffer failed\n", __FUNCTION__);
+      (*BindBufferARB)(GL_ARRAY_BUFFER_ARB, 0);
+      (*DeleteBuffersARB)(1, &bufferID);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* All done. */
+   (*BindBufferARB)(GL_ARRAY_BUFFER_ARB, 0);
+   (*DeleteBuffersARB)(1, &bufferID);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+   return GL_TRUE;
+
+#undef BUFFER_DATA_SIZE
+}
+
+/**************************************************************************
+ * Exercises for occlusion query
+ */
+static GLboolean
+exercise_occlusion_query(void)
+{
+   GLuint queryObject;
+   GLint queryReady;
+   GLuint querySampleCount;
+   GLint queryCurrent;
+   GLint queryCounterBits;
+
+   /* Get the function pointers we need.  These are from
+    * GL_ARB_vertex_buffer_object and are required in all
+    * cases.
+    */
+   DECLARE_GLFUNC_PTR(GenQueriesARB, PFNGLGENQUERIESARBPROC);
+   DECLARE_GLFUNC_PTR(BeginQueryARB, PFNGLBEGINQUERYARBPROC);
+   DECLARE_GLFUNC_PTR(GetQueryivARB, PFNGLGETQUERYIVARBPROC);
+   DECLARE_GLFUNC_PTR(EndQueryARB, PFNGLENDQUERYARBPROC);
+   DECLARE_GLFUNC_PTR(IsQueryARB, PFNGLISQUERYARBPROC);
+   DECLARE_GLFUNC_PTR(GetQueryObjectivARB, PFNGLGETQUERYOBJECTIVARBPROC);
+   DECLARE_GLFUNC_PTR(GetQueryObjectuivARB, PFNGLGETQUERYOBJECTUIVARBPROC);
+   DECLARE_GLFUNC_PTR(DeleteQueriesARB, PFNGLDELETEQUERIESARBPROC);
+
+   /* Make sure we have all the function pointers we need. */
+   if (GenQueriesARB == NULL ||
+      BeginQueryARB == NULL ||
+      GetQueryivARB == NULL ||
+      EndQueryARB == NULL ||
+      IsQueryARB == NULL ||
+      GetQueryObjectivARB == NULL ||
+      GetQueryObjectuivARB == NULL ||
+      DeleteQueriesARB == NULL) {
+      fprintf(stderr, "%s: don't have all the Query functions\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   /* Create a query object, and start a query. */
+   (*GenQueriesARB)(1, &queryObject);
+   (*BeginQueryARB)(GL_SAMPLES_PASSED_ARB, queryObject);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* While we're in the query, check the functions that are supposed
+    * to return which query we're in and how many bits of resolution
+    * we get.
+    */
+   (*GetQueryivARB)(GL_SAMPLES_PASSED_ARB, GL_CURRENT_QUERY_ARB, &queryCurrent);
+   if (queryCurrent != queryObject) {
+      fprintf(stderr, "%s: current query 0x%x != set query 0x%x\n",
+         __FUNCTION__, queryCurrent, queryObject);
+      (*EndQueryARB)(GL_SAMPLES_PASSED_ARB);
+      (*DeleteQueriesARB)(1, &queryObject);
+      return GL_FALSE;
+   }
+   (*GetQueryivARB)(GL_SAMPLES_PASSED_ARB, GL_QUERY_COUNTER_BITS_ARB, 
+      &queryCounterBits);
+   if (queryCounterBits < 1) {
+      fprintf(stderr, "%s: query counter bits is too small (%d)\n",
+         __FUNCTION__, queryCounterBits);
+      (*EndQueryARB)(GL_SAMPLES_PASSED_ARB);
+      (*DeleteQueriesARB)(1, &queryObject);
+      return GL_FALSE;
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Finish up the query.  Since we didn't draw anything, the result
+    * should be 0 passed samples.
+    */
+   (*EndQueryARB)(GL_SAMPLES_PASSED_ARB);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Routine existence test */
+   if (!(*IsQueryARB)(queryObject)) {
+      fprintf(stderr, "%s: query object 0x%x fails existence test\n",
+         __FUNCTION__, queryObject);
+      (*DeleteQueriesARB)(1, &queryObject);
+      return GL_FALSE;
+   }
+
+   /* Loop until the query is ready, then get back the result.  We use
+    * the signed query for the boolean value of whether the result is
+    * available, but the unsigned query to actually pull the result;
+    * this is just to test both entrypoints, but in a real query you may
+    * need the extra bit of resolution.
+    */
+   queryReady = GL_FALSE;
+   do {
+      (*GetQueryObjectivARB)(queryObject, GL_QUERY_RESULT_AVAILABLE_ARB, 
+         &queryReady);
+   } while (!queryReady);
+   (*GetQueryObjectuivARB)(queryObject, GL_QUERY_RESULT_ARB, &querySampleCount);
+   (*DeleteQueriesARB)(1, &queryObject);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* If sample count isn't 0, something's funny. */
+   if (querySampleCount > 0) {
+      fprintf(stderr, "%s: expected query result of 0, got %ud\n",
+         __FUNCTION__, querySampleCount);
+      return GL_FALSE;
+   }
+
+   /* Here, all is well. */
+   return GL_TRUE;
+}
+
+/**************************************************************************
  * The following functions are used to check that the named OpenGL function
  * actually does what it's supposed to do.
- * The naming of these functions is signficant.  The getprocaddress.py script
+ * The naming of these functions is significant.  The getprocaddress.py script
  * scans this file and extracts these function names.
  */
 
+static GLboolean
+test_WeightPointerARB(generic_func func)
+{
+   /* Assume we have at least 2 vertex units (or this extension makes
+    * no sense), and establish a set of 2-element vector weights.
+    * We use floats that can be represented exactly in binary
+    * floating point formats so we can compare correctly later.
+    * We also make sure the 0th entry matches the default weights,
+    * so we can restore the default easily.
+    */
+#define USE_VERTEX_UNITS 2
+#define USE_WEIGHT_INDEX 3
+   static GLfloat weights[] = {
+      1.0,   0.0,
+      0.875, 0.125,
+      0.75,  0.25,
+      0.625, 0.375,
+      0.5,   0.5,
+      0.375, 0.625,
+      0.25,  0.75,
+      0.125, 0.875, 
+      0.0,   1.0,
+   };
+   GLint numVertexUnits;
+   GLfloat *currentWeights;
+   int i;
+   int errorCount = 0;
+
+   PFNGLWEIGHTPOINTERARBPROC WeightPointerARB = (PFNGLWEIGHTPOINTERARBPROC) func;
+
+   /* Make sure we have at least two vertex units */
+   glGetIntegerv(GL_MAX_VERTEX_UNITS_ARB, &numVertexUnits);
+   if (numVertexUnits < USE_VERTEX_UNITS) {
+      fprintf(stderr, "%s: need %d vertex units, got %d\n", 
+         __FUNCTION__, USE_VERTEX_UNITS, numVertexUnits);
+      return GL_FALSE;
+   }
+   
+   /* Make sure we allocate enough room to query all the current weights */
+   currentWeights = (GLfloat *)malloc(numVertexUnits * sizeof(GLfloat));
+   if (currentWeights == NULL) {
+      fprintf(stderr, "%s: couldn't allocate room for %d floats\n",
+         __FUNCTION__, numVertexUnits);
+      return GL_FALSE;
+   }
+
+   /* Set up the pointer, enable the state, and try to send down a
+    * weight vector (we'll arbitrarily send index 2).
+    */
+   (*WeightPointerARB)(USE_VERTEX_UNITS, GL_FLOAT, 0, weights);
+   glEnableClientState(GL_WEIGHT_ARRAY_ARB);
+   glArrayElement(USE_WEIGHT_INDEX);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Verify that it changed the current state. */
+   glGetFloatv(GL_CURRENT_WEIGHT_ARB, currentWeights);
+   for (i = 0; i < numVertexUnits; i++) {
+      if (i < USE_VERTEX_UNITS) {
+         /* This is one of the units we explicitly set. */
+         if (currentWeights[i] != weights[USE_VERTEX_UNITS*USE_WEIGHT_INDEX + i]) {
+            fprintf(stderr, "%s: current weight at index %d is %f, should be %f\n",
+               __FUNCTION__, i, currentWeights[i], 
+               weights[USE_VERTEX_UNITS*USE_WEIGHT_INDEX + i]);
+            errorCount++;
+         }
+      }
+      else {
+         /* All other weights should be 0. */
+         if (currentWeights[i] != 0.0) {
+            fprintf(stderr, "%s: current weight at index %d is %f, should be %f\n",
+               __FUNCTION__, i, 0.0,
+               weights[USE_VERTEX_UNITS*USE_WEIGHT_INDEX + i]);
+            errorCount++;
+         }
+      }
+   }
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Restore the old state.  We know the default set of weights is in
+    * index 0.
+    */
+   glArrayElement(0);
+   glDisableClientState(GL_WEIGHT_ARRAY_ARB);
+   (*WeightPointerARB)(0, GL_FLOAT, 0, NULL);
+   free(currentWeights);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* We're fine if we didn't get any mismatches. */
+   if (errorCount == 0) {
+      return GL_TRUE;
+   }
+   else {
+      return GL_FALSE;
+   }
+}
+
+/* Wrappers on the exercise_occlusion_query function */
+static GLboolean
+test_GenQueriesARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_BeginQueryARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_GetQueryivARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_EndQueryARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_IsQueryARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_GetQueryObjectivARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_GetQueryObjectuivARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+static GLboolean
+test_DeleteQueriesARB(generic_func func)
+{
+   (void) func;
+   return exercise_occlusion_query();
+}
+
+/* Wrappers on the exercise_buffer_objects() function */
+static GLboolean
+test_GenBuffersARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_BindBufferARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_BufferDataARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_MapBufferARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_UnmapBufferARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_DeleteBuffersARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_GetBufferParameterivARB(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer);
+}
+static GLboolean
+test_FlushMappedBufferRange(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer_Range);
+}
+static GLboolean
+test_MapBufferRange(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer_Range);
+}
+static GLboolean
+test_BufferParameteriAPPLE(generic_func func)
+{
+   (void) func;
+   return exercise_buffer_objects(Use_Map_Buffer_Range);
+}
+
+/* Wrappers on the exercise_framebuffer() function */
+static GLboolean
+test_BindFramebufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_BindRenderbufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_CheckFramebufferStatusEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_DeleteFramebuffersEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_DeleteRenderbuffersEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_FramebufferRenderbufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_GenFramebuffersEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_GenRenderbuffersEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_IsFramebufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_IsRenderbufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_RenderbufferStorageEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+static GLboolean
+test_BlitFramebufferEXT(generic_func func)
+{
+   (void) func;
+   return exercise_framebuffer();
+}
+
+/* These are wrappers on the exercise_CompressedTextures function. 
+ * Unfortunately, we cannot test the 1D counterparts, because the
+ * texture compressions available all support 2D and higher only.
+ */
+static GLboolean
+test_CompressedTexImage2DARB(generic_func func)
+{
+   (void) func;
+   return exercise_CompressedTextures(GL_TEXTURE_2D);
+}
+static GLboolean
+test_CompressedTexSubImage2DARB(generic_func func)
+{
+   (void) func;
+   return exercise_CompressedTextures(GL_TEXTURE_2D);
+}
+static GLboolean
+test_CompressedTexImage3DARB(generic_func func)
+{
+   (void) func;
+   return exercise_CompressedTextures(GL_TEXTURE_3D);
+}
+static GLboolean
+test_CompressedTexSubImage3DARB(generic_func func)
+{
+   (void) func;
+   return exercise_CompressedTextures(GL_TEXTURE_3D);
+}
+static GLboolean
+test_GetCompressedTexImageARB(generic_func func)
+{
+   (void) func;
+   return exercise_CompressedTextures(GL_TEXTURE_3D);
+}
+
+/* Wrappers on exercise_fences(). */
+static GLboolean
+test_DeleteFencesNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_GenFencesNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_SetFenceNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_TestFenceNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_FinishFenceNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_GetFenceivNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+static GLboolean
+test_IsFenceNV(generic_func func)
+{
+   (void) func;
+   return exercise_fences();
+}
+
+/* A bunch of glUniform*() tests */
+static GLboolean
+test_Uniform1iv(generic_func func)
+{
+   PFNGLUNIFORM1IVARBPROC Uniform1ivARB = (PFNGLUNIFORM1IVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform int uniformColor;" 
+      "void main() {gl_FragColor.r = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[1] = {1};
+   GLint queriedUniform[1];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * so we must set it using integer versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform1ivARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 1, uniform, 1, queriedUniform);
+}
+
+static GLboolean
+test_Uniform1i(generic_func func)
+{
+   PFNGLUNIFORM1IARBPROC Uniform1iARB = (PFNGLUNIFORM1IARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform int uniformColor;"
+      "void main() {gl_FragColor.r = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[1] = {1};
+   GLint queriedUniform[4];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * so we must set it using integer versions
+    * of the Uniform* functions.
+    */
+   (*Uniform1iARB)(uniformLocation, uniform[0]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 1, uniform, 1, queriedUniform);
+}
+
+static GLboolean
+test_Uniform1fv(generic_func func)
+{
+   PFNGLUNIFORM1FVARBPROC Uniform1fvARB = (PFNGLUNIFORM1FVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform float uniformColor;"
+      "void main() {gl_FragColor.r = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[1] = {1.1};
+   GLfloat queriedUniform[1];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * so we must set it using float versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform1fvARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 1, uniform, 1, queriedUniform);
+}
+
+static GLboolean
+test_Uniform1f(generic_func func)
+{
+   PFNGLUNIFORM1FARBPROC Uniform1fARB = (PFNGLUNIFORM1FARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform float uniformColor;"
+      "void main() {gl_FragColor.r = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[1] = {1.1};
+   GLfloat queriedUniform[1];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * so we must set it using float versions
+    * of the Uniform* functions.
+    */
+   (*Uniform1fARB)(uniformLocation, uniform[0]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 1, uniform, 1, queriedUniform);
+}
+
+static GLboolean
+test_Uniform2iv(generic_func func)
+{
+   PFNGLUNIFORM2IVARBPROC Uniform2ivARB = (PFNGLUNIFORM2IVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec2 uniformColor;" 
+      "void main() {gl_FragColor.rg = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[2] = {1,2};
+   GLint queriedUniform[2];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector 2 (ivec2), so we must set it using integer versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform2ivARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 2, uniform, 2, queriedUniform);
+}
+
+static GLboolean
+test_Uniform2i(generic_func func)
+{
+   PFNGLUNIFORM2IARBPROC Uniform2iARB = (PFNGLUNIFORM2IARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec2 uniformColor;"
+      "void main() {gl_FragColor.rg = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[2] = {1,2};
+   GLint queriedUniform[4];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector 2 (ivec2), so we must set it using integer versions
+    * of the Uniform* functions.
+    */
+   (*Uniform2iARB)(uniformLocation, uniform[0], uniform[1]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 2, uniform, 2, queriedUniform);
+}
+
+static GLboolean
+test_Uniform2fv(generic_func func)
+{
+   PFNGLUNIFORM2FVARBPROC Uniform2fvARB = (PFNGLUNIFORM2FVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec2 uniformColor;"
+      "void main() {gl_FragColor.rg = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[2] = {1.1,2.2};
+   GLfloat queriedUniform[2];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * vector 2 (vec2), so we must set it using float versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform2fvARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 2, uniform, 2, queriedUniform);
+}
+
+static GLboolean
+test_Uniform2f(generic_func func)
+{
+   PFNGLUNIFORM2FARBPROC Uniform2fARB = (PFNGLUNIFORM2FARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec2 uniformColor;"
+      "void main() {gl_FragColor.rg = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[2] = {1.1,2.2};
+   GLfloat queriedUniform[2];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * vector 2 (vec2), so we must set it using float versions
+    * of the Uniform* functions.
+    */
+   (*Uniform2fARB)(uniformLocation, uniform[0], uniform[1]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 2, uniform, 2, queriedUniform);
+}
+
+static GLboolean
+test_Uniform3iv(generic_func func)
+{
+   PFNGLUNIFORM3IVARBPROC Uniform3ivARB = (PFNGLUNIFORM3IVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec3 uniformColor;" 
+      "void main() {gl_FragColor.rgb = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[3] = {1,2,3};
+   GLint queriedUniform[3];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector 3 (ivec3), so we must set it using integer versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform3ivARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 3, uniform, 3, queriedUniform);
+}
+
+static GLboolean
+test_Uniform3i(generic_func func)
+{
+   PFNGLUNIFORM3IARBPROC Uniform3iARB = (PFNGLUNIFORM3IARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec3 uniformColor;"
+      "void main() {gl_FragColor.rgb = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[3] = {1,2,3};
+   GLint queriedUniform[4];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector 3 (ivec3), so we must set it using integer versions
+    * of the Uniform* functions.
+    */
+   (*Uniform3iARB)(uniformLocation, uniform[0], uniform[1], uniform[2]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 3, uniform, 3, queriedUniform);
+}
+
+static GLboolean
+test_Uniform3fv(generic_func func)
+{
+   PFNGLUNIFORM3FVARBPROC Uniform3fvARB = (PFNGLUNIFORM3FVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec3 uniformColor;"
+      "void main() {gl_FragColor.rgb = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[3] = {1.1,2.2,3.3};
+   GLfloat queriedUniform[3];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * vector 3 (vec3), so we must set it using float versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform3fvARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 3, uniform, 3, queriedUniform);
+}
+
+static GLboolean
+test_Uniform3f(generic_func func)
+{
+   PFNGLUNIFORM3FARBPROC Uniform3fARB = (PFNGLUNIFORM3FARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec3 uniformColor;"
+      "void main() {gl_FragColor.rgb = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[3] = {1.1,2.2,3.3};
+   GLfloat queriedUniform[3];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * vector 3 (vec3), so we must set it using float versions
+    * of the Uniform* functions.
+    */
+   (*Uniform3fARB)(uniformLocation, uniform[0], uniform[1], uniform[2]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 3, uniform, 3, queriedUniform);
+}
+
+static GLboolean
+test_Uniform4iv(generic_func func)
+{
+   PFNGLUNIFORM4IVARBPROC Uniform4ivARB = (PFNGLUNIFORM4IVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec4 uniformColor; void main() {gl_FragColor = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[4] = {1,2,3,4};
+   GLint queriedUniform[4];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector (ivec4), so we must set it using integer versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform4ivARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 4, uniform, 4, queriedUniform);
+}
+
+static GLboolean
+test_Uniform4i(generic_func func)
+{
+   PFNGLUNIFORM4IARBPROC Uniform4iARB = (PFNGLUNIFORM4IARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformivARB, PFNGLGETUNIFORMIVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform ivec4 uniformColor; void main() {gl_FragColor = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLint uniform[4] = {1,2,3,4};
+   GLint queriedUniform[4];
+
+   if (GetUniformivARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector (ivec4), so we must set it using integer versions
+    * of the Uniform* functions.
+    */
+   (*Uniform4iARB)(uniformLocation, uniform[0], uniform[1], uniform[2],
+      uniform[3]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformivARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_ints(__FUNCTION__, 4, uniform, 4, queriedUniform);
+}
+
+static GLboolean
+test_Uniform4fv(generic_func func)
+{
+   PFNGLUNIFORM4FVARBPROC Uniform4fvARB = (PFNGLUNIFORM4FVARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec4 uniformColor; void main() {gl_FragColor = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[4] = {1.1,2.2,3.3,4.4};
+   GLfloat queriedUniform[4];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is a float
+    * vector (vec4), so we must set it using float versions
+    * of the Uniform* functions.  The "1" means we're setting
+    * one vector's worth of information.
+    */
+   (*Uniform4fvARB)(uniformLocation, 1, uniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 4, uniform, 4, queriedUniform);
+}
+
+static GLboolean
+test_Uniform4f(generic_func func)
+{
+   PFNGLUNIFORM4FARBPROC Uniform4fARB = (PFNGLUNIFORM4FARBPROC) func;
+   DECLARE_GLFUNC_PTR(GetUniformfvARB, PFNGLGETUNIFORMFVARBPROC);
+
+   /* This is a trivial fragment shader that sets the color of the
+    * fragment to the uniform value passed in.
+    */
+   static const char *fragmentShaderText = 
+      "uniform vec4 uniformColor; void main() {gl_FragColor = uniformColor;}";
+   static const char *uniformName = "uniformColor";
+
+   GLhandleARB program;
+   GLint uniformLocation;
+   const GLfloat uniform[4] = {1.1,2.2,3.3,4.4};
+   GLfloat queriedUniform[4];
+
+   if (GetUniformfvARB == NULL) {
+      return GL_FALSE;
+   }
+
+   /* Call a helper function to compile up the shader and give
+    * us back the validated program and uniform location.
+    * If it fails, something's wrong and we can't continue.
+    */
+   if (!exercise_uniform_start(fragmentShaderText, uniformName, 
+      &program, &uniformLocation)) {
+      return GL_FALSE;
+   }
+
+   /* Set the value of the program uniform.  Note that you must
+    * use a compatible type.  Our uniform above is an integer
+    * vector (ivec4), so we must set it using integer versions
+    * of the Uniform* functions.
+    */
+   (*Uniform4fARB)(uniformLocation, uniform[0], uniform[1], uniform[2],
+      uniform[3]);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Query it back */
+   (*GetUniformfvARB)(program, uniformLocation, queriedUniform);
+   CheckGLError(__LINE__, __FILE__, __FUNCTION__);
+
+   /* Clean up before we check to see whether it came back unscathed */
+   exercise_uniform_end(program);
+
+   /* Now check to see whether the uniform came back as expected.  This
+    * will return GL_TRUE if all is well, or GL_FALSE if the comparison failed.
+    */
+   return compare_floats(__FUNCTION__, 4, uniform, 4, queriedUniform);
+}
 
 static GLboolean
 test_ActiveTextureARB(generic_func func)
@@ -107,6 +2694,40 @@ test_VertexAttrib1fvARB(generic_func func)
 }
 
 static GLboolean
+test_VertexAttrib1dvARB(generic_func func)
+{
+   PFNGLVERTEXATTRIB1DVARBPROC vertexAttrib1dvARB = (PFNGLVERTEXATTRIB1DVARBPROC) func;
+   PFNGLGETVERTEXATTRIBDVARBPROC getVertexAttribdvARB = (PFNGLGETVERTEXATTRIBDVARBPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvARB");
+
+   const GLdouble v[1] = {25.0};
+   const GLdouble def[1] = {0};
+   GLdouble res[4];
+   GLboolean pass;
+   (*vertexAttrib1dvARB)(6, v);
+   (*getVertexAttribdvARB)(6, GL_CURRENT_VERTEX_ATTRIB_ARB, res);
+   pass = (res[0] == 25.0 && res[1] == 0.0 && res[2] == 0.0 && res[3] == 1.0);
+   (*vertexAttrib1dvARB)(6, def);
+   return pass;
+}
+
+static GLboolean
+test_VertexAttrib1svARB(generic_func func)
+{
+   PFNGLVERTEXATTRIB1SVARBPROC vertexAttrib1svARB = (PFNGLVERTEXATTRIB1SVARBPROC) func;
+   PFNGLGETVERTEXATTRIBIVARBPROC getVertexAttribivARB = (PFNGLGETVERTEXATTRIBIVARBPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivARB");
+
+   const GLshort v[1] = {25.0};
+   const GLshort def[1] = {0};
+   GLint res[4];
+   GLboolean pass;
+   (*vertexAttrib1svARB)(6, v);
+   (*getVertexAttribivARB)(6, GL_CURRENT_VERTEX_ATTRIB_ARB, res);
+   pass = (res[0] == 25 && res[1] == 0 && res[2] == 0 && res[3] == 1);
+   (*vertexAttrib1svARB)(6, def);
+   return pass;
+}
+
+static GLboolean
 test_VertexAttrib4NubvARB(generic_func func)
 {
    PFNGLVERTEXATTRIB4NUBVARBPROC vertexAttrib4NubvARB = (PFNGLVERTEXATTRIB4NUBVARBPROC) func;
@@ -177,7 +2798,6 @@ test_VertexAttrib4NsvARB(generic_func func)
    return pass;
 }
 
-
 static GLboolean
 test_VertexAttrib4NusvARB(generic_func func)
 {
@@ -195,42 +2815,110 @@ test_VertexAttrib4NusvARB(generic_func func)
    return pass;
 }
 
+static GLboolean
+test_VertexAttrib1sNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB1SNVPROC vertexAttrib1sNV = (PFNGLVERTEXATTRIB1SNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 0, 0, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib1sNV)(6, v[0]);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib1sNV)(6, def[0]);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
 
 static GLboolean
-test_VertexAttrib4ubNV(generic_func func)
+test_VertexAttrib1fNV(generic_func func)
 {
-   PFNGLVERTEXATTRIB4UBNVPROC vertexAttrib4ubNV = (PFNGLVERTEXATTRIB4UBNVPROC) func;
+   PFNGLVERTEXATTRIB1FNVPROC vertexAttrib1fNV = (PFNGLVERTEXATTRIB1FNVPROC) func;
    PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
 
-   const GLubyte v[4] = {255, 0, 255, 0};
-   const GLubyte def[4] = {0, 0, 0, 255};
+   const GLfloat v[4] = {2.5, 0.0, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
    GLfloat res[4];
-   GLboolean pass;
-   (*vertexAttrib4ubNV)(6, v[0], v[1], v[2], v[3]);
+   (*vertexAttrib1fNV)(6, v[0]);
    (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
-   pass = (res[0] == 1.0 && res[1] == 0.0 && res[2] == 1.0 && res[3] == 0.0);
-   (*vertexAttrib4ubNV)(6, def[0], def[1], def[2], def[3]);
-   return pass;
+   (*vertexAttrib1fNV)(6, def[0]);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
 }
 
+static GLboolean
+test_VertexAttrib1dNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB1DNVPROC vertexAttrib1dNV = (PFNGLVERTEXATTRIB1DNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 0.0, 0.0, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib1dNV)(6, v[0]);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib1dNV)(6, def[0]);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
 
 static GLboolean
 test_VertexAttrib2sNV(generic_func func)
 {
    PFNGLVERTEXATTRIB2SNVPROC vertexAttrib2sNV = (PFNGLVERTEXATTRIB2SNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 0, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib2sNV)(6, v[0], v[1]);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib2sNV)(6, def[0], def[1]);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib2fNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB2FNVPROC vertexAttrib2fNV = (PFNGLVERTEXATTRIB2FNVPROC) func;
    PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
 
-   const GLshort v[2] = {2, -4,};
-   const GLshort def[2] = {0, 0};
+   const GLfloat v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
    GLfloat res[4];
-   GLboolean pass;
-   (*vertexAttrib2sNV)(6, v[0], v[1]);
+   (*vertexAttrib2fNV)(6, v[0], v[1]);
    (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
-   pass = (EQUAL(res[0], 2) && EQUAL(res[1], -4) && EQUAL(res[2], 0) && res[3] == 1.0);
-   (*vertexAttrib2sNV)(6, def[0], def[1]);
-   return pass;
+   (*vertexAttrib2fNV)(6, def[0], def[1]);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib2dNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB2DNVPROC vertexAttrib2dNV = (PFNGLVERTEXATTRIB2DNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib2dNV)(6, v[0], v[1]);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib2dNV)(6, def[0], def[1]);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
 }
 
+static GLboolean
+test_VertexAttrib3sNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB3SNVPROC vertexAttrib3sNV = (PFNGLVERTEXATTRIB3SNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib3sNV)(6, v[0], v[1], v[2]);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib3sNV)(6, def[0], def[1], def[2]);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
 
 static GLboolean
 test_VertexAttrib3fNV(generic_func func)
@@ -238,35 +2926,467 @@ test_VertexAttrib3fNV(generic_func func)
    PFNGLVERTEXATTRIB3FNVPROC vertexAttrib3fNV = (PFNGLVERTEXATTRIB3FNVPROC) func;
    PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
 
-   const GLfloat v[3] = {0.2, 0.4, 0.8};
-   const GLfloat def[3] = {0, 0, 0};
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
    GLfloat res[4];
-   GLboolean pass;
    (*vertexAttrib3fNV)(6, v[0], v[1], v[2]);
    (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
-   pass = (EQUAL(res[0], 0.2) && EQUAL(res[1], 0.4) && EQUAL(res[2], 0.8) && res[3] == 1.0);
    (*vertexAttrib3fNV)(6, def[0], def[1], def[2]);
-   return pass;
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
 }
 
+static GLboolean
+test_VertexAttrib3dNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB3DNVPROC vertexAttrib3dNV = (PFNGLVERTEXATTRIB3DNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib3dNV)(6, v[0], v[1], v[2]);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib3dNV)(6, def[0], def[1], def[2]);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4sNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4SNVPROC vertexAttrib4sNV = (PFNGLVERTEXATTRIB4SNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 5};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib4sNV)(6, v[0], v[1], v[2], v[3]);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4sNV)(6, def[0], def[1], def[2], def[3]);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4fNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4FNVPROC vertexAttrib4fNV = (PFNGLVERTEXATTRIB4FNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttrib4fNV)(6, v[0], v[1], v[2], v[3]);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4fNV)(6, def[0], def[1], def[2], def[3]);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4dNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4DNVPROC vertexAttrib4dNV = (PFNGLVERTEXATTRIB4DNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib4dNV)(6, v[0], v[1], v[2], v[3]);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4dNV)(6, def[0], def[1], def[2], def[3]);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4ubNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4UBNVPROC vertexAttrib4ubNV = (PFNGLVERTEXATTRIB4UBNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLubyte v[4] = {255, 0, 255, 0};
+   const GLubyte def[4] = {0, 0, 0, 255};
+   GLfloat res[4];
+   /* There's no byte-value query; so we use the float-value query.
+    * Bytes are interpreted as steps between 0 and 1, so the
+    * expected float values will be 0.0 for byte value 0 and 1.0 for
+    * byte value 255.
+    */
+   GLfloat expectedResults[4] = {1.0, 0.0, 1.0, 0.0};
+   (*vertexAttrib4ubNV)(6, v[0], v[1], v[2], v[3]);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4ubNV)(6, def[0], def[1], def[2], def[3]);
+   return compare_floats(__FUNCTION__, 4, expectedResults, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib1fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB1FVNVPROC vertexAttrib1fvNV = (PFNGLVERTEXATTRIB1FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 0.0, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttrib1fvNV)(6, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib1fvNV)(6, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib1dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB1DVNVPROC vertexAttrib1dvNV = (PFNGLVERTEXATTRIB1DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 0.0, 0.0, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib1dvNV)(6, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib1dvNV)(6, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib2svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB2SVNVPROC vertexAttrib2svNV = (PFNGLVERTEXATTRIB2SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 0, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib2svNV)(6, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib2svNV)(6, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib2fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB2FVNVPROC vertexAttrib2fvNV = (PFNGLVERTEXATTRIB2FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttrib2fvNV)(6, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib2fvNV)(6, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib2dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB2DVNVPROC vertexAttrib2dvNV = (PFNGLVERTEXATTRIB2DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib2dvNV)(6, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib2dvNV)(6, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib3svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB3SVNVPROC vertexAttrib3svNV = (PFNGLVERTEXATTRIB3SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib3svNV)(6, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib3svNV)(6, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib3fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB3FVNVPROC vertexAttrib3fvNV = (PFNGLVERTEXATTRIB3FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttrib3fvNV)(6, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib3fvNV)(6, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib3dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB3DVNVPROC vertexAttrib3dvNV = (PFNGLVERTEXATTRIB3DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib3dvNV)(6, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib3dvNV)(6, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4SVNVPROC vertexAttrib4svNV = (PFNGLVERTEXATTRIB4SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 5};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttrib4svNV)(6, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4svNV)(6, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4FVNVPROC vertexAttrib4fvNV = (PFNGLVERTEXATTRIB4FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttrib4fvNV)(6, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4fvNV)(6, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
 
 static GLboolean
 test_VertexAttrib4dvNV(generic_func func)
 {
    PFNGLVERTEXATTRIB4DVNVPROC vertexAttrib4dvNV = (PFNGLVERTEXATTRIB4DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttrib4dvNV)(6, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4dvNV)(6, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttrib4ubvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIB4UBVNVPROC vertexAttrib4ubvNV = (PFNGLVERTEXATTRIB4UBVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLubyte v[4] = {255, 0, 255, 0};
+   const GLubyte def[4] = {0, 0, 0, 255};
+   GLfloat res[4];
+   /* There's no byte-value query; so we use the float-value query.
+    * Bytes are interpreted as steps between 0 and 1, so the
+    * expected float values will be 0.0 for byte value 0 and 1.0 for
+    * byte value 255.
+    */
+   GLfloat expectedResults[4] = {1.0, 0.0, 1.0, 0.0};
+   (*vertexAttrib4ubvNV)(6, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttrib4ubvNV)(6, def);
+   return compare_floats(__FUNCTION__, 4, expectedResults, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs1fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS1FVNVPROC vertexAttribs1fvNV = (PFNGLVERTEXATTRIBS1FVNVPROC) func;
    PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
 
-   const GLdouble v[4] = {0.2, 0.4, 0.8, 1.2};
+   const GLfloat v[4] = {2.5, 0.0, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttribs1fvNV)(6, 1, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs1fvNV)(6, 1, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs1dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS1DVNVPROC vertexAttribs1dvNV = (PFNGLVERTEXATTRIBS1DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 0.0, 0.0, 1.0};
    const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttribs1dvNV)(6, 1, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs1dvNV)(6, 1, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs2svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS2SVNVPROC vertexAttribs2svNV = (PFNGLVERTEXATTRIBS2SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 0, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttribs2svNV)(6, 1, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs2svNV)(6, 1, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs2fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS2FVNVPROC vertexAttribs2fvNV = (PFNGLVERTEXATTRIBS2FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
    GLfloat res[4];
-   GLboolean pass;
-   (*vertexAttrib4dvNV)(6, v);
+   (*vertexAttribs2fvNV)(6, 1, v);
    (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
-   pass = (EQUAL(res[0], 0.2) && EQUAL(res[1], 0.4) && EQUAL(res[2], 0.8) && EQUAL(res[3], 1.2));
-   (*vertexAttrib4dvNV)(6, def);
-   return pass;
+   (*vertexAttribs2fvNV)(6, 1, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs2dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS2DVNVPROC vertexAttribs2dvNV = (PFNGLVERTEXATTRIBS2DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 0.0, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttribs2dvNV)(6, 1, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs2dvNV)(6, 1, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
 }
 
+static GLboolean
+test_VertexAttribs3svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS3SVNVPROC vertexAttribs3svNV = (PFNGLVERTEXATTRIBS3SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 1};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttribs3svNV)(6, 1, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs3svNV)(6, 1, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs3fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS3FVNVPROC vertexAttribs3fvNV = (PFNGLVERTEXATTRIBS3FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttribs3fvNV)(6, 1, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs3fvNV)(6, 1, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs3dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS3DVNVPROC vertexAttribs3dvNV = (PFNGLVERTEXATTRIBS3DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 1.0};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttribs3dvNV)(6, 1, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs3dvNV)(6, 1, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs4svNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS4SVNVPROC vertexAttribs4svNV = (PFNGLVERTEXATTRIBS4SVNVPROC) func;
+   PFNGLGETVERTEXATTRIBIVNVPROC getVertexAttribivNV = (PFNGLGETVERTEXATTRIBIVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribivNV");
+
+   const GLshort v[4] = {2, 4, 7, 5};
+   const GLshort def[4] = {0, 0, 0, 1};
+   GLint res[4];
+   (*vertexAttribs4svNV)(6, 1, v);
+   (*getVertexAttribivNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs4svNV)(6, 1, def);
+   return compare_shorts_to_ints(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs4fvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS4FVNVPROC vertexAttribs4fvNV = (PFNGLVERTEXATTRIBS4FVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLfloat v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLfloat def[4] = {0, 0, 0, 1};
+   GLfloat res[4];
+   (*vertexAttribs4fvNV)(6, 1, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs4fvNV)(6, 1, def);
+   return compare_floats(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs4dvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS4DVNVPROC vertexAttribs4dvNV = (PFNGLVERTEXATTRIBS4DVNVPROC) func;
+   PFNGLGETVERTEXATTRIBDVNVPROC getVertexAttribdvNV = (PFNGLGETVERTEXATTRIBDVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribdvNV");
+
+   const GLdouble v[4] = {2.5, 4.25, 7.125, 5.0625};
+   const GLdouble def[4] = {0, 0, 0, 1};
+   GLdouble res[4];
+   (*vertexAttribs4dvNV)(6, 1, v);
+   (*getVertexAttribdvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs4dvNV)(6, 1, def);
+   return compare_doubles(__FUNCTION__, 4, v, 4, res);
+}
+
+static GLboolean
+test_VertexAttribs4ubvNV(generic_func func)
+{
+   PFNGLVERTEXATTRIBS4UBVNVPROC vertexAttribs4ubvNV = (PFNGLVERTEXATTRIBS4UBVNVPROC) func;
+   PFNGLGETVERTEXATTRIBFVNVPROC getVertexAttribfvNV = (PFNGLGETVERTEXATTRIBFVNVPROC) glXGetProcAddressARB((const GLubyte *) "glGetVertexAttribfvNV");
+
+   const GLubyte v[4] = {255, 0, 255, 0};
+   const GLubyte def[4] = {0, 0, 0, 255};
+   GLfloat res[4];
+   /* There's no byte-value query; so we use the float-value query.
+    * Bytes are interpreted as steps between 0 and 1, so the
+    * expected float values will be 0.0 for byte value 0 and 1.0 for
+    * byte value 255.
+    */
+   GLfloat expectedResults[4] = {1.0, 0.0, 1.0, 0.0};
+   (*vertexAttribs4ubvNV)(6, 1, v);
+   (*getVertexAttribfvNV)(6, GL_CURRENT_ATTRIB_NV, res);
+   (*vertexAttribs4ubvNV)(6, 1, def);
+   return compare_floats(__FUNCTION__, 4, expectedResults, 4, res);
+}
 
 static GLboolean
 test_StencilFuncSeparateATI(generic_func func)
@@ -387,17 +3507,32 @@ extension_supported(const char *haystack, const char *needle)
 }
 
 
-static void
+/* Run all the known extension function tests, if the extension is supported.
+ * Return a count of how many failed.
+ */
+static int
 check_functions( const char *extensions )
 {
    struct name_test_pair *entry;
-   int failures = 0, passes = 0;
-   int totalFail = 0, totalPass = 0;
+   int failures = 0, passes = 0, untested = 0;
+   int totalFail = 0, totalPass = 0, totalUntested = 0, totalUnsupported = 0;
    int doTests;
-
+   const char *version = (const char *) glGetString(GL_VERSION);
+
+   /* The functions list will have "real" entries (consisting of
+    * a GL function name and a pointer to an exercise function for
+    * that GL function), and "group" entries (indicated as
+    * such by having a "-" as the first character of the name).
+    * "Group" names always start with the "-" character, and can
+    * be numeric (e.g. "-1.0", "-2.1"), indicating that a particular
+    * OpenGL version is required for the following functions; or can be
+    * an extension name (e.g. "-GL_ARB_multitexture") that means
+    * that the named extension is required for the following functions.
+    */
    for (entry = functions; entry->name; entry++) {
+      /* Check if this is a group indicator */
       if (entry->name[0] == '-') {
-         const char *version = (const char *) glGetString(GL_VERSION);
+         /* A group indicator; check if it's an OpenGL version group */
          if (entry->name[1] == '1') {
             /* check GL version 1.x */
             if (version[0] == '1' &&
@@ -419,14 +3554,27 @@ check_functions( const char *extensions )
             /* check if the named extension is available */
             doTests = extension_supported(extensions, entry->name+1);
          }
+
+         /* doTests is now set if we're starting an OpenGL version
+          * group, and the running OpenGL version is at least the
+          * version required; or if we're starting an OpenGL extension
+          * group, and the extension is supported.
+          */
          if (doTests)
             printf("Testing %s functions\n", entry->name + 1);
-         totalFail += failures;
-         totalPass += passes;
+
+         /* Each time we hit a title function, reset the function
+          * counts.
+          */
          failures = 0;
          passes = 0;
+         untested = 0;
       }
       else if (doTests) {
+         /* Here, we know we're trying to exercise a function for
+          * a supported extension.  See whether we have a test for
+          * it, and try to run it.
+          */
          generic_func funcPtr = (generic_func) glXGetProcAddressARB((const GLubyte *) entry->name);
          if (funcPtr) {
             if (entry->test) {
@@ -436,21 +3584,36 @@ check_functions( const char *extensions )
                if (b) {
                   printf(" Pass\n");
                   passes++;
+                  totalPass++;
                }
                else {
                   printf(" FAIL!!!\n");
                   failures++;
+                  totalFail++;
                }
             }
             else {
-               passes++;
+               untested++;
+               totalUntested++;
             }
          }
          else {
             printf("   glXGetProcAddress(%s) failed!\n", entry->name);
             failures++;
+            totalFail++;
          }
       }
+      else {
+         /* Here, we have a function that belongs to a group that
+          * is known to be unsupported.
+          */
+         totalUnsupported++;
+      }
+
+      /* Make sure a poor test case doesn't leave any lingering
+       * OpenGL errors.
+       */
+      CheckGLError(__LINE__, __FILE__, __FUNCTION__);
 
       if (doTests && (!(entry+1)->name || (entry+1)->name[0] == '-')) {
          if (failures > 0) {
@@ -459,18 +3622,29 @@ check_functions( const char *extensions )
          if (passes > 0) {
             printf("   %d passed.\n", passes);
          }
+         if (untested > 0) {
+            printf("   %d untested.\n", untested);
+         }
       }
    }
-   totalFail += failures;
-   totalPass += passes;
 
    printf("-----------------------------\n");
-   printf("Total: %d pass  %d fail\n", totalPass, totalFail);
+   printf("Total: %d pass  %d fail  %d untested  %d unsupported  %d total\n", 
+      totalPass, totalFail, totalUntested, totalUnsupported,
+      totalPass + totalFail + totalUntested + totalUnsupported);
+
+   return totalFail;
 }
 
 
+/* Return an error code */
+#define ERROR_NONE 0
+#define ERROR_NO_VISUAL 1
+#define ERROR_NO_CONTEXT 2
+#define ERROR_NO_MAKECURRENT 3
+#define ERROR_FAILED 4
 
-static void
+static int
 print_screen_info(Display *dpy, int scrnum, Bool allowDirect)
 {
    Window win;
@@ -496,6 +3670,7 @@ print_screen_info(Display *dpy, int scrnum, Bool allowDirect)
    GLXContext ctx;
    XVisualInfo *visinfo;
    int width = 100, height = 100;
+   int failures;
 
    root = RootWindow(dpy, scrnum);
 
@@ -504,7 +3679,7 @@ print_screen_info(Display *dpy, int scrnum, Bool allowDirect)
       visinfo = glXChooseVisual(dpy, scrnum, attribDouble);
       if (!visinfo) {
          fprintf(stderr, "Error: couldn't find RGB GLX visual\n");
-         return;
+         return ERROR_NO_VISUAL;
       }
    }
 
@@ -521,26 +3696,29 @@ print_screen_info(Display *dpy, int scrnum, Bool allowDirect)
    if (!ctx) {
       fprintf(stderr, "Error: glXCreateContext failed\n");
       XDestroyWindow(dpy, win);
-      return;
+      return ERROR_NO_CONTEXT;
    }
 
-   if (glXMakeCurrent(dpy, win, ctx)) {
-      check_functions( (const char *) glGetString(GL_EXTENSIONS) );
-   }
-   else {
+   if (!glXMakeCurrent(dpy, win, ctx)) {
       fprintf(stderr, "Error: glXMakeCurrent failed\n");
+      glXDestroyContext(dpy, ctx);
+      XDestroyWindow(dpy, win);
+      return ERROR_NO_MAKECURRENT;
    }
 
+   failures = check_functions( (const char *) glGetString(GL_EXTENSIONS) );
    glXDestroyContext(dpy, ctx);
    XDestroyWindow(dpy, win);
-}
 
+   return (failures == 0 ? ERROR_NONE : ERROR_FAILED);
+}
 
 int
 main(int argc, char *argv[])
 {
    char *displayName = NULL;
    Display *dpy;
+   int returnCode;
 
    dpy = XOpenDisplay(displayName);
    if (!dpy) {
@@ -548,9 +3726,9 @@ main(int argc, char *argv[])
       return -1;
    }
 
-   print_screen_info(dpy, 0, GL_TRUE);
+   returnCode = print_screen_info(dpy, 0, GL_TRUE);
 
    XCloseDisplay(dpy);
 
-   return 0;
+   return returnCode;
 }
diff --git a/progs/tests/getprocaddress.py b/progs/tests/getprocaddress.py
index 8adfc51bd60..699195bd48c 100644
--- a/progs/tests/getprocaddress.py
+++ b/progs/tests/getprocaddress.py
@@ -52,7 +52,7 @@ static struct name_test_pair functions[] = {"""
 		prev_category = None
 		
 
-		for f in api.functionIterateByOffset():
+		for f in api.functionIterateByCategory():
 			[category, num] = api.get_category_for_name( f.name )
 			if category != prev_category:
 				print '   { "-%s", NULL},' % category
diff --git a/progs/tests/getteximage.c b/progs/tests/getteximage.c
index e4818a8fabb..71f29b4ac84 100644
--- a/progs/tests/getteximage.c
+++ b/progs/tests/getteximage.c
@@ -15,7 +15,7 @@ static int Win;
 
 
 static void
-TestGetTexImage(void)
+TestGetTexImage(GLboolean npot)
 {
    GLuint iter;
    GLubyte *data = (GLubyte *) malloc(1024 * 1024 * 4);
@@ -27,8 +27,8 @@ TestGetTexImage(void)
 
    for (iter = 0; iter < 8; iter++) {
       GLint p = (iter % 8) + 3;
-      GLint w = (1 << p);
-      GLint h = (1 << p);
+      GLint w = npot ? (p * 20) : (1 << p);
+      GLint h = npot ? (p * 10) : (1 << p);
       GLuint i;
       GLint level = 0;
 
@@ -83,63 +83,94 @@ ColorsEqual(const GLubyte ref[4], const GLubyte act[4])
 
 
 static void
-TestGetTexImageRTT(void)
+TestGetTexImageRTT(GLboolean npot)
 {
    GLuint iter;
-   GLuint fb, tex;
-   GLint w = 512;
-   GLint h = 256;
-   GLint level = 0;
-   
-   glGenTextures(1, &tex);
-   glGenFramebuffersEXT(1, &fb);
-
-   glBindTexture(GL_TEXTURE_2D, tex);
-   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
-   glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
-   glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, w, h, 0,
-                GL_RGBA, GL_UNSIGNED_BYTE, NULL);
-
-   glBindFramebuffer(GL_FRAMEBUFFER_EXT, fb);
-   glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT,
-                             GL_TEXTURE_2D, tex, level);
 
    printf("Render to texture + glGetTexImage:\n");
-   printf("  Testing %d x %d tex image\n", w, h);
+
    for (iter = 0; iter < 8; iter++) {
-      GLubyte color[4];
-      GLubyte *data2 = (GLubyte *) malloc(w * h * 4);
-      GLuint i;
 
-      /* random clear color */
-      for (i = 0; i < 4; i++) {
-         color[i] = rand() % 256;
+      GLuint fb, tex;
+      GLint w, h;
+      GLint level = 0;
+
+      if (npot) {
+         w = 200 + iter * 40;
+         h = 200 + iter * 12;
+      }
+      else {
+         w = 4 << iter;
+         h = 4 << iter;
       }
 
-      glClearColor(color[0] / 255.0,
-                   color[1] / 255.0,
-                   color[2] / 255.0,
-                   color[3] / 255.0);
+      glGenTextures(1, &tex);
+      glGenFramebuffersEXT(1, &fb);
 
-      glClear(GL_COLOR_BUFFER_BIT);
+      glBindTexture(GL_TEXTURE_2D, tex);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+      glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, w, h, 0,
+                   GL_RGBA, GL_UNSIGNED_BYTE, NULL);
 
-      /* get */
-      glGetTexImage(GL_TEXTURE_2D, level, GL_RGBA, GL_UNSIGNED_BYTE, data2);
+      glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb);
+      glFramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT,
+                                GL_TEXTURE_2D, tex, level);
 
-      /* compare */
-      for (i = 0; i < w * h; i += 4) {
-         if (!ColorsEqual(color, data2 + i * 4)) {
-            printf("Render to texture failure!\n");
-            abort();
+      glViewport(0, 0, w, h);
+
+      printf("  Testing %d x %d tex image\n", w, h);
+      {
+         static const GLubyte blue[4] = {0, 0, 255, 255};
+         GLubyte color[4];
+         GLubyte *data2 = (GLubyte *) malloc(w * h * 4);
+         GLuint i;
+
+         /* random clear color */
+         for (i = 0; i < 4; i++) {
+            color[i] = rand() % 256;
+         }
+
+         glClearColor(color[0] / 255.0,
+                      color[1] / 255.0,
+                      color[2] / 255.0,
+                      color[3] / 255.0);
+
+         glClear(GL_COLOR_BUFFER_BIT);
+
+         /* draw polygon over top half, in blue */
+         glColor4ubv(blue);
+         glRectf(0, 0.5, 1.0, 1.0);
+
+         /* get */
+         glGetTexImage(GL_TEXTURE_2D, level, GL_RGBA, GL_UNSIGNED_BYTE, data2);
+
+         /* compare */
+         for (i = 0; i < w * h; i += 4) {
+            if (i < w * h / 2) {
+               /* lower half */
+               if (!ColorsEqual(color, data2 + i * 4)) {
+                  printf("Render to texture failure (expected clear color)!\n");
+                  abort();
+               }
+            }
+            else {
+               /* upper half */
+               if (!ColorsEqual(blue, data2 + i * 4)) {
+                  printf("Render to texture failure (expected blue)!\n");
+                  abort();
+               }
+            }
          }
+
+         free(data2);
       }
 
-      free(data2);
-   }
+      glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0);
+      glDeleteFramebuffersEXT(1, &fb);
+      glDeleteTextures(1, &tex);
 
-   glBindFramebuffer(GL_FRAMEBUFFER_EXT, 0);
-   glDeleteFramebuffersEXT(1, &fb);
-   glDeleteTextures(1, &tex);
+   }
 
    printf("Passed\n");
 }
@@ -152,11 +183,16 @@ Draw(void)
 {
    glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
 
-   TestGetTexImage();
+   TestGetTexImage(GL_FALSE);
+   if (glutExtensionSupported("GL_ARB_texture_non_power_of_two"))
+      TestGetTexImage(GL_TRUE);
 
    if (glutExtensionSupported("GL_EXT_framebuffer_object") ||
-       glutExtensionSupported("GL_ARB_framebuffer_object"))
-      TestGetTexImageRTT();
+       glutExtensionSupported("GL_ARB_framebuffer_object")) {
+      TestGetTexImageRTT(GL_FALSE);
+      if (glutExtensionSupported("GL_ARB_texture_non_power_of_two"))
+         TestGetTexImageRTT(GL_TRUE);
+   }
 
    glutDestroyWindow(Win);
    exit(0);
@@ -171,10 +207,10 @@ Reshape(int width, int height)
    glViewport(0, 0, width, height);
    glMatrixMode(GL_PROJECTION);
    glLoadIdentity();
-   glFrustum(-1.0, 1.0, -1.0, 1.0, 5.0, 25.0);
+   glOrtho(0, 1, 0, 1, -1, 1);
    glMatrixMode(GL_MODELVIEW);
    glLoadIdentity();
-   glTranslatef(0.0, 0.0, -15.0);
+   glTranslatef(0.0, 0.0, 0.0);
 }
 
 
diff --git a/progs/tests/texcompsub.c b/progs/tests/texcompsub.c
index 79a5f958a12..50106bf1e22 100644
--- a/progs/tests/texcompsub.c
+++ b/progs/tests/texcompsub.c
@@ -6,6 +6,7 @@
 #include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include <GL/glew.h>
 #include <GL/glut.h>
 
@@ -31,6 +32,8 @@ CheckError(int line)
 static void
 LoadCompressedImage(void)
 {
+   unsigned char ImgDataTemp[ImgSize / 4];
+   unsigned i;
    const GLenum filter = GL_LINEAR;
    glTexImage2D(Target, 0, CompFormat, ImgWidth, ImgHeight, 0,
                 GL_RGB, GL_UNSIGNED_BYTE, NULL);
@@ -40,11 +43,24 @@ LoadCompressedImage(void)
                                 0, 0, /* pos */
                                 ImgWidth, ImgHeight / 2,
                                 CompFormat, ImgSize / 2, ImgData + ImgSize / 2);
-   /* top half */
+
+   /* top left */
+   for (i = 0; i < ImgHeight / 8; i++) {
+      memcpy(&ImgDataTemp[i * ImgWidth], &ImgData[i * 2 * ImgWidth], ImgWidth);
+   }
    glCompressedTexSubImage2DARB(Target, 0,
                                 0, ImgHeight / 2, /* pos */
-                                ImgWidth, ImgHeight / 2,
-                                CompFormat, ImgSize / 2, ImgData);
+                                ImgWidth / 2, ImgHeight / 2,
+                                CompFormat, ImgSize / 4, ImgDataTemp);
+
+   /* top right */
+   for (i = 0; i < ImgHeight / 8; i++) {
+      memcpy(&ImgDataTemp[i * ImgWidth], &ImgData[i * 2 * ImgWidth + ImgWidth], ImgWidth);
+   }
+   glCompressedTexSubImage2DARB(Target, 0,
+                                ImgWidth / 2, ImgHeight / 2, /* pos */
+                                ImgWidth / 2, ImgHeight / 2,
+                                CompFormat, ImgSize / 4, ImgDataTemp);
 
    glTexParameteri(Target, GL_TEXTURE_MIN_FILTER, filter);
    glTexParameteri(Target, GL_TEXTURE_MAG_FILTER, filter);
diff --git a/progs/tests/texwrap.c b/progs/tests/texwrap.c
index 12f045b72e0..92c8a2f14c2 100644
--- a/progs/tests/texwrap.c
+++ b/progs/tests/texwrap.c
@@ -258,8 +258,26 @@ static void Init( void )
    }
 
    glBindTexture(GL_TEXTURE_2D, BORDER_TEXTURE);
+#ifdef TEST_PBO_DLIST
+   /* test fetching teximage from PBO in display list */
+   {
+      GLuint b = 42, l = 10;
+
+      glBindBufferARB(GL_PIXEL_UNPACK_BUFFER, b);
+      glBufferDataARB(GL_PIXEL_UNPACK_BUFFER, sizeof(BorderImage),
+                      BorderImage, GL_STREAM_DRAW);
+
+      glNewList(l, GL_COMPILE);
+      glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, SIZE+2, SIZE+2, 1,
+                GL_RGBA, GL_UNSIGNED_BYTE, (void *) 0/* BorderImage*/);
+      glEndList();
+      glCallList(l);
+      glBindBufferARB(GL_PIXEL_UNPACK_BUFFER, 0);
+   }
+#else
    glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, SIZE+2, SIZE+2, 1,
                 GL_RGBA, GL_UNSIGNED_BYTE, (void *) BorderImage);
+#endif
 
    for (i = 0; i < SIZE; i++) {
       for (j = 0; j < SIZE; j++) {
diff --git a/progs/trivial/.gitignore b/progs/trivial/.gitignore
index dce733a70ae..4d6e405c500 100644
--- a/progs/trivial/.gitignore
+++ b/progs/trivial/.gitignore
@@ -5,10 +5,17 @@ clear-random
 clear-repeat
 clear-scissor
 clear-undefined
+createwin
+dlist-begin-call-end
 dlist-dangling
 dlist-degenerate
 dlist-edgeflag
 dlist-edgeflag-dangling
+dlist-flat-tri
+dlist-mat-tri
+dlist-recursive-call
+dlist-tri-flat-tri
+dlist-tri-mat-tri
 draw2arrays
 drawarrays
 drawelements
@@ -31,6 +38,7 @@ lineloop
 lineloop-clip
 lineloop-elts
 linestrip
+linestrip-clip
 linestrip-flat-stipple
 linestrip-stipple
 linestrip-stipple-wide
@@ -71,8 +79,10 @@ quadstrip-cont
 quadstrip-flat
 readtex.c
 readtex.h
+readpixels
 tri
 tri-alpha
+tri-alpha-tex
 tri-array-interleaved
 tri-blend
 tri-blend-color
@@ -80,6 +90,7 @@ tri-blend-max
 tri-blend-min
 tri-blend-revsub
 tri-blend-sub
+tri-clear
 tri-clip
 tri-cull
 tri-cull-both
@@ -94,6 +105,7 @@ tri-fog
 tri-fp
 tri-fp-const-imm
 tri-lit
+tri-lit-material
 tri-logicop-none
 tri-logicop-xor
 tri-mask-tri
@@ -102,6 +114,7 @@ tri-orig
 tri-query
 tri-repeat
 tri-scissor-tri
+tri-square
 tri-stencil
 tri-stipple
 tri-tex
@@ -111,6 +124,7 @@ tri-unfilled
 tri-unfilled-clip
 tri-unfilled-edgeflag
 tri-unfilled-fog
+tri-unfilled-point
 tri-unfilled-smooth
 tri-unfilled-tri
 tri-unfilled-tri-lit
@@ -119,6 +133,7 @@ tri-unfilled-userclip-stip
 tri-userclip
 tri-viewport
 tri-z
+tri-z-9
 tri-z-eq
 trifan
 trifan-flat
diff --git a/progs/trivial/Makefile b/progs/trivial/Makefile
index 3bd8faff99a..70728616d28 100644
--- a/progs/trivial/Makefile
+++ b/progs/trivial/Makefile
@@ -18,6 +18,7 @@ SOURCES = \
 	clear-repeat.c \
 	clear-random.c \
 	clear.c \
+	createwin.c \
 	dlist-dangling.c \
 	dlist-flat-tri.c \
 	dlist-mat-tri.c \
@@ -48,6 +49,7 @@ SOURCES = \
 	lineloop-clip.c \
 	lineloop-elts.c \
 	lineloop.c \
+	linestrip-clip.c \
 	linestrip-flat-stipple.c \
 	linestrip-stipple-wide.c \
 	linestrip-stipple.c \
@@ -87,7 +89,9 @@ SOURCES = \
 	quadstrip-cont.c \
 	quadstrip-flat.c \
 	quadstrip.c \
+	readpixels.c \
 	tri-alpha.c \
+	tri-alpha-tex.c \
 	tri-array-interleaved.c \
 	tri-blend-color.c \
 	tri-blend-max.c \
@@ -95,6 +99,7 @@ SOURCES = \
 	tri-blend-revsub.c \
 	tri-blend-sub.c \
 	tri-blend.c \
+	tri-clear.c \
 	tri-clip.c \
 	tri-cull-both.c \
 	tri-cull.c \
@@ -117,6 +122,7 @@ SOURCES = \
 	tri-query.c \
 	tri-repeat.c \
 	tri-scissor-tri.c \
+	tri-square.c \
 	tri-stencil.c \
 	tri-stipple.c \
 	tri-multitex-vbo.c \
@@ -126,6 +132,7 @@ SOURCES = \
 	tri-unfilled-fog.c \
 	tri-unfilled-edgeflag.c \
 	tri-unfilled-clip.c \
+	tri-unfilled-point.c \
 	tri-unfilled-smooth.c \
 	tri-unfilled-tri.c \
 	tri-unfilled-tri-lit.c \
@@ -134,6 +141,7 @@ SOURCES = \
 	tri-unfilled.c \
 	tri-userclip.c \
 	tri-viewport.c \
+	tri-z-9.c \
 	tri-z-eq.c \
 	tri-z.c \
 	tri.c \
diff --git a/progs/util/shaderutil.c b/progs/util/shaderutil.c
index 13b68d90e0b..489e71cc30c 100644
--- a/progs/util/shaderutil.c
+++ b/progs/util/shaderutil.c
@@ -9,21 +9,12 @@
 #include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include <GL/glew.h>
 #include <GL/glut.h>
 #include "shaderutil.h"
 
 
-static void
-Init(void)
-{
-   static GLboolean firstCall = GL_TRUE;
-   if (firstCall) {
-      firstCall = GL_FALSE;
-   }
-}
-
-
 GLboolean
 ShadersSupported(void)
 {
@@ -47,8 +38,6 @@ CompileShaderText(GLenum shaderType, const char *text)
    GLuint shader;
    GLint stat;
 
-   Init();
-
    shader = glCreateShader(shaderType);
    glShaderSource(shader, 1, (const GLchar **) &text, NULL);
    glCompileShader(shader);
@@ -79,9 +68,6 @@ CompileShaderFile(GLenum shaderType, const char *filename)
    GLuint shader;
    FILE *f;
 
-   Init();
-
-
    f = fopen(filename, "r");
    if (!f) {
       fprintf(stderr, "Unable to open shader file %s\n", filename);
@@ -136,7 +122,7 @@ LinkShaders(GLuint vertShader, GLuint fragShader)
 
 
 void
-InitUniforms(GLuint program, struct uniform_info uniforms[])
+SetUniformValues(GLuint program, struct uniform_info uniforms[])
 {
    GLuint i;
 
@@ -144,28 +130,133 @@ InitUniforms(GLuint program, struct uniform_info uniforms[])
       uniforms[i].location
          = glGetUniformLocation(program, uniforms[i].name);
 
-      printf("Uniform %s location: %d\n", uniforms[i].name,
-             uniforms[i].location);
-
-      switch (uniforms[i].size) {
-      case 1:
-         if (uniforms[i].type == GL_INT)
-            glUniform1i(uniforms[i].location,
-                             (GLint) uniforms[i].value[0]);
-         else
-            glUniform1fv(uniforms[i].location, 1, uniforms[i].value);
+      switch (uniforms[i].type) {
+      case GL_INT:
+      case GL_SAMPLER_1D:
+      case GL_SAMPLER_2D:
+      case GL_SAMPLER_3D:
+      case GL_SAMPLER_CUBE:
+      case GL_SAMPLER_2D_RECT_ARB:
+         glUniform1i(uniforms[i].location,
+                     (GLint) uniforms[i].value[0]);
+         break;
+      case GL_FLOAT:
+         glUniform1fv(uniforms[i].location, 1, uniforms[i].value);
          break;
-      case 2:
+      case GL_FLOAT_VEC2:
          glUniform2fv(uniforms[i].location, 1, uniforms[i].value);
          break;
-      case 3:
+      case GL_FLOAT_VEC3:
          glUniform3fv(uniforms[i].location, 1, uniforms[i].value);
          break;
-      case 4:
+      case GL_FLOAT_VEC4:
          glUniform4fv(uniforms[i].location, 1, uniforms[i].value);
          break;
       default:
-         abort();
+         if (strncmp(uniforms[i].name, "gl_", 3) == 0) {
+            /* built-in uniform: ignore */
+         }
+         else {
+            fprintf(stderr,
+                    "Unexpected uniform data type in SetUniformValues\n");
+            abort();
+         }
       }
    }
 }
+
+
+/** Get list of uniforms used in the program */
+GLuint
+GetUniforms(GLuint program, struct uniform_info uniforms[])
+{
+   GLint n, max, i;
+
+   glGetProgramiv(program, GL_ACTIVE_UNIFORMS, &n);
+   glGetProgramiv(program, GL_ACTIVE_UNIFORM_MAX_LENGTH, &max);
+
+   for (i = 0; i < n; i++) {
+      GLint size, len;
+      GLenum type;
+      char name[100];
+
+      glGetActiveUniform(program, i, 100, &len, &size, &type, name);
+
+      uniforms[i].name = strdup(name);
+      uniforms[i].size = size;
+      uniforms[i].type = type;
+      uniforms[i].location = glGetUniformLocation(program, name);
+   }
+
+   uniforms[i].name = NULL; /* end of list */
+
+   return n;
+}
+
+
+void
+PrintUniforms(const struct uniform_info uniforms[])
+{
+   GLint i;
+
+   printf("Uniforms:\n");
+
+   for (i = 0; uniforms[i].name; i++) {
+      printf("  %d: %s size=%d type=0x%x loc=%d value=%g, %g, %g, %g\n",
+             i,
+             uniforms[i].name,
+             uniforms[i].size,
+             uniforms[i].type,
+             uniforms[i].location,
+             uniforms[i].value[0],
+             uniforms[i].value[1],
+             uniforms[i].value[2],
+             uniforms[i].value[3]);
+   }
+}
+
+
+/** Get list of attribs used in the program */
+GLuint
+GetAttribs(GLuint program, struct attrib_info attribs[])
+{
+   GLint n, max, i;
+
+   glGetProgramiv(program, GL_ACTIVE_ATTRIBUTES, &n);
+   glGetProgramiv(program, GL_ACTIVE_ATTRIBUTE_MAX_LENGTH, &max);
+
+   for (i = 0; i < n; i++) {
+      GLint size, len;
+      GLenum type;
+      char name[100];
+
+      glGetActiveAttrib(program, i, 100, &len, &size, &type, name);
+
+      attribs[i].name = strdup(name);
+      attribs[i].size = size;
+      attribs[i].type = type;
+      attribs[i].location = glGetAttribLocation(program, name);
+   }
+
+   attribs[i].name = NULL; /* end of list */
+
+   return n;
+}
+
+
+void
+PrintAttribs(const struct attrib_info attribs[])
+{
+   GLint i;
+
+   printf("Attribs:\n");
+
+   for (i = 0; attribs[i].name; i++) {
+      printf("  %d: %s size=%d type=0x%x loc=%d\n",
+             i,
+             attribs[i].name,
+             attribs[i].size,
+             attribs[i].type,
+             attribs[i].location);
+   }
+}
diff --git a/progs/util/shaderutil.h b/progs/util/shaderutil.h
index cfb8c1f3b06..0a6be026759 100644
--- a/progs/util/shaderutil.h
+++ b/progs/util/shaderutil.h
@@ -6,8 +6,8 @@
 struct uniform_info
 {
    const char *name;
-   GLuint size;
-   GLenum type;  /**< GL_FLOAT or GL_INT */
+   GLuint size;  /**< number of value[] elements: 1, 2, 3 or 4 */
+   GLenum type;  /**< GL_FLOAT, GL_FLOAT_VEC4, GL_INT, etc */
    GLfloat value[4];
    GLint location;  /**< filled in by InitUniforms() */
 };
@@ -15,6 +15,15 @@ struct uniform_info
 #define END_OF_UNIFORMS   { NULL, 0, GL_NONE, { 0, 0, 0, 0 }, -1 }
 
 
+struct attrib_info
+{
+   const char *name;
+   GLuint size;  /**< number of value[] elements: 1, 2, 3 or 4 */
+   GLenum type;  /**< GL_FLOAT, GL_FLOAT_VEC4, GL_INT, etc */
+   GLint location;
+};
+
+
 extern GLboolean
 ShadersSupported(void);
 
@@ -28,7 +37,18 @@ extern GLuint
 LinkShaders(GLuint vertShader, GLuint fragShader);
 
 extern void
-InitUniforms(GLuint program, struct uniform_info uniforms[]);
+SetUniformValues(GLuint program, struct uniform_info uniforms[]);
+
+extern GLuint
+GetUniforms(GLuint program, struct uniform_info uniforms[]);
+
+extern void
+PrintUniforms(const struct uniform_info uniforms[]);
 
+extern GLuint
+GetAttribs(GLuint program, struct attrib_info attribs[]);
+
+extern void
+PrintAttribs(const struct attrib_info attribs[]);
 
 #endif /* SHADER_UTIL_H */
diff --git a/progs/vp/arl-static.txt b/progs/vp/arl-static.txt
index aea87b79a49..83aebf689ea 100644
--- a/progs/vp/arl-static.txt
+++ b/progs/vp/arl-static.txt
@@ -1,5 +1,5 @@
 !!ARBvp1.0
-PARAM arr[5]  = { {.0,.1,.2,.3}, {.4,.5,.6,.7}, {.8,.9,.10,.1}, {.12,.3,.4,.14}, {.5,.8,.1,.9},  {.2,0,.4,.1}, {.6,.1,.8,.9}};
+PARAM arr[7]  = { {.0,.1,.2,.3}, {.4,.5,.6,.7}, {.8,.9,.10,.1}, {.12,.3,.4,.14}, {.5,.8,.1,.9},  {.2,0,.4,.1}, {.6,.1,.8,.9}};
 ADDRESS addr;
 ARL addr.x, {3}.x;
 MOV result.color, arr[addr.x];
diff --git a/progs/vp/arl-unused.txt b/progs/vp/arl-unused.txt
index 7bdbb8e86c7..c2afe3c0924 100644
--- a/progs/vp/arl-unused.txt
+++ b/progs/vp/arl-unused.txt
@@ -1,5 +1,5 @@
 !!ARBvp1.0
-PARAM arr[5]  = { {.0,.1,.2,.3}, {.4,.5,.6,.7}, {.8,.9,.10,.1}, {.12,.3,.4,.14}, {.5,.8,.1,.9},  {.2,0,.4,.1}, {.6,.1,.8,.9}};
+PARAM arr[7]  = { {.0,.1,.2,.3}, {.4,.5,.6,.7}, {.8,.9,.10,.1}, {.12,.3,.4,.14}, {.5,.8,.1,.9},  {.2,0,.4,.1}, {.6,.1,.8,.9}};
 ADDRESS addr;
 ARL addr.x, {3}.x;  # not actually used
 MOV result.color, arr[3];
diff --git a/progs/xdemos/Makefile b/progs/xdemos/Makefile
index 6581df80398..53e1c54ef3c 100644
--- a/progs/xdemos/Makefile
+++ b/progs/xdemos/Makefile
@@ -26,6 +26,7 @@ PROGS = \
 	glxsnoop \
 	glxswapcontrol \
 	manywin \
+	multictx \
 	offset \
 	overlay \
 	pbinfo \
diff --git a/progs/xdemos/glxcontexts.c b/progs/xdemos/glxcontexts.c
index 481749be3d8..9f83679acd3 100644
--- a/progs/xdemos/glxcontexts.c
+++ b/progs/xdemos/glxcontexts.c
@@ -378,6 +378,9 @@ make_window( Display *dpy, const char *name,
    scrnum = DefaultScreen( dpy );
    root = RootWindow( dpy, scrnum );
 
+   if (visinfo)
+	   XFree(visinfo);
+
    visinfo = glXChooseVisual( dpy, scrnum, attribs );
    if (!visinfo) {
       printf("Error: couldn't get an RGB, Double-buffered visual\n");
diff --git a/progs/xdemos/glxgears.c b/progs/xdemos/glxgears.c
index bc84ee3dbd2..088f25a357a 100644
--- a/progs/xdemos/glxgears.c
+++ b/progs/xdemos/glxgears.c
@@ -23,10 +23,7 @@
  * This is a port of the infamous "gears" demo to straight GLX (i.e. no GLUT)
  * Port by Brian Paul  23 March 2001
  *
- * Command line options:
- *    -info      print GL implementation information
- *    -stereo    use stereo enabled GLX visual
- *
+ * See usage() below for command line options.
  */
 
 
@@ -45,10 +42,6 @@ typedef int (*PFNGLXGETSWAPINTERVALMESAPROC)(void);
 #endif
 
 
-static int is_glx_extension_supported(Display *dpy, const char *query);
-
-static void query_vsync(Display *dpy);
-
 #define BENCHMARK
 
 #ifdef BENCHMARK
@@ -572,7 +565,7 @@ make_window( Display *dpy, const char *name,
 /**
  * Determine whether or not a GLX extension is supported.
  */
-int
+static int
 is_glx_extension_supported(Display *dpy, const char *query)
 {
    const int scrnum = DefaultScreen(dpy);
@@ -592,7 +585,7 @@ is_glx_extension_supported(Display *dpy, const char *query)
 /**
  * Attempt to determine whether or not the display is synched to vblank.
  */
-void
+static void
 query_vsync(Display *dpy)
 {
    int interval = 0;
diff --git a/progs/xdemos/glxswapcontrol.c b/progs/xdemos/glxswapcontrol.c
index df9f7ad784b..cd60d460a06 100644
--- a/progs/xdemos/glxswapcontrol.c
+++ b/progs/xdemos/glxswapcontrol.c
@@ -587,6 +587,9 @@ event_loop(Display *dpy, Window win)
       angle += 2.0;
 
       draw();
+
+      glXSwapBuffers(dpy, win);
+
       if ( get_frame_usage != NULL ) {
 	 GLfloat   temp;
 	 
@@ -594,8 +597,6 @@ event_loop(Display *dpy, Window win)
 	 frame_usage += temp;
       }
 
-      glXSwapBuffers(dpy, win);
-
       /* calc framerate */
       {
          static int t0 = -1;
diff --git a/progs/xdemos/multictx.c b/progs/xdemos/multictx.c
new file mode 100644
index 00000000000..75255b28605
--- /dev/null
+++ b/progs/xdemos/multictx.c
@@ -0,0 +1,585 @@
+/*
+ * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Test rendering with two contexts into one window.
+ * Setup different rendering state for each context to check that
+ * context switching is handled properly.
+ *
+ * Brian Paul
+ * 6 Aug 2009
+ */
+
+
+#include <math.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <X11/Xlib.h>
+#include <X11/keysym.h>
+#include <GL/gl.h>
+#include <GL/glx.h>
+
+
+
+#ifndef M_PI
+#define M_PI 3.14159265
+#endif
+
+
+/** Event handler results: */
+#define NOP 0
+#define EXIT 1
+#define DRAW 2
+
+static GLfloat view_rotx = 0.0, view_roty = 210.0, view_rotz = 0.0;
+static GLint gear1, gear2;
+static GLfloat angle = 0.0;
+
+static GLboolean animate = GL_TRUE;	/* Animation */
+
+
+static double
+current_time(void)
+{
+   struct timeval tv;
+#ifdef __VMS
+   (void) gettimeofday(&tv, NULL );
+#else
+   struct timezone tz;
+   (void) gettimeofday(&tv, &tz);
+#endif
+   return (double) tv.tv_sec + tv.tv_usec / 1000000.0;
+}
+
+
+/*
+ *
+ *  Draw a gear wheel.  You'll probably want to call this function when
+ *  building a display list since we do a lot of trig here.
+ * 
+ *  Input:  inner_radius - radius of hole at center
+ *          outer_radius - radius at center of teeth
+ *          width - width of gear
+ *          teeth - number of teeth
+ *          tooth_depth - depth of tooth
+ */
+static void
+gear(GLfloat inner_radius, GLfloat outer_radius, GLfloat width,
+     GLint teeth, GLfloat tooth_depth)
+{
+   GLint i;
+   GLfloat r0, r1, r2;
+   GLfloat angle, da;
+   GLfloat u, v, len;
+
+   r0 = inner_radius;
+   r1 = outer_radius - tooth_depth / 2.0;
+   r2 = outer_radius + tooth_depth / 2.0;
+
+   da = 2.0 * M_PI / teeth / 4.0;
+
+   glShadeModel(GL_FLAT);
+
+   glNormal3f(0.0, 0.0, 1.0);
+
+   /* draw front face */
+   glBegin(GL_QUAD_STRIP);
+   for (i = 0; i <= teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+      glVertex3f(r0 * cos(angle), r0 * sin(angle), width * 0.5);
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), width * 0.5);
+      if (i < teeth) {
+	 glVertex3f(r0 * cos(angle), r0 * sin(angle), width * 0.5);
+	 glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		    width * 0.5);
+      }
+   }
+   glEnd();
+
+   /* draw front sides of teeth */
+   glBegin(GL_QUADS);
+   da = 2.0 * M_PI / teeth / 4.0;
+   for (i = 0; i < teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), width * 0.5);
+      glVertex3f(r2 * cos(angle + da), r2 * sin(angle + da), width * 0.5);
+      glVertex3f(r2 * cos(angle + 2 * da), r2 * sin(angle + 2 * da),
+		 width * 0.5);
+      glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		 width * 0.5);
+   }
+   glEnd();
+
+   glNormal3f(0.0, 0.0, -1.0);
+
+   /* draw back face */
+   glBegin(GL_QUAD_STRIP);
+   for (i = 0; i <= teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), -width * 0.5);
+      glVertex3f(r0 * cos(angle), r0 * sin(angle), -width * 0.5);
+      if (i < teeth) {
+	 glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		    -width * 0.5);
+	 glVertex3f(r0 * cos(angle), r0 * sin(angle), -width * 0.5);
+      }
+   }
+   glEnd();
+
+   /* draw back sides of teeth */
+   glBegin(GL_QUADS);
+   da = 2.0 * M_PI / teeth / 4.0;
+   for (i = 0; i < teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+
+      glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		 -width * 0.5);
+      glVertex3f(r2 * cos(angle + 2 * da), r2 * sin(angle + 2 * da),
+		 -width * 0.5);
+      glVertex3f(r2 * cos(angle + da), r2 * sin(angle + da), -width * 0.5);
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), -width * 0.5);
+   }
+   glEnd();
+
+   /* draw outward faces of teeth */
+   glBegin(GL_QUAD_STRIP);
+   for (i = 0; i < teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), width * 0.5);
+      glVertex3f(r1 * cos(angle), r1 * sin(angle), -width * 0.5);
+      u = r2 * cos(angle + da) - r1 * cos(angle);
+      v = r2 * sin(angle + da) - r1 * sin(angle);
+      len = sqrt(u * u + v * v);
+      u /= len;
+      v /= len;
+      glNormal3f(v, -u, 0.0);
+      glVertex3f(r2 * cos(angle + da), r2 * sin(angle + da), width * 0.5);
+      glVertex3f(r2 * cos(angle + da), r2 * sin(angle + da), -width * 0.5);
+      glNormal3f(cos(angle), sin(angle), 0.0);
+      glVertex3f(r2 * cos(angle + 2 * da), r2 * sin(angle + 2 * da),
+		 width * 0.5);
+      glVertex3f(r2 * cos(angle + 2 * da), r2 * sin(angle + 2 * da),
+		 -width * 0.5);
+      u = r1 * cos(angle + 3 * da) - r2 * cos(angle + 2 * da);
+      v = r1 * sin(angle + 3 * da) - r2 * sin(angle + 2 * da);
+      glNormal3f(v, -u, 0.0);
+      glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		 width * 0.5);
+      glVertex3f(r1 * cos(angle + 3 * da), r1 * sin(angle + 3 * da),
+		 -width * 0.5);
+      glNormal3f(cos(angle), sin(angle), 0.0);
+   }
+
+   glVertex3f(r1 * cos(0), r1 * sin(0), width * 0.5);
+   glVertex3f(r1 * cos(0), r1 * sin(0), -width * 0.5);
+
+   glEnd();
+
+   glShadeModel(GL_SMOOTH);
+
+   /* draw inside radius cylinder */
+   glBegin(GL_QUAD_STRIP);
+   for (i = 0; i <= teeth; i++) {
+      angle = i * 2.0 * M_PI / teeth;
+      glNormal3f(-cos(angle), -sin(angle), 0.0);
+      glVertex3f(r0 * cos(angle), r0 * sin(angle), -width * 0.5);
+      glVertex3f(r0 * cos(angle), r0 * sin(angle), width * 0.5);
+   }
+   glEnd();
+}
+
+
+static void
+draw(int ctx)
+{
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   glPushMatrix();
+   glRotatef(view_rotx, 1.0, 0.0, 0.0);
+   glRotatef(view_roty + angle, 0.0, 1.0, 0.0);
+   glRotatef(view_rotz, 0.0, 0.0, 1.0);
+
+   if (ctx == 0) {
+      glDisable(GL_CULL_FACE);
+      glPushMatrix();
+      glRotatef(angle, 0.0, 0.0, 1.0);
+      glCallList(gear1);
+      glPopMatrix();
+      /* This should not effect the other context's rendering */
+      glEnable(GL_CULL_FACE);
+      glCullFace(GL_FRONT_AND_BACK);
+   }
+   else {
+      glPushMatrix();
+      glRotatef(-2.0 * angle - 9.0, 0.0, 0.0, 1.0);
+      glCallList(gear2);
+      glPopMatrix();
+   }
+
+   glPopMatrix();
+
+   /* this flush is important since we'll be switching contexts next */
+   glFlush();
+}
+
+
+
+static void
+draw_frame(Display *dpy, Window win, GLXContext ctx1, GLXContext ctx2)
+{
+   static double tRot0 = -1.0;
+   double dt, t = current_time();
+
+   if (tRot0 < 0.0)
+      tRot0 = t;
+   dt = t - tRot0;
+   tRot0 = t;
+
+   if (animate) {
+      /* advance rotation for next frame */
+      angle += 70.0 * dt;  /* 70 degrees per second */
+      if (angle > 3600.0)
+         angle -= 3600.0;
+   }
+
+   glXMakeCurrent(dpy, (GLXDrawable) win, ctx1);
+   draw(0);
+
+   glXMakeCurrent(dpy, (GLXDrawable) win, ctx2);
+   draw(1);
+
+   glXSwapBuffers(dpy, win);
+}
+
+
+/* new window size or exposure */
+static void
+reshape(Display *dpy, Window win,
+        GLXContext ctx1, GLXContext ctx2, int width, int height)
+{
+   int i;
+
+   width /= 2;
+
+   /* loop: left half of window, right half of window */
+   for (i = 0; i < 2; i++) {
+      if (i == 0)
+         glXMakeCurrent(dpy, win, ctx1);
+      else
+         glXMakeCurrent(dpy, win, ctx2);
+
+      glViewport(width * i, 0, width, height);
+      glScissor(width * i, 0, width, height);
+
+      {
+         GLfloat h = (GLfloat) height / (GLfloat) width;
+
+         glMatrixMode(GL_PROJECTION);
+         glLoadIdentity();
+         glFrustum(-1.0, 1.0, -h, h, 5.0, 60.0);
+      }
+
+      glMatrixMode(GL_MODELVIEW);
+      glLoadIdentity();
+      glTranslatef(0.0, 0.0, -30.0);
+   }
+}
+   
+
+
+static void
+init(Display *dpy, Window win, GLXContext ctx1, GLXContext ctx2)
+{
+   static GLfloat pos[4] = { 5.0, 5.0, 10.0, 0.0 };
+   static GLfloat red[4] = { 0.8, 0.1, 0.0, 1.0 };
+   static GLfloat green[4] = { 0.0, 0.8, 0.2, 0.5 };
+   /*static GLfloat blue[4] = { 0.2, 0.2, 1.0, 1.0 };*/
+
+   /* first ctx */
+   {
+      static GLuint stipple[32] = {
+         0x00ff00ff, 0x00ff00ff, 0x00ff00ff, 0x00ff00ff,
+         0x00ff00ff, 0x00ff00ff, 0x00ff00ff, 0x00ff00ff,
+
+         0xff00ff00, 0xff00ff00, 0xff00ff00, 0xff00ff00,
+         0xff00ff00, 0xff00ff00, 0xff00ff00, 0xff00ff00,
+
+         0x00ff00ff, 0x00ff00ff, 0x00ff00ff, 0x00ff00ff,
+         0x00ff00ff, 0x00ff00ff, 0x00ff00ff, 0x00ff00ff,
+
+         0xff00ff00, 0xff00ff00, 0xff00ff00, 0xff00ff00,
+         0xff00ff00, 0xff00ff00, 0xff00ff00, 0xff00ff00
+      };
+
+      glXMakeCurrent(dpy, win, ctx1);
+
+      glLightfv(GL_LIGHT0, GL_POSITION, pos);
+      glEnable(GL_LIGHTING);
+      glEnable(GL_LIGHT0);
+      glEnable(GL_DEPTH_TEST);
+
+      gear1 = glGenLists(1);
+      glNewList(gear1, GL_COMPILE);
+      glMaterialfv(GL_FRONT, GL_AMBIENT_AND_DIFFUSE, red);
+      gear(1.0, 4.0, 1.0, 20, 0.7);
+      glEndList();
+
+      glEnable(GL_NORMALIZE);
+      glEnable(GL_SCISSOR_TEST);
+      glClearColor(0.4, 0.4, 0.4, 1.0);
+
+      glPolygonStipple((GLubyte *) stipple);
+      glEnable(GL_POLYGON_STIPPLE);
+   }
+
+   /* second ctx */
+   {
+      glXMakeCurrent(dpy, win, ctx2);
+
+      glLightfv(GL_LIGHT0, GL_POSITION, pos);
+      glEnable(GL_LIGHTING);
+      glEnable(GL_LIGHT0);
+      glEnable(GL_DEPTH_TEST);
+
+      gear2 = glGenLists(1);
+      glNewList(gear2, GL_COMPILE);
+      glMaterialfv(GL_FRONT, GL_AMBIENT_AND_DIFFUSE, green);
+      gear(1.5, 3.0, 1.5, 16, 0.7);
+      glEndList();
+
+      glEnable(GL_NORMALIZE);
+      glEnable(GL_SCISSOR_TEST);
+      glClearColor(0.6, 0.6, 0.6, 1.0);
+
+      glEnable(GL_BLEND);
+      glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
+   }
+}
+
+
+/**
+ * Create an RGB, double-buffered window.
+ * Return the window and two context handles.
+ */
+static void
+make_window_and_contexts( Display *dpy, const char *name,
+                          int x, int y, int width, int height,
+                          Window *winRet,
+                          GLXContext *ctxRet1,
+                          GLXContext *ctxRet2)
+{
+   int attribs[] = { GLX_RGBA,
+                     GLX_RED_SIZE, 1,
+                     GLX_GREEN_SIZE, 1,
+                     GLX_BLUE_SIZE, 1,
+                     GLX_DOUBLEBUFFER,
+                     GLX_DEPTH_SIZE, 1,
+                     None };
+   int scrnum;
+   XSetWindowAttributes attr;
+   unsigned long mask;
+   Window root;
+   Window win;
+   XVisualInfo *visinfo;
+
+   scrnum = DefaultScreen( dpy );
+   root = RootWindow( dpy, scrnum );
+
+   visinfo = glXChooseVisual( dpy, scrnum, attribs );
+   if (!visinfo) {
+      printf("Error: couldn't get an RGB, Double-buffered visual\n");
+      exit(1);
+   }
+
+   /* window attributes */
+   attr.background_pixel = 0;
+   attr.border_pixel = 0;
+   attr.colormap = XCreateColormap( dpy, root, visinfo->visual, AllocNone);
+   attr.event_mask = StructureNotifyMask | ExposureMask | KeyPressMask;
+   mask = CWBackPixel | CWBorderPixel | CWColormap | CWEventMask;
+
+   win = XCreateWindow( dpy, root, x, y, width, height,
+		        0, visinfo->depth, InputOutput,
+		        visinfo->visual, mask, &attr );
+
+   /* set hints and properties */
+   {
+      XSizeHints sizehints;
+      sizehints.x = x;
+      sizehints.y = y;
+      sizehints.width  = width;
+      sizehints.height = height;
+      sizehints.flags = USSize | USPosition;
+      XSetNormalHints(dpy, win, &sizehints);
+      XSetStandardProperties(dpy, win, name, name,
+                              None, (char **)NULL, 0, &sizehints);
+   }
+
+   *winRet = win;
+   *ctxRet1 = glXCreateContext( dpy, visinfo, NULL, True );
+   *ctxRet2 = glXCreateContext( dpy, visinfo, NULL, True );
+
+   if (!*ctxRet1 || !*ctxRet2) {
+      printf("Error: glXCreateContext failed\n");
+      exit(1);
+   }
+
+   XFree(visinfo);
+}
+
+
+/**
+ * Handle one X event.
+ * \return NOP, EXIT or DRAW
+ */
+static int
+handle_event(Display *dpy, Window win, GLXContext ctx1, GLXContext ctx2,
+             XEvent *event)
+{
+   (void) dpy;
+   (void) win;
+
+   switch (event->type) {
+   case Expose:
+      return DRAW;
+   case ConfigureNotify:
+      reshape(dpy, win, ctx1, ctx2,
+              event->xconfigure.width, event->xconfigure.height);
+      break;
+   case KeyPress:
+      {
+         char buffer[10];
+         int r, code;
+         code = XLookupKeysym(&event->xkey, 0);
+         if (code == XK_Left) {
+            view_roty += 5.0;
+         }
+         else if (code == XK_Right) {
+            view_roty -= 5.0;
+         }
+         else if (code == XK_Up) {
+            view_rotx += 5.0;
+         }
+         else if (code == XK_Down) {
+            view_rotx -= 5.0;
+         }
+         else {
+            r = XLookupString(&event->xkey, buffer, sizeof(buffer),
+                              NULL, NULL);
+            if (buffer[0] == 27) {
+               /* escape */
+               return EXIT;
+            }
+            else if (buffer[0] == 'a' || buffer[0] == 'A') {
+               animate = !animate;
+            }
+         }
+         return DRAW;
+      }
+   }
+   return NOP;
+}
+
+
+static void
+event_loop(Display *dpy, Window win, GLXContext ctx1, GLXContext ctx2)
+{
+   while (1) {
+      int op;
+      while (!animate || XPending(dpy) > 0) {
+         XEvent event;
+         XNextEvent(dpy, &event);
+         op = handle_event(dpy, win, ctx1, ctx2, &event);
+         if (op == EXIT)
+            return;
+         else if (op == DRAW)
+            break;
+      }
+
+      draw_frame(dpy, win, ctx1, ctx2);
+   }
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   unsigned int winWidth = 800, winHeight = 400;
+   int x = 0, y = 0;
+   Display *dpy;
+   Window win;
+   GLXContext ctx1, ctx2;
+   char *dpyName = NULL;
+   GLboolean printInfo = GL_FALSE;
+   int i;
+
+   for (i = 1; i < argc; i++) {
+      if (strcmp(argv[i], "-display") == 0) {
+         dpyName = argv[i+1];
+         i++;
+      }
+      else {
+         return 1;
+      }
+   }
+
+   dpy = XOpenDisplay(dpyName);
+   if (!dpy) {
+      printf("Error: couldn't open display %s\n",
+	     dpyName ? dpyName : getenv("DISPLAY"));
+      return -1;
+   }
+
+   make_window_and_contexts(dpy, "multictx", x, y, winWidth, winHeight,
+                            &win, &ctx1, &ctx2);
+   XMapWindow(dpy, win);
+
+   if (printInfo) {
+      printf("GL_RENDERER   = %s\n", (char *) glGetString(GL_RENDERER));
+      printf("GL_VERSION    = %s\n", (char *) glGetString(GL_VERSION));
+      printf("GL_VENDOR     = %s\n", (char *) glGetString(GL_VENDOR));
+      printf("GL_EXTENSIONS = %s\n", (char *) glGetString(GL_EXTENSIONS));
+   }
+
+   init(dpy, win, ctx1, ctx2);
+
+   /* Set initial projection/viewing transformation.
+    * We can't be sure we'll get a ConfigureNotify event when the window
+    * first appears.
+    */
+   reshape(dpy, win, ctx1, ctx2, winWidth, winHeight);
+
+   event_loop(dpy, win, ctx1, ctx2);
+
+   glDeleteLists(gear1, 1);
+   glDeleteLists(gear2, 1);
+   glXDestroyContext(dpy, ctx1);
+   glXDestroyContext(dpy, ctx2);
+   XDestroyWindow(dpy, win);
+   XCloseDisplay(dpy);
+
+   return 0;
+}
diff --git a/scons/gallium.py b/scons/gallium.py
index e9e799dc78d..bf6172b4d77 100644
--- a/scons/gallium.py
+++ b/scons/gallium.py
@@ -462,6 +462,8 @@ def generate(env):
         shlinkflags += [
             '-Wl,-Bsymbolic',
         ]
+        # Handle circular dependencies in the libraries
+        env['_LIBFLAGS'] = '-Wl,--start-group ' + env['_LIBFLAGS'] + ' -Wl,--end-group'
     if platform == 'windows' and msvc:
         # See also:
         # - http://msdn2.microsoft.com/en-us/library/y0zzbyt4.aspx
diff --git a/src/egl/drivers/demo/Makefile b/src/egl/drivers/demo/Makefile
index 26694c92fa6..444dfb35bde 100644
--- a/src/egl/drivers/demo/Makefile
+++ b/src/egl/drivers/demo/Makefile
@@ -4,7 +4,7 @@ TOP = ../../../..
 include $(TOP)/configs/current
 
 
-INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/egl/main
+INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/egl/main $(X11_INCLUDES)
 
 
 SOURCES = demo.c
diff --git a/src/egl/drivers/demo/demo.c b/src/egl/drivers/demo/demo.c
index f316974d836..e8c0c1df5ee 100644
--- a/src/egl/drivers/demo/demo.c
+++ b/src/egl/drivers/demo/demo.c
@@ -236,7 +236,7 @@ demoDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
 {
    DemoSurface *fs = LookupDemoSurface(surface);
    _eglUnlinkSurface(&fs->Base);
-   if (!fs->Base.IsBound)
+   if (!_eglIsSurfaceBound(&fs->Base))
       free(fs);
    return EGL_TRUE;
 }
@@ -247,7 +247,7 @@ demoDestroyContext(_EGLDriver *drv, EGLDisplay dpy, EGLContext context)
 {
    DemoContext *fc = LookupDemoContext(context);
    _eglUnlinkContext(&fc->Base);
-   if (!fc->Base.IsBound)
+   if (!_eglIsContextBound(&fc->Base))
       free(fc);
    return EGL_TRUE;
 }
diff --git a/src/egl/drivers/dri/Makefile b/src/egl/drivers/dri/Makefile
index 567edfae973..7339c97c77d 100644
--- a/src/egl/drivers/dri/Makefile
+++ b/src/egl/drivers/dri/Makefile
@@ -55,6 +55,7 @@ install:
 clean:
 	-rm -f *.o
 	-rm -f *.so
+	-rm -f depend depend.bak
 
 depend: $(SOURCES) $(HEADERS)
 	@ echo "running $(MKDEP)"
diff --git a/src/egl/drivers/dri/egldri.c b/src/egl/drivers/dri/egldri.c
index 3f9617a8f98..9e400be6248 100644
--- a/src/egl/drivers/dri/egldri.c
+++ b/src/egl/drivers/dri/egldri.c
@@ -296,7 +296,7 @@ _eglDRIDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
 
    fs->drawable.destroyDrawable(disp, fs->drawable.private);
 
-   if (!fs->Base.IsBound)
+   if (!_eglIsSurfaceBound(&fs->Base))
       free(fs);
    return EGL_TRUE;
 }
@@ -312,7 +312,7 @@ _eglDRIDestroyContext(_EGLDriver *drv, EGLDisplay dpy, EGLContext context)
 
    fc->driContext.destroyContext(disp, 0, fc->driContext.private);
 
-   if (!fc->Base.IsBound)
+   if (!_eglIsContextBound(&fc->Base))
       free(fc);
    return EGL_TRUE;
 }
diff --git a/src/egl/drivers/glx/egl_glx.c b/src/egl/drivers/glx/egl_glx.c
index 207b1ea7793..5ed4b6883f8 100644
--- a/src/egl/drivers/glx/egl_glx.c
+++ b/src/egl/drivers/glx/egl_glx.c
@@ -739,7 +739,7 @@ GLX_eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
    return EGL_TRUE;
    if (surf) {
       _eglUnlinkSurface(surf);
-      if (!surf->IsBound)
+      if (!_eglIsSurfaceBound(surf))
          free(surf);
 
       return EGL_TRUE;
diff --git a/src/egl/drivers/xdri/egl_xdri.c b/src/egl/drivers/xdri/egl_xdri.c
index e040efdd438..d8d29fcef49 100644
--- a/src/egl/drivers/xdri/egl_xdri.c
+++ b/src/egl/drivers/xdri/egl_xdri.c
@@ -958,7 +958,7 @@ xdri_eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
    struct xdri_egl_surface *xdri_surf = lookup_surface(surface);
    if (xdri_surf) {
       _eglUnlinkSurface(&xdri_surf->Base);
-      if (!xdri_surf->Base.IsBound) {
+      if (!_eglIsSurfaceBound(&xdri_surf->Base)) {
          /*
          st_unreference_framebuffer(surf->Framebuffer);
          */
diff --git a/src/egl/main/Makefile b/src/egl/main/Makefile
index e1ff8794b35..ab61d68f2b6 100644
--- a/src/egl/main/Makefile
+++ b/src/egl/main/Makefile
@@ -4,7 +4,7 @@ TOP = ../../..
 include $(TOP)/configs/current
 
 
-INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/mesa/glapi
+INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/mesa/glapi $(X11_INCLUDES)
 
 HEADERS = \
 	eglcompiler.h \
@@ -20,6 +20,7 @@ HEADERS = \
 	eglhash.h \
 	eglmisc.h \
 	eglmode.h \
+	eglmutex.h \
 	eglscreen.h \
 	eglstring.h \
 	eglsurface.h \
@@ -74,6 +75,7 @@ install: default
 clean:
 	-rm -f *.o *.so*
 	-rm -f core.*
+	-rm -f depend depend.bak
 
 
 depend: $(SOURCES) $(HEADERS)
diff --git a/src/egl/main/eglapi.c b/src/egl/main/eglapi.c
index f0a6f7f9355..fde6b7316c8 100644
--- a/src/egl/main/eglapi.c
+++ b/src/egl/main/eglapi.c
@@ -50,7 +50,6 @@ EGLDisplay EGLAPIENTRY
 eglGetDisplay(NativeDisplayType nativeDisplay)
 {
    _EGLDisplay *dpy;
-   _eglInitGlobals();
    dpy = _eglFindDisplay(nativeDisplay);
    if (!dpy) {
       dpy = _eglNewDisplay(nativeDisplay);
diff --git a/src/egl/main/eglcompiler.h b/src/egl/main/eglcompiler.h
index 0b19afedfd3..6b639b75c66 100644
--- a/src/egl/main/eglcompiler.h
+++ b/src/egl/main/eglcompiler.h
@@ -3,6 +3,39 @@
 
 
 /**
+ * Get standard integer types
+ */
+#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L)
+#  include <stdint.h>
+#elif defined(_MSC_VER)
+   typedef __int8             int8_t;
+   typedef unsigned __int8    uint8_t;
+   typedef __int16            int16_t;
+   typedef unsigned __int16   uint16_t;
+#  ifndef __eglplatform_h_
+     typedef __int32            int32_t;
+#  endif
+   typedef unsigned __int32   uint32_t;
+   typedef __int64            int64_t;
+   typedef unsigned __int64   uint64_t;
+
+#  if defined(_WIN64)
+     typedef __int64            intptr_t;
+     typedef unsigned __int64   uintptr_t;
+#  else
+     typedef __int32            intptr_t;
+     typedef unsigned __int32   uintptr_t;
+#  endif
+
+#  define INT64_C(__val) __val##i64
+#  define UINT64_C(__val) __val##ui64
+#else
+/* hope the best instead of adding a bunch of ifdef's */
+#  include <stdint.h>
+#endif
+
+
+/**
  * Function inlining
  */
 #if defined(__GNUC__)
diff --git a/src/egl/main/eglconfig.c b/src/egl/main/eglconfig.c
index f2f32585c73..bbc585b55e9 100644
--- a/src/egl/main/eglconfig.c
+++ b/src/egl/main/eglconfig.c
@@ -34,7 +34,7 @@ void
 _eglInitConfig(_EGLConfig *config, EGLint id)
 {
    memset(config, 0, sizeof(*config));
-   config->Handle = (EGLConfig) id;
+   config->Handle = (EGLConfig) _eglUIntToPointer((unsigned int) id);
    _eglSetConfigAttrib(config, EGL_CONFIG_ID,               id);
    _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGB,     EGL_DONT_CARE);
    _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGBA,    EGL_DONT_CARE);
diff --git a/src/egl/main/eglconfigutil.c b/src/egl/main/eglconfigutil.c
index 138dc729e74..c9d00e79826 100644
--- a/src/egl/main/eglconfigutil.c
+++ b/src/egl/main/eglconfigutil.c
@@ -156,6 +156,7 @@ _eglFillInConfigs(_EGLConfig * configs,
             {0x000000FF, 0x0000FF00, 0x00FF0000, 0xFF000000},  /* 8_8_8_8_REV */
          };
 
+#if 0
    static const uint32_t masks_table_bgr[8][4] = {
             {0x00000000, 0x00000000, 0x00000000, 0x00000000},
             {0x00000000, 0x00000000, 0x00000000, 0x00000000},
@@ -177,6 +178,7 @@ _eglFillInConfigs(_EGLConfig * configs,
             {0x00000000, 0x00000000, 0x00000000, 0x00000000},
             {0x00FF0000, 0x0000FF00, 0x000000FF, 0xFF000000},  /* 8_8_8_8_REV */
          };
+#endif
 
    static const uint8_t bytes_per_pixel[8] = {
             0, 0, 0, 2, 2, 4, 0, 4
diff --git a/src/egl/main/eglcontext.c b/src/egl/main/eglcontext.c
index 9ab4286d3a8..88de60d69bb 100644
--- a/src/egl/main/eglcontext.c
+++ b/src/egl/main/eglcontext.c
@@ -96,7 +96,7 @@ _eglDestroyContext(_EGLDriver *drv, EGLDisplay dpy, EGLContext ctx)
    _EGLContext *context = _eglLookupContext(ctx);
    if (context) {
       _eglUnlinkContext(context);
-      if (!context->IsBound)
+      if (!_eglIsContextBound(context))
          free(context);
       return EGL_TRUE;
    }
@@ -146,10 +146,11 @@ _eglQueryContext(_EGLDriver *drv, EGLDisplay dpy, EGLContext ctx,
  * Then, the driver will do its device-dependent Make-Current stuff.
  */
 EGLBoolean
-_eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface d,
+_eglMakeCurrent(_EGLDriver *drv, EGLDisplay display, EGLSurface d,
                 EGLSurface r, EGLContext context)
 {
    _EGLThreadInfo *t = _eglGetCurrentThread();
+   _EGLDisplay *dpy = _eglLookupDisplay(display);
    _EGLContext *ctx = _eglLookupContext(context);
    _EGLSurface *draw = _eglLookupSurface(d);
    _EGLSurface *read = _eglLookupSurface(r);
@@ -160,21 +161,23 @@ _eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface d,
 
    if (_eglIsCurrentThreadDummy())
       return _eglError(EGL_BAD_ALLOC, "eglMakeCurrent");
+   if (dpy == NULL)
+      return _eglError(EGL_BAD_DISPLAY, "eglMakeCurrent");
 
-   /* error checking */
    if (ctx) {
+      /* error checking */
+      if (ctx->Binding && ctx->Binding != t)
+         return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
       if (draw == NULL || read == NULL) {
-         _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
-         return EGL_FALSE;
-      }
-      if (draw->Config != ctx->Config) {
-         _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
-         return EGL_FALSE;
-      }
-      if (read->Config != ctx->Config) {
-         _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
-         return EGL_FALSE;
+         EGLint err = (d == EGL_NO_SURFACE || r == EGL_NO_SURFACE)
+                      ? EGL_BAD_MATCH : EGL_BAD_SURFACE;
+         return _eglError(err, "eglMakeCurrent");
       }
+      if (draw->Config != ctx->Config || read->Config != ctx->Config)
+         return _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
+      if ((draw->Binding && draw->Binding->Binding != t) ||
+          (read->Binding && read->Binding->Binding != t))
+         return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
 
 #ifdef EGL_VERSION_1_4
       /* OpenGL and OpenGL ES are conflicting */
@@ -194,6 +197,10 @@ _eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface d,
       apiIndex = _eglConvertApiToIndex(ctx->ClientAPI);
    }
    else {
+      if (context != EGL_NO_CONTEXT)
+         return _eglError(EGL_BAD_CONTEXT, "eglMakeCurrent");
+      if (draw != NULL || read != NULL)
+         return _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
       apiIndex = t->CurrentAPIIndex;
    }
 
@@ -201,60 +208,47 @@ _eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface d,
    if (oldContext) {
       oldDrawSurface = oldContext->DrawSurface;
       oldReadSurface = oldContext->ReadSurface;
-   }
+      assert(oldDrawSurface);
+      assert(oldReadSurface);
 
-   /*
-    * check if the old context or surfaces need to be deleted
-    */
-   if (oldDrawSurface != NULL) {
-      oldDrawSurface->IsBound = EGL_FALSE;
+      /* break old bindings */
+      t->CurrentContexts[apiIndex] = NULL;
+      oldContext->Binding = NULL;
+      oldContext->DrawSurface = NULL;
+      oldContext->ReadSurface = NULL;
+      oldDrawSurface->Binding = NULL;
+      oldReadSurface->Binding = NULL;
+
+      /*
+       * check if the old context or surfaces need to be deleted
+       * FIXME They are linked so that they can be unlinked.  This is ugly.
+       */
       if (!_eglIsSurfaceLinked(oldDrawSurface)) {
-         /* make sure we don't try to rebind a deleted surface */
-         if (draw == oldDrawSurface || draw == oldReadSurface) {
-            draw = NULL;
-         }
-         /* really delete surface now */
-         drv->API.DestroySurface(drv, dpy, oldDrawSurface->Handle);
+         assert(draw != oldDrawSurface && read != oldDrawSurface);
+         drv->API.DestroySurface(drv, display,
+                                 _eglLinkSurface(oldDrawSurface, dpy));
       }
-   }
-   if (oldReadSurface != NULL && oldReadSurface != oldDrawSurface) {
-      oldReadSurface->IsBound = EGL_FALSE;
-      if (!_eglIsSurfaceLinked(oldReadSurface)) {
-         /* make sure we don't try to rebind a deleted surface */
-         if (read == oldDrawSurface || read == oldReadSurface) {
-            read = NULL;
-         }
-         /* really delete surface now */
-         drv->API.DestroySurface(drv, dpy, oldReadSurface->Handle);
+      if (oldReadSurface != oldDrawSurface &&
+          !_eglIsSurfaceLinked(oldReadSurface)) {
+         assert(draw != oldReadSurface && read != oldReadSurface);
+         drv->API.DestroySurface(drv, display,
+                                 _eglLinkSurface(oldReadSurface, dpy));
       }
-   }
-   if (oldContext != NULL) {
-      oldContext->IsBound = EGL_FALSE;
       if (!_eglIsContextLinked(oldContext)) {
-         /* make sure we don't try to rebind a deleted context */
-         if (ctx == oldContext) {
-            ctx = NULL;
-         }
-         /* really delete context now */
-         drv->API.DestroyContext(drv, dpy, _eglGetContextHandle(oldContext));
+         assert(ctx != oldContext);
+         drv->API.DestroyContext(drv, display,
+                                 _eglLinkContext(oldContext, dpy));
       }
    }
 
+   /* build new bindings */
    if (ctx) {
-      /* check read/draw again, in case we deleted them above */
-      if (draw == NULL || read == NULL) {
-         _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
-         return EGL_FALSE;
-      }
+      t->CurrentContexts[apiIndex] = ctx;
+      ctx->Binding = t;
       ctx->DrawSurface = draw;
       ctx->ReadSurface = read;
-      ctx->IsBound = EGL_TRUE;
-      draw->IsBound = EGL_TRUE;
-      read->IsBound = EGL_TRUE;
-      t->CurrentContexts[apiIndex] = ctx;
-   }
-   else {
-      t->CurrentContexts[apiIndex] = NULL;
+      draw->Binding = ctx;
+      read->Binding = ctx;
    }
 
    return EGL_TRUE;
diff --git a/src/egl/main/eglcontext.h b/src/egl/main/eglcontext.h
index 2fb28d38b91..4276c0980e2 100644
--- a/src/egl/main/eglcontext.h
+++ b/src/egl/main/eglcontext.h
@@ -15,12 +15,12 @@ struct _egl_context
    _EGLDisplay *Display;
    _EGLContext *Next;
 
-   _EGLConfig *Config;
-
+   /* The bound status of the context */
+   _EGLThreadInfo *Binding;
    _EGLSurface *DrawSurface;
    _EGLSurface *ReadSurface;
 
-   EGLBoolean IsBound;
+   _EGLConfig *Config;
 
    EGLint ClientAPI; /**< EGL_OPENGL_ES_API, EGL_OPENGL_API, EGL_OPENVG_API */
    EGLint ClientVersion; /**< 1 = OpenGLES 1.x, 2 = OpenGLES 2.x */
@@ -51,4 +51,15 @@ _eglMakeCurrent(_EGLDriver *drv, EGLDisplay dpy, EGLSurface draw, EGLSurface rea
 extern EGLBoolean
 _eglCopyContextMESA(_EGLDriver *drv, EGLDisplay dpy, EGLContext source, EGLContext dest, EGLint mask);
 
+
+/**
+ * Return true if the context is bound to a thread.
+ */
+static INLINE EGLBoolean
+_eglIsContextBound(_EGLContext *ctx)
+{
+   return (ctx->Binding != NULL);
+}
+
+
 #endif /* EGLCONTEXT_INCLUDED */
diff --git a/src/egl/main/eglcurrent.c b/src/egl/main/eglcurrent.c
index 96152db19fb..4431f964f69 100644
--- a/src/egl/main/eglcurrent.c
+++ b/src/egl/main/eglcurrent.c
@@ -3,61 +3,127 @@
 #include "eglcurrent.h"
 #include "eglcontext.h"
 #include "egllog.h"
+#include "eglmutex.h"
+#include "eglglobals.h"
 
 
+/* This should be kept in sync with _eglInitThreadInfo() */
+#define _EGL_THREAD_INFO_INITIALIZER \
+   { EGL_SUCCESS, { NULL }, 1 }
+
 /* a fallback thread info to guarantee that every thread always has one */
-static _EGLThreadInfo dummy_thread;
+static _EGLThreadInfo dummy_thread = _EGL_THREAD_INFO_INITIALIZER;
 
 
 #ifdef GLX_USE_TLS
 static __thread const _EGLThreadInfo *_egl_TSD;
    __attribute__ ((tls_model("initial-exec")));
 
-static INLINE EGLBoolean _eglInitTSD(void) { return EGL_TRUE; }
-static INLINE void _eglFiniTSD(void) { }
-static INLINE void _eglSetTSD(const _EGLThreadInfo *t) { _egl_TSD = t; }
+static INLINE void _eglSetTSD(const _EGLThreadInfo *t)
+{
+   _egl_TSD = t;
+}
 
 static INLINE _EGLThreadInfo *_eglGetTSD(void)
 {
    return (_EGLThreadInfo *) _egl_TSD;
 }
 
+static INLINE void _eglFiniTSD(void)
+{
+}
+
+static INLINE EGLBoolean _eglInitTSD(void (*dtor)(_EGLThreadInfo *))
+{
+   /* TODO destroy TSD */
+   (void) dtor;
+   (void) _eglFiniTSD;
+   return EGL_TRUE;
+}
+
 #elif PTHREADS
 #include <pthread.h>
 
+static _EGL_DECLARE_MUTEX(_egl_TSDMutex);
+static EGLBoolean _egl_TSDInitialized;
 static pthread_key_t _egl_TSD;
+static void (*_egl_FreeTSD)(_EGLThreadInfo *);
 
-static INLINE EGLBoolean _eglInitTSD(void)
+static INLINE void _eglSetTSD(const _EGLThreadInfo *t)
 {
-   return (pthread_key_create(&_egl_TSD, NULL) == 0);
+   pthread_setspecific(_egl_TSD, (const void *) t);
 }
 
-static INLINE void _eglFiniTSD(void)
+static INLINE _EGLThreadInfo *_eglGetTSD(void)
 {
-   pthread_key_delete(_egl_TSD);
+   return (_EGLThreadInfo *) pthread_getspecific(_egl_TSD);
 }
 
-static INLINE void _eglSetTSD(const _EGLThreadInfo *t)
+static INLINE void _eglFiniTSD(void)
 {
-   pthread_setspecific(_egl_TSD, (const void *) t);
+   _eglLockMutex(&_egl_TSDMutex);
+   if (_egl_TSDInitialized) {
+      _EGLThreadInfo *t = _eglGetTSD();
+
+      _egl_TSDInitialized = EGL_FALSE;
+      if (t && _egl_FreeTSD)
+         _egl_FreeTSD((void *) t);
+      pthread_key_delete(_egl_TSD);
+   }
+   _eglUnlockMutex(&_egl_TSDMutex);
 }
 
-static INLINE _EGLThreadInfo *_eglGetTSD(void)
+static INLINE EGLBoolean _eglInitTSD(void (*dtor)(_EGLThreadInfo *))
 {
-   return (_EGLThreadInfo *) pthread_getspecific(_egl_TSD);
+   if (!_egl_TSDInitialized) {
+      _eglLockMutex(&_egl_TSDMutex);
+
+      /* check again after acquiring lock */
+      if (!_egl_TSDInitialized) {
+         if (pthread_key_create(&_egl_TSD, (void (*)(void *)) dtor) != 0) {
+            _eglUnlockMutex(&_egl_TSDMutex);
+            return EGL_FALSE;
+         }
+         _egl_FreeTSD = dtor;
+         _eglAddAtExitCall(_eglFiniTSD);
+         _egl_TSDInitialized = EGL_TRUE;
+      }
+
+      _eglUnlockMutex(&_egl_TSDMutex);
+   }
+
+   return EGL_TRUE;
 }
 
 #else /* PTHREADS */
 static const _EGLThreadInfo *_egl_TSD;
+static void (*_egl_FreeTSD)(_EGLThreadInfo *);
 
-static INLINE EGLBoolean _eglInitTSD(void) { return EGL_TRUE; }
-static INLINE void _eglFiniTSD(void) { }
-static INLINE void _eglSetTSD(const _EGLThreadInfo *t) { _egl_TSD = t; }
+static INLINE void _eglSetTSD(const _EGLThreadInfo *t)
+{
+   _egl_TSD = t;
+}
 
 static INLINE _EGLThreadInfo *_eglGetTSD(void)
 {
    return (_EGLThreadInfo *) _egl_TSD;
 }
+
+static INLINE void _eglFiniTSD(void)
+{
+   if (_egl_FreeTSD && _egl_TSD)
+      _egl_FreeTSD((_EGLThreadInfo *) _egl_TSD);
+}
+
+static INLINE EGLBoolean _eglInitTSD(void (*dtor)(_EGLThreadInfo *))
+{
+   if (!_egl_FreeTSD && dtor) {
+      _egl_FreeTSD = dtor;
+      _eglAddAtExitCall(_eglFiniTSD);
+   }
+   return EGL_TRUE;
+}
+
 #endif /* !PTHREADS */
 
 
@@ -98,24 +164,17 @@ _eglDestroyThreadInfo(_EGLThreadInfo *t)
 
 
 /**
- * Initialize "current thread" management.
+ * Make sure TSD is initialized and return current value.
  */
-EGLBoolean
-_eglInitCurrent(void)
+static INLINE _EGLThreadInfo *
+_eglCheckedGetTSD(void)
 {
-   _eglInitThreadInfo(&dummy_thread);
-   return _eglInitTSD();
-}
-
+   if (_eglInitTSD(&_eglDestroyThreadInfo) != EGL_TRUE) {
+      _eglLog(_EGL_FATAL, "failed to initialize \"current\" system");
+      return NULL;
+   }
 
-/**
- * Finish "current thread" management.
- */
-void
-_eglFiniCurrent(void)
-{
-   /* TODO trace and release all threads... */
-   _eglFiniTSD();
+   return _eglGetTSD();
 }
 
 
@@ -129,7 +188,7 @@ _eglFiniCurrent(void)
 _EGLThreadInfo *
 _eglGetCurrentThread(void)
 {
-   _EGLThreadInfo *t = _eglGetTSD();
+   _EGLThreadInfo *t = _eglCheckedGetTSD();
    if (!t) {
       t = _eglCreateThreadInfo();
       _eglSetTSD(t);
@@ -145,7 +204,7 @@ _eglGetCurrentThread(void)
 void
 _eglDestroyCurrentThread(void)
 {
-   _EGLThreadInfo *t = _eglGetTSD();
+   _EGLThreadInfo *t = _eglCheckedGetTSD();
    if (t) {
       _eglDestroyThreadInfo(t);
       _eglSetTSD(NULL);
@@ -162,7 +221,7 @@ _eglDestroyCurrentThread(void)
 EGLBoolean
 _eglIsCurrentThreadDummy(void)
 {
-   _EGLThreadInfo *t = _eglGetTSD();
+   _EGLThreadInfo *t = _eglCheckedGetTSD();
    return (!t || t == &dummy_thread);
 }
 
diff --git a/src/egl/main/eglcurrent.h b/src/egl/main/eglcurrent.h
index f9fdf7bd0f0..8eb241029ec 100644
--- a/src/egl/main/eglcurrent.h
+++ b/src/egl/main/eglcurrent.h
@@ -20,14 +20,6 @@ struct _egl_thread_info
 };
 
 
-extern EGLBoolean
-_eglInitCurrent(void);
-
-
-extern void
-_eglFiniCurrent(void);
-
-
 /**
  * Return true if a client API enum can be converted to an index.
  */
diff --git a/src/egl/main/egldisplay.c b/src/egl/main/egldisplay.c
index 89de609d0b4..feae1d60409 100644
--- a/src/egl/main/egldisplay.c
+++ b/src/egl/main/egldisplay.c
@@ -1,4 +1,3 @@
-
 /**
  * Functions related to EGLDisplay.
  */
@@ -13,6 +12,67 @@
 #include "eglglobals.h"
 #include "eglhash.h"
 #include "eglstring.h"
+#include "eglmutex.h"
+#include "egllog.h"
+
+
+static _EGL_DECLARE_MUTEX(_eglDisplayInitMutex);
+static _EGLHashtable *_eglDisplayHash;
+/* TODO surface hash table should be per-display */
+static _EGLHashtable *_eglSurfaceHash;
+
+
+/**
+ * Finish display management.
+ */
+static void
+_eglFiniDisplay(void)
+{
+   _eglLockMutex(&_eglDisplayInitMutex);
+   if (_eglDisplayHash) {
+      EGLuint key = _eglHashFirstEntry(_eglDisplayHash);
+
+      while (key) {
+         _EGLDisplay *dpy = (_EGLDisplay *)
+            _eglHashLookup(_eglDisplayHash, key);
+         assert(dpy);
+
+         if (dpy->ContextList || dpy->SurfaceList)
+            _eglLog(_EGL_DEBUG, "Display %u is destroyed with resources", key);
+
+         _eglCleanupDisplay(dpy);
+         free(dpy);
+
+         key = _eglHashNextEntry(_eglDisplayHash, key);
+      }
+
+      _eglDeleteHashTable(_eglDisplayHash);
+      _eglDisplayHash = NULL;
+      _eglDeleteHashTable(_eglSurfaceHash);
+      _eglSurfaceHash = NULL;
+   }
+   _eglUnlockMutex(&_eglDisplayInitMutex);
+}
+
+
+/* This can be avoided if hash table can be statically initialized */
+static INLINE void
+_eglInitDisplay(void)
+{
+   if (!_eglDisplayHash) {
+      _eglLockMutex(&_eglDisplayInitMutex);
+
+      /* check again after acquiring lock */
+      if (!_eglDisplayHash) {
+         _eglDisplayHash = _eglNewHashTable();
+         _eglSurfaceHash = _eglNewHashTable();
+
+         _eglAddAtExitCall(_eglFiniDisplay);
+      }
+
+      _eglUnlockMutex(&_eglDisplayInitMutex);
+   }
+}
 
 
 /**
@@ -31,6 +91,9 @@ _eglNewDisplay(NativeDisplayType nativeDisplay)
       dpy->Xdpy = (Display *) nativeDisplay;
 #endif
 
+      _eglInitDisplay();
+      dpy->SurfaceHash = _eglSurfaceHash;
+
       dpy->DriverName = _eglChooseDriver(dpy);
       if (!dpy->DriverName) {
          free(dpy);
@@ -49,11 +112,14 @@ EGLDisplay
 _eglLinkDisplay(_EGLDisplay *dpy)
 {
    EGLuint key;
-   key = _eglHashGenKey(_eglGlobal.Displays);
+
+   _eglInitDisplay();
+
+   key = _eglHashGenKey(_eglDisplayHash);
    assert(key);
    /* "link" the display to the hash table */
-   _eglHashInsert(_eglGlobal.Displays, key, dpy);
-   dpy->Handle = (EGLDisplay) key;
+   _eglHashInsert(_eglDisplayHash, key, dpy);
+   dpy->Handle = (EGLDisplay) _eglUIntToPointer(key);
 
    return dpy->Handle;
 }
@@ -66,7 +132,11 @@ _eglLinkDisplay(_EGLDisplay *dpy)
 void
 _eglUnlinkDisplay(_EGLDisplay *dpy)
 {
-   _eglHashRemove(_eglGlobal.Displays, (EGLuint) dpy->Handle);
+   EGLuint key = _eglPointerToUInt((void *) dpy->Handle);
+
+   _eglInitDisplay();
+
+   _eglHashRemove(_eglDisplayHash, key);
    dpy->Handle = EGL_NO_DISPLAY;
 }
 
@@ -83,7 +153,7 @@ _eglGetDisplayHandle(_EGLDisplay *display)
       return EGL_NO_DISPLAY;
 }
 
- 
+
 /**
  * Lookup a handle to find the linked display.
  * Return NULL if the handle has no corresponding linked display.
@@ -91,8 +161,11 @@ _eglGetDisplayHandle(_EGLDisplay *display)
 _EGLDisplay *
 _eglLookupDisplay(EGLDisplay dpy)
 {
-   EGLuint key = (EGLuint) dpy;
-   return (_EGLDisplay *) _eglHashLookup(_eglGlobal.Displays, key);
+   EGLuint key = _eglPointerToUInt((void *) dpy);
+
+   _eglInitDisplay();
+
+   return (_EGLDisplay *) _eglHashLookup(_eglDisplayHash, key);
 }
 
 
@@ -103,17 +176,20 @@ _eglLookupDisplay(EGLDisplay dpy)
 _EGLDisplay *
 _eglFindDisplay(NativeDisplayType nativeDisplay)
 {
-   EGLuint key = _eglHashFirstEntry(_eglGlobal.Displays);
+   EGLuint key;
+
+   _eglInitDisplay();
 
    /* Walk the hash table.  Should switch to list if it is a problem. */
+   key = _eglHashFirstEntry(_eglDisplayHash);
    while (key) {
       _EGLDisplay *dpy = (_EGLDisplay *)
-            _eglHashLookup(_eglGlobal.Displays, key);
+            _eglHashLookup(_eglDisplayHash, key);
       assert(dpy);
 
       if (dpy->NativeDisplay == nativeDisplay)
          return dpy;
-      key = _eglHashNextEntry(_eglGlobal.Displays, key);
+      key = _eglHashNextEntry(_eglDisplayHash, key);
    }
 
    return NULL;
@@ -224,7 +300,7 @@ _eglUnlinkContext(_EGLContext *ctx)
 EGLContext
 _eglGetContextHandle(_EGLContext *ctx)
 {
-   return (EGLContext) (ctx && ctx->Display) ? ctx : EGL_NO_CONTEXT;
+   return (EGLContext) ((ctx && ctx->Display) ? ctx : EGL_NO_CONTEXT);
 }
 
 
@@ -253,11 +329,11 @@ _eglLinkSurface(_EGLSurface *surf, _EGLDisplay *dpy)
    surf->Next = dpy->SurfaceList;
    dpy->SurfaceList = surf;
 
-   key = _eglHashGenKey(_eglGlobal.Surfaces);
+   key = _eglHashGenKey(dpy->SurfaceHash);
    assert(key);
-   _eglHashInsert(_eglGlobal.Surfaces, key, surf);
+   _eglHashInsert(dpy->SurfaceHash, key, surf);
 
-   surf->Handle = (EGLSurface) key;
+   surf->Handle = (EGLSurface) _eglUIntToPointer(key);
    return surf->Handle;
 }
 
@@ -270,8 +346,9 @@ void
 _eglUnlinkSurface(_EGLSurface *surf)
 {
    _EGLSurface *prev;
+   EGLuint key = _eglPointerToUInt((void *) surf->Handle);
 
-   _eglHashRemove(_eglGlobal.Surfaces, (EGLuint) surf->Handle);
+   _eglHashRemove(surf->Display->SurfaceHash, key);
    surf->Handle = EGL_NO_SURFACE;
 
    prev = surf->Display->SurfaceList;
@@ -314,7 +391,6 @@ _eglGetSurfaceHandle(_EGLSurface *surface)
 _EGLSurface *
 _eglLookupSurface(EGLSurface surf)
 {
-   _EGLSurface *c = (_EGLSurface *) _eglHashLookup(_eglGlobal.Surfaces,
-                                                   (EGLuint) surf);
-   return c;
+   EGLuint key = _eglPointerToUInt((void *) surf);
+   return (_EGLSurface *) _eglHashLookup(_eglSurfaceHash, key);
 }
diff --git a/src/egl/main/egldisplay.h b/src/egl/main/egldisplay.h
index 372ed3cd79a..70c59ef5e46 100644
--- a/src/egl/main/egldisplay.h
+++ b/src/egl/main/egldisplay.h
@@ -6,6 +6,7 @@
 #endif
 
 #include "egltypedefs.h"
+#include "eglhash.h"
 
 
 struct _egl_display 
@@ -26,6 +27,10 @@ struct _egl_display
    /* lists of linked contexts and surface */
    _EGLContext *ContextList;
    _EGLSurface *SurfaceList;
+
+   /* hash table to map surfaces to handles */
+   _EGLHashtable *SurfaceHash;
+
 #ifdef _EGL_PLATFORM_X
    Display *Xdpy;
 #endif
@@ -125,4 +130,25 @@ _eglIsSurfaceLinked(_EGLSurface *surf)
 }
 
 
+/**
+ * Cast an unsigned int to a pointer.
+ */
+static INLINE void *
+_eglUIntToPointer(unsigned int v)
+{
+   return (void *) ((uintptr_t) v);
+}
+
+
+/**
+ * Cast a pointer to an unsigned int.  The pointer must be one that is
+ * returned by _eglUIntToPointer.
+ */
+static INLINE unsigned int
+_eglPointerToUInt(const void *p)
+{
+   return (unsigned int) ((uintptr_t) p);
+}
+
+
 #endif /* EGLDISPLAY_INCLUDED */
diff --git a/src/egl/main/eglglobals.c b/src/egl/main/eglglobals.c
index 55de394ef5f..e93b48e03b8 100644
--- a/src/egl/main/eglglobals.c
+++ b/src/egl/main/eglglobals.c
@@ -1,41 +1,53 @@
 #include <stdlib.h>
+#include <assert.h>
 #include "eglglobals.h"
+#include "egldisplay.h"
 #include "egllog.h"
+#include "eglmutex.h"
 
-struct _egl_global _eglGlobal = 
-{
-   EGL_FALSE
-};
 
-/**
- * Init the fields in the _eglGlobal struct
- * May be safely called more than once.
- */
-void
-_eglInitGlobals(void)
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
+
+
+static _EGL_DECLARE_MUTEX(_eglGlobalMutex);
+struct _egl_global _eglGlobal =
 {
-   if (!_eglGlobal.Initialized) {
-      _eglGlobal.Displays = _eglNewHashTable();
-      _eglGlobal.Surfaces = _eglNewHashTable();
-      _eglGlobal.FreeScreenHandle = 1;
-      _eglGlobal.Initialized = EGL_TRUE;
+   &_eglGlobalMutex,       /* Mutex */
+   1,                      /* FreeScreenHandle */
+   0x0,                    /* ClientAPIsMask */
+   { 0x0 },                /* ClientAPIs */
+   0,                      /* NumDrivers */
+   { NULL },               /* Drivers */
+   0,                      /* NumAtExitCalls */
+   { NULL },               /* AtExitCalls */
+};
 
-      _eglGlobal.ClientAPIsMask = 0x0;
 
-      if (!_eglInitCurrent())
-         _eglLog(_EGL_FATAL, "failed to initialize \"current\" system");
-   }
+static void
+_eglAtExit(void)
+{
+   EGLint i;
+   for (i = _eglGlobal.NumAtExitCalls - 1; i >= 0; i--)
+      _eglGlobal.AtExitCalls[i]();
 }
 
 
-/**
- * Should call this via an atexit handler.
- */
 void
-_eglDestroyGlobals(void)
+_eglAddAtExitCall(void (*func)(void))
 {
-   _eglFiniCurrent();
-   /* XXX TODO walk over table entries, deleting each */
-   _eglDeleteHashTable(_eglGlobal.Displays);
-   _eglDeleteHashTable(_eglGlobal.Surfaces);
+   if (func) {
+      static EGLBoolean registered = EGL_FALSE;
+
+      _eglLockMutex(_eglGlobal.Mutex);
+
+      if (!registered) {
+         atexit(_eglAtExit);
+         registered = EGL_TRUE;
+      }
+
+      assert(_eglGlobal.NumAtExitCalls < ARRAY_SIZE(_eglGlobal.AtExitCalls));
+      _eglGlobal.AtExitCalls[_eglGlobal.NumAtExitCalls++] = func;
+
+      _eglUnlockMutex(_eglGlobal.Mutex);
+   }
 }
diff --git a/src/egl/main/eglglobals.h b/src/egl/main/eglglobals.h
index a9443cfbdd8..1e2c6742630 100644
--- a/src/egl/main/eglglobals.h
+++ b/src/egl/main/eglglobals.h
@@ -4,6 +4,7 @@
 #include "egltypedefs.h"
 #include "eglhash.h"
 #include "eglcurrent.h"
+#include "eglmutex.h"
 
 
 /**
@@ -11,12 +12,7 @@
  */
 struct _egl_global
 {
-   EGLBoolean Initialized;
-
-   /* these are private to egldisplay.c */
-   _EGLHashtable *Displays;
-   _EGLHashtable *Surfaces;
-
+   _EGLMutex *Mutex;
    EGLScreenMESA FreeScreenHandle;
 
    /* bitmaks of supported APIs (supported by _some_ driver) */
@@ -26,6 +22,9 @@ struct _egl_global
 
    EGLint NumDrivers;
    _EGLDriver *Drivers[10];
+
+   EGLint NumAtExitCalls;
+   void (*AtExitCalls[10])(void);
 };
 
 
@@ -33,11 +32,7 @@ extern struct _egl_global _eglGlobal;
 
 
 extern void
-_eglInitGlobals(void);
-
-
-extern void
-_eglDestroyGlobals(void);
+_eglAddAtExitCall(void (*func)(void));
 
 
 #endif /* EGLGLOBALS_INCLUDED */
diff --git a/src/egl/main/eglmutex.h b/src/egl/main/eglmutex.h
new file mode 100644
index 00000000000..29faba0f241
--- /dev/null
+++ b/src/egl/main/eglmutex.h
@@ -0,0 +1,52 @@
+#ifndef EGLMUTEX_INCLUDED
+#define EGLMUTEX_INCLUDED
+
+#include "eglcompiler.h"
+
+#ifdef PTHREADS
+#include <pthread.h>
+
+typedef pthread_mutex_t _EGLMutex;
+
+static INLINE void _eglInitMutex(_EGLMutex *m)
+{
+   pthread_mutex_init(m, NULL);
+}
+
+static INLINE void
+_eglDestroyMutex(_EGLMutex *m)
+{
+   pthread_mutex_destroy(m);
+}
+
+static INLINE void
+_eglLockMutex(_EGLMutex *m)
+{
+   pthread_mutex_lock(m);
+}
+
+static INLINE void
+_eglUnlockMutex(_EGLMutex *m)
+{
+   pthread_mutex_unlock(m);
+}
+
+#define _EGL_MUTEX_INITIALIZER PTHREAD_MUTEX_INITIALIZER
+#define _EGL_DECLARE_MUTEX(m) \
+   _EGLMutex m = _EGL_MUTEX_INITIALIZER
+
+#else
+
+typedef int _EGLMutex;
+static INLINE void _eglInitMutex(_EGLMutex *m) { (void) m; }
+static INLINE void _eglDestroyMutex(_EGLMutex *m) { (void) m; }
+static INLINE void _eglLockMutex(_EGLMutex *m) { (void) m; }
+static INLINE void _eglUnlockMutex(_EGLMutex *m) { (void) m; }
+
+#define _EGL_MUTEX_INITIALIZER 0
+#define _EGL_DECLARE_MUTEX(m) \
+   _EGLMutex m = _EGL_MUTEX_INITIALIZER
+
+#endif
+
+#endif /* EGLMUTEX_INCLUDED */
diff --git a/src/egl/main/eglsurface.c b/src/egl/main/eglsurface.c
index 9821e636287..39470511276 100644
--- a/src/egl/main/eglsurface.c
+++ b/src/egl/main/eglsurface.c
@@ -26,7 +26,8 @@ _eglInitSurface(_EGLDriver *drv, _EGLSurface *surf, EGLint type,
 {
    const char *func;
    EGLint width = 0, height = 0, largest = 0;
-   EGLint texFormat = 0, texTarget = 0, mipmapTex = 0;
+   EGLint texFormat = EGL_NO_TEXTURE, texTarget = EGL_NO_TEXTURE;
+   EGLint mipmapTex = EGL_FALSE;
    EGLint renderBuffer = EGL_BACK_BUFFER;
 #ifdef EGL_VERSION_1_2
    EGLint colorspace = EGL_COLORSPACE_sRGB;
@@ -413,7 +414,7 @@ _eglDestroySurface(_EGLDriver *drv, EGLDisplay dpy, EGLSurface surface)
    _EGLSurface *surf = _eglLookupSurface(surface);
    if (surf) {
       _eglUnlinkSurface(surf);
-      if (!surf->IsBound)
+      if (!_eglIsSurfaceBound(surf))
          free(surf);
       return EGL_TRUE;
    }
diff --git a/src/egl/main/eglsurface.h b/src/egl/main/eglsurface.h
index f9413eb9d75..88641768449 100644
--- a/src/egl/main/eglsurface.h
+++ b/src/egl/main/eglsurface.h
@@ -15,12 +15,12 @@ struct _egl_surface
    _EGLSurface *Next;
    EGLSurface Handle;
 
-   _EGLConfig *Config;
-
-   /* May need reference counting here */
-   EGLBoolean IsBound;
+   /* The bound status of the surface */
+   _EGLContext *Binding;
    EGLBoolean BoundToTexture;
 
+   _EGLConfig *Config;
+
    EGLint Type; /* one of EGL_WINDOW_BIT, EGL_PIXMAP_BIT or EGL_PBUFFER_BIT */
    EGLint Width, Height;
    EGLint TextureFormat, TextureTarget;
@@ -100,5 +100,16 @@ _eglCreatePbufferFromClientBuffer(_EGLDriver *drv, EGLDisplay dpy,
 #endif /* EGL_VERSION_1_2 */
 
 
+/**
+ * Return true if the surface is bound to a thread.
+ * A surface bound to a texutre is not considered bound by
+ * this function.
+ */
+static INLINE EGLBoolean
+_eglIsSurfaceBound(_EGLSurface *surf)
+{
+   return (surf->Binding != NULL);
+}
+
 
 #endif /* EGLSURFACE_INCLUDED */
diff --git a/src/gallium/Makefile.template b/src/gallium/Makefile.template
index 98487d43bd6..2e3da436cd7 100644
--- a/src/gallium/Makefile.template
+++ b/src/gallium/Makefile.template
@@ -31,8 +31,8 @@ INCLUDES = \
 
 default: depend lib$(LIBNAME).a
 
-lib$(LIBNAME).a: $(OBJECTS) Makefile $(TOP)/src/gallium/Makefile.template
-	$(MKLIB) -o $(LIBNAME) -static $(OBJECTS)
+lib$(LIBNAME).a: $(OBJECTS) $(EXTRA_OBJECTS) Makefile $(TOP)/src/gallium/Makefile.template
+	$(MKLIB) -o $(LIBNAME) -static $(OBJECTS) $(EXTRA_OBJECTS)
 
 depend: $(C_SOURCES) $(CPP_SOURCES) $(ASM_SOURCES) $(SYMLINKS)
 	rm -f depend
diff --git a/src/gallium/auxiliary/draw/draw_pipe.c b/src/gallium/auxiliary/draw/draw_pipe.c
index 2e3f5b2fc07..1c6d657297c 100644
--- a/src/gallium/auxiliary/draw/draw_pipe.c
+++ b/src/gallium/auxiliary/draw/draw_pipe.c
@@ -158,6 +158,60 @@ static void do_triangle( struct draw_context *draw,
 
 
 
+#define QUAD(i0,i1,i2,i3)                       \
+   do_triangle( draw,                           \
+                ( DRAW_PIPE_RESET_STIPPLE |     \
+                  DRAW_PIPE_EDGE_FLAG_0 |       \
+                  DRAW_PIPE_EDGE_FLAG_2 ),      \
+                verts + stride * elts[i0],      \
+                verts + stride * elts[i1],      \
+                verts + stride * elts[i3]);     \
+   do_triangle( draw,                           \
+                ( DRAW_PIPE_EDGE_FLAG_0 |       \
+                  DRAW_PIPE_EDGE_FLAG_1 ),      \
+                verts + stride * elts[i1],      \
+                verts + stride * elts[i2],      \
+                verts + stride * elts[i3])
+
+#define TRIANGLE(flags,i0,i1,i2)                                        \
+   do_triangle( draw,                                                   \
+                elts[i0],  /* flags */                          \
+                verts + stride * (elts[i0] & ~DRAW_PIPE_FLAG_MASK),     \
+                verts + stride * elts[i1],                              \
+                verts + stride * elts[i2])
+
+#define LINE(flags,i0,i1)                                       \
+   do_line( draw,                                               \
+            elts[i0],                                   \
+            verts + stride * (elts[i0] & ~DRAW_PIPE_FLAG_MASK), \
+            verts + stride * elts[i1])
+
+#define POINT(i0)                               \
+   do_point( draw,                              \
+             verts + stride * elts[i0] )
+
+#define FUNC pipe_run
+#define ARGS                                    \
+    struct draw_context *draw,                  \
+    unsigned prim,                              \
+    struct vertex_header *vertices,             \
+    unsigned stride,                            \
+    const ushort *elts
+
+#define LOCAL_VARS                                           \
+   char *verts = (char *)vertices;                           \
+   boolean flatfirst = (draw->rasterizer->flatshade &&       \
+                        draw->rasterizer->flatshade_first);  \
+   unsigned i;                                               \
+   ushort flags
+
+#define FLUSH
+
+#include "draw_pt_decompose.h"
+#undef ARGS
+#undef LOCAL_VARS
+
+
 
 /* Code to run the pipeline on a fairly arbitary collection of vertices.
  *
@@ -178,34 +232,12 @@ void draw_pipeline_run( struct draw_context *draw,
                         unsigned count )
 {
    char *verts = (char *)vertices;
-   unsigned i;
 
    draw->pipeline.verts = verts;
    draw->pipeline.vertex_stride = stride;
    draw->pipeline.vertex_count = vertex_count;
    
-   switch (prim) {
-   case PIPE_PRIM_POINTS:
-      for (i = 0; i < count; i++) 
-         do_point( draw, 
-                   verts + stride * elts[i] );
-      break;
-   case PIPE_PRIM_LINES:
-      for (i = 0; i+1 < count; i += 2) 
-         do_line( draw, 
-                  elts[i+0],  /* flags */
-                  verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK),
-                  verts + stride * elts[i+1]);
-      break;
-   case PIPE_PRIM_TRIANGLES:
-      for (i = 0; i+2 < count; i += 3)
-         do_triangle( draw, 
-                      elts[i+0],  /* flags */
-                      verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK),
-                      verts + stride * elts[i+1],
-                      verts + stride * elts[i+2]);
-      break;
-   }
+   pipe_run(draw, prim, vertices, stride, elts, count);
    
    draw->pipeline.verts = NULL;
    draw->pipeline.vertex_count = 0;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
index 30a6d2919d9..283502cdf3e 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
@@ -256,7 +256,10 @@ pstip_transform_inst(struct tgsi_transform_context *ctx,
          uint size = 4;
          immed = tgsi_default_full_immediate();
          immed.Immediate.NrTokens = 1 + size; /* one for the token itself */
-         immed.u.Pointer = (void *) value;
+         immed.u[0].Float = value[0];
+         immed.u[1].Float = value[1];
+         immed.u[2].Float = value[2];
+         immed.u[3].Float = value[3];
          ctx->emit_immediate(ctx, &immed);
       }
 
diff --git a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
index 014d8c7346d..7d76a7dbf39 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
@@ -28,6 +28,30 @@
 /* Authors:  Keith Whitwell <keith@tungstengraphics.com>
  */
 
+/**
+ * Notes on wide points and sprite mode:
+ *
+ * In wide point/sprite mode we effectively need to convert each incoming
+ * vertex into four outgoing vertices specifying the corners of a quad.
+ * Since we don't (yet) have geometry shaders, we have to handle this here
+ * in the draw module.
+ *
+ * For sprites, it also means that this is where we have to handle texcoords
+ * for the vertices of the quad.  OpenGL's GL_COORD_REPLACE state specifies
+ * if/how enabled texcoords are automatically generated for sprites.  We pass
+ * that info through gallium in the pipe_rasterizer_state::sprite_coord_mode
+ * array.
+ *
+ * Additionally, GLSL's gl_PointCoord fragment attribute has to be handled
+ * here as well.  This is basically an additional texture/generic attribute
+ * that varies .x from 0 to 1 horizontally across the point and varies .y
+ * vertically from 0 to 1 down the sprite.
+ *
+ * With geometry shaders, the state tracker could create a GS to do
+ * most/all of this.
+ */
+
+
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "pipe/p_defines.h"
@@ -52,7 +76,7 @@ struct widepoint_stage {
 
    int psize_slot;
 
-   int point_coord_fs_input;  /**< input for pointcoord (and fog) */
+   int point_coord_fs_input;  /**< input for pointcoord */
 };
 
 
@@ -64,8 +88,6 @@ widepoint_stage( struct draw_stage *stage )
 }
 
 
-
-
 /**
  * Set the vertex texcoords for sprite mode.
  * Coords may be left untouched or set to a right-side-up or upside-down
@@ -89,10 +111,12 @@ static void set_texcoords(const struct widepoint_stage *wide,
    }
 
    if (wide->point_coord_fs_input >= 0) {
-      /* put gl_PointCoord into extra vertex output's zw components */
-      uint k = wide->stage.draw->extra_vp_outputs.slot;
-      v->data[k][2] = tc[0];
-      v->data[k][3] = tc[1];
+      /* put gl_PointCoord into the extra vertex slot */
+      uint slot = wide->stage.draw->extra_vp_outputs.slot;
+      v->data[slot][0] = tc[0];
+      v->data[slot][1] = tc[1];
+      v->data[slot][2] = 0.0F;
+      v->data[slot][3] = 1.0F;
    }
 }
 
@@ -182,10 +206,10 @@ static void widepoint_point( struct draw_stage *stage,
 
 
 static int
-find_fog_input_attrib(struct draw_context *draw)
+find_pntc_input_attrib(struct draw_context *draw)
 {
-   /* Scan the fragment program's input decls to find the fogcoord
-    * attribute.  The z/w components will store the point coord.
+   /* Scan the fragment program's input decls to find the pointcoord
+    * attribute.  The xy components will store the point coord.
     */
    return 0; /* XXX fix this */
 }
@@ -229,8 +253,8 @@ static void widepoint_first_point( struct draw_stage *stage,
       }
       wide->num_texcoords = j;
 
-      /* find fragment shader PointCoord/Fog input */
-      wide->point_coord_fs_input = find_fog_input_attrib(draw);
+      /* find fragment shader PointCoord input */
+      wide->point_coord_fs_input = find_pntc_input_attrib(draw);
 
       /* setup extra vp output (point coord implemented as a texcoord) */
       draw->extra_vp_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index 9e37a26c1e2..62e04a65f30 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -1758,24 +1758,24 @@ emit_instruction( struct aos_compilation *cp,
    case TGSI_OPCODE_SUB:
       return emit_SUB(cp, inst);
  
-   case TGSI_OPCODE_LERP:
+   case TGSI_OPCODE_LRP:
 //      return emit_LERP(cp, inst);
       return FALSE;
 
-   case TGSI_OPCODE_FRAC:
+   case TGSI_OPCODE_FRC:
       return emit_FRC(cp, inst);
 
    case TGSI_OPCODE_CLAMP:
 //      return emit_CLAMP(cp, inst);
       return FALSE;
 
-   case TGSI_OPCODE_FLOOR:
+   case TGSI_OPCODE_FLR:
       return emit_FLR(cp, inst);
 
    case TGSI_OPCODE_ROUND:
       return emit_RND(cp, inst);
 
-   case TGSI_OPCODE_EXPBASE2:
+   case TGSI_OPCODE_EX2:
 #if FAST_MATH
       return emit_EXPBASE2(cp, inst);
 #elif 0
@@ -1787,13 +1787,13 @@ emit_instruction( struct aos_compilation *cp,
       return FALSE;
 #endif
 
-   case TGSI_OPCODE_LOGBASE2:
+   case TGSI_OPCODE_LG2:
       return emit_LG2(cp, inst);
 
-   case TGSI_OPCODE_POWER:
+   case TGSI_OPCODE_POW:
       return emit_POW(cp, inst);
 
-   case TGSI_OPCODE_CROSSPRODUCT:
+   case TGSI_OPCODE_XPD:
       return emit_XPD(cp, inst);
 
    case TGSI_OPCODE_ABS:
@@ -1891,8 +1891,9 @@ static boolean note_immediate( struct aos_compilation *cp,
    unsigned pos = cp->num_immediates++;
    unsigned j;
 
+   assert( imm->Immediate.NrTokens <= 4 + 1 );
    for (j = 0; j < imm->Immediate.NrTokens - 1; j++) {
-      cp->vaos->machine->immediate[pos][j] = imm->u.ImmediateFloat32[j].Float;
+      cp->vaos->machine->immediate[pos][j] = imm->u[j].Float;
    }
 
    return TRUE;
diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c
index 727977bc3af..b3535c0e48e 100644
--- a/src/gallium/auxiliary/draw/draw_vs_llvm.c
+++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c
@@ -119,7 +119,7 @@ draw_create_vs_llvm(struct draw_context *draw,
    vs->base.create_varient = draw_vs_varient_generic;
    vs->base.run_linear = vs_llvm_run_linear;
    vs->base.delete = vs_llvm_delete;
-   vs->machine = &draw->vs.machine;
+   vs->machine = draw->vs.machine;
 
    {
       struct gallivm_ir *ir = gallivm_ir_new(GALLIVM_VS);
diff --git a/src/gallium/auxiliary/gallivm/instructionssoa.cpp b/src/gallium/auxiliary/gallivm/instructionssoa.cpp
index 2d2af3085e6..721b7d2d833 100644
--- a/src/gallium/auxiliary/gallivm/instructionssoa.cpp
+++ b/src/gallium/auxiliary/gallivm/instructionssoa.cpp
@@ -128,7 +128,7 @@ void InstructionsSoa::createFunctionMap()
    m_functionsMap[TGSI_OPCODE_DP4]   = "dp4";
    m_functionsMap[TGSI_OPCODE_MIN]   = "min";
    m_functionsMap[TGSI_OPCODE_MAX]   = "max";
-   m_functionsMap[TGSI_OPCODE_POWER] = "pow";
+   m_functionsMap[TGSI_OPCODE_POW]   = "pow";
    m_functionsMap[TGSI_OPCODE_LIT]   = "lit";
    m_functionsMap[TGSI_OPCODE_RSQ]   = "rsq";
    m_functionsMap[TGSI_OPCODE_SLT]   = "slt";
@@ -311,7 +311,7 @@ std::vector<llvm::Value*> InstructionsSoa::mul(const std::vector<llvm::Value*> i
 std::vector<llvm::Value*> InstructionsSoa::pow(const std::vector<llvm::Value*> in1,
                                                const std::vector<llvm::Value*> in2)
 {
-   llvm::Function *func = function(TGSI_OPCODE_POWER);
+   llvm::Function *func = function(TGSI_OPCODE_POW);
    return callBuiltin(func, in1, in2);
 }
 
diff --git a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
index 5b08200d142..bf84401e112 100644
--- a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
+++ b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
@@ -160,10 +160,11 @@ translate_immediate(Storage *storage,
 {
    float vec[4];
    int i;
+   assert( imm->Immediate.NrTokens <= 4 + 1 );
    for (i = 0; i < imm->Immediate.NrTokens - 1; ++i) {
       switch (imm->Immediate.DataType) {
       case TGSI_IMM_FLOAT32:
-         vec[i] = imm->u.ImmediateFloat32[i].Float;
+         vec[i] = imm->u[i].Float;
          break;
       default:
          assert(0);
@@ -179,10 +180,11 @@ translate_immediateir(StorageSoa *storage,
 {
    float vec[4];
    int i;
+   assert( imm->Immediate.NrTokens <= 4 + 1 );
    for (i = 0; i < imm->Immediate.NrTokens - 1; ++i) {
       switch (imm->Immediate.DataType) {
       case TGSI_IMM_FLOAT32:
-         vec[i] = imm->u.ImmediateFloat32[i].Float;
+         vec[i] = imm->u[i].Float;
          break;
       default:
          assert(0);
@@ -336,7 +338,7 @@ translate_instruction(llvm::Module *module,
       out = instr->sub(inputs[0], inputs[1]);
    }
       break;
-   case TGSI_OPCODE_LERP: {
+   case TGSI_OPCODE_LRP: {
       out = instr->lerp(inputs[0], inputs[1], inputs[2]);
    }
       break;
@@ -348,17 +350,11 @@ translate_instruction(llvm::Module *module,
       out = instr->cnd0(inputs[0], inputs[1], inputs[2]);
    }
       break;
-   case TGSI_OPCODE_DOT2ADD: {
+   case TGSI_OPCODE_DP2A: {
       out = instr->dot2add(inputs[0], inputs[1], inputs[2]);
    }
       break;
-   case TGSI_OPCODE_INDEX:
-      break;
-   case TGSI_OPCODE_NEGATE: {
-      out = instr->neg(inputs[0]);
-   }
-      break;
-   case TGSI_OPCODE_FRAC: {
+   case TGSI_OPCODE_FRC: {
       out = instr->frc(inputs[0]);
    }
       break;
@@ -366,30 +362,28 @@ translate_instruction(llvm::Module *module,
       out = instr->clamp(inputs[0]);
    }
       break;
-   case TGSI_OPCODE_FLOOR: {
+   case TGSI_OPCODE_FLR: {
       out = instr->floor(inputs[0]);
    }
       break;
    case TGSI_OPCODE_ROUND:
       break;
-   case TGSI_OPCODE_EXPBASE2: {
+   case TGSI_OPCODE_EX2: {
       out = instr->ex2(inputs[0]);
    }
       break;
-   case TGSI_OPCODE_LOGBASE2: {
+   case TGSI_OPCODE_LG2: {
       out = instr->lg2(inputs[0]);
    }
       break;
-   case TGSI_OPCODE_POWER: {
+   case TGSI_OPCODE_POW: {
       out = instr->pow(inputs[0], inputs[1]);
    }
       break;
-   case TGSI_OPCODE_CROSSPRODUCT: {
+   case TGSI_OPCODE_XPD: {
       out = instr->cross(inputs[0], inputs[1]);
    }
       break;
-   case TGSI_OPCODE_MULTIPLYMATRIX:
-      break;
    case TGSI_OPCODE_ABS: {
       out = instr->abs(inputs[0]);
    }
@@ -522,7 +516,7 @@ translate_instruction(llvm::Module *module,
       return;  //just update the state
    }
       break;
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       break;
    case TGSI_OPCODE_REP:
       break;
@@ -538,7 +532,7 @@ translate_instruction(llvm::Module *module,
       return; //just update the state
    }
       break;
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       break;
    case TGSI_OPCODE_ENDREP:
       break;
@@ -580,7 +574,7 @@ translate_instruction(llvm::Module *module,
       break;
    case TGSI_OPCODE_ENDPRIM:
       break;
-   case TGSI_OPCODE_BGNLOOP2: {
+   case TGSI_OPCODE_BGNLOOP: {
       instr->beginLoop();
       storage->setCurrentBlock(instr->currentBlock());
       return;
@@ -593,7 +587,7 @@ translate_instruction(llvm::Module *module,
       return;
    }
       break;
-   case TGSI_OPCODE_ENDLOOP2: {
+   case TGSI_OPCODE_ENDLOOP: {
       instr->endLoop();
       storage->setCurrentBlock(instr->currentBlock());
       return;
@@ -617,14 +611,6 @@ translate_instruction(llvm::Module *module,
       break;
    case TGSI_OPCODE_NOP:
       break;
-   case TGSI_OPCODE_M4X3:
-      break;
-   case TGSI_OPCODE_M3X4:
-      break;
-   case TGSI_OPCODE_M3X3:
-      break;
-   case TGSI_OPCODE_M3X2:
-      break;
    case TGSI_OPCODE_CALLNZ:
       break;
    case TGSI_OPCODE_IFC:
@@ -778,44 +764,38 @@ translate_instructionir(llvm::Module *module,
       out = instr->sub(inputs[0], inputs[1]);
    }
       break;
-   case TGSI_OPCODE_LERP: {
+   case TGSI_OPCODE_LRP: {
    }
       break;
    case TGSI_OPCODE_CND:
       break;
    case TGSI_OPCODE_CND0:
       break;
-   case TGSI_OPCODE_DOT2ADD:
+   case TGSI_OPCODE_DP2A:
       break;
-   case TGSI_OPCODE_INDEX:
-      break;
-   case TGSI_OPCODE_NEGATE:
-      break;
-   case TGSI_OPCODE_FRAC: {
+   case TGSI_OPCODE_FRC: {
    }
       break;
    case TGSI_OPCODE_CLAMP:
       break;
-   case TGSI_OPCODE_FLOOR: {
+   case TGSI_OPCODE_FLR: {
    }
       break;
    case TGSI_OPCODE_ROUND:
       break;
-   case TGSI_OPCODE_EXPBASE2: {
+   case TGSI_OPCODE_EX2: {
    }
       break;
-   case TGSI_OPCODE_LOGBASE2: {
+   case TGSI_OPCODE_LG2: {
    }
       break;
-   case TGSI_OPCODE_POWER: {
+   case TGSI_OPCODE_POW: {
       out = instr->pow(inputs[0], inputs[1]);
    }
       break;
-   case TGSI_OPCODE_CROSSPRODUCT: {
+   case TGSI_OPCODE_XPD: {
    }
       break;
-   case TGSI_OPCODE_MULTIPLYMATRIX:
-      break;
    case TGSI_OPCODE_ABS: {
       out = instr->abs(inputs[0]);
    }
@@ -910,7 +890,7 @@ translate_instructionir(llvm::Module *module,
    case TGSI_OPCODE_IF: {
    }
       break;
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       break;
    case TGSI_OPCODE_REP:
       break;
@@ -920,7 +900,7 @@ translate_instructionir(llvm::Module *module,
    case TGSI_OPCODE_ENDIF: {
    }
       break;
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       break;
    case TGSI_OPCODE_ENDREP:
       break;
@@ -961,13 +941,13 @@ translate_instructionir(llvm::Module *module,
       break;
    case TGSI_OPCODE_ENDPRIM:
       break;
-   case TGSI_OPCODE_BGNLOOP2: {
+   case TGSI_OPCODE_BGNLOOP: {
    }
       break;
    case TGSI_OPCODE_BGNSUB: {
    }
       break;
-   case TGSI_OPCODE_ENDLOOP2: {
+   case TGSI_OPCODE_ENDLOOP: {
    }
       break;
    case TGSI_OPCODE_ENDSUB: {
@@ -983,14 +963,6 @@ translate_instructionir(llvm::Module *module,
       break;
    case TGSI_OPCODE_NOP:
       break;
-   case TGSI_OPCODE_M4X3:
-      break;
-   case TGSI_OPCODE_M3X4:
-      break;
-   case TGSI_OPCODE_M3X3:
-      break;
-   case TGSI_OPCODE_M3X2:
-      break;
    case TGSI_OPCODE_NRM4:
       break;
    case TGSI_OPCODE_CALLNZ:
diff --git a/src/gallium/auxiliary/tgsi/Makefile b/src/gallium/auxiliary/tgsi/Makefile
index b4900e8dbaa..5f0a580b096 100644
--- a/src/gallium/auxiliary/tgsi/Makefile
+++ b/src/gallium/auxiliary/tgsi/Makefile
@@ -16,6 +16,7 @@ C_SOURCES = \
 	tgsi_sse2.c \
 	tgsi_text.c \
 	tgsi_transform.c \
+	tgsi_ureg.c \
 	tgsi_util.c
 
 include ../../Makefile.template
diff --git a/src/gallium/auxiliary/tgsi/SConscript b/src/gallium/auxiliary/tgsi/SConscript
index 8200cce42f5..b6bc2924f06 100644
--- a/src/gallium/auxiliary/tgsi/SConscript
+++ b/src/gallium/auxiliary/tgsi/SConscript
@@ -16,6 +16,7 @@ tgsi = env.ConvenienceLibrary(
 		'tgsi_sse2.c',
 		'tgsi_text.c',
 		'tgsi_transform.c',
+		'tgsi_ureg.c',
 		'tgsi_util.c',
 	])
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
index a3f4947c734..802ec371189 100644
--- a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
+++ b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
@@ -665,9 +665,18 @@ TGSI Instruction Specification
   TBD
 
 
-1.9.8  LOOP - Loop
+1.9.8  BGNFOR - Begin a For-Loop
 
-  TBD
+  dst.x = floor(src.x)
+  dst.y = floor(src.y)
+  dst.z = floor(src.z)
+
+  if (dst.y <= 0)
+    pc = [matching ENDFOR] + 1
+  endif
+
+  Note: The destination must be a loop register.
+        The source must be a constant register.
 
 
 1.9.9  REP - Repeat
@@ -685,9 +694,16 @@ TGSI Instruction Specification
   TBD
 
 
-1.9.12  ENDLOOP - End Loop
+1.9.12  ENDFOR - End a For-Loop
 
-  TBD
+  dst.x = dst.x + dst.z
+  dst.y = dst.y - 1.0
+
+  if (dst.y > 0)
+    pc = [matching BGNFOR instruction] + 1
+  endif
+
+  Note: The destination must be a loop register.
 
 
 1.9.13  ENDREP - End Repeat
@@ -840,7 +856,7 @@ TGSI Instruction Specification
 ----------
 
 
-1.13.1  BGNLOOP2 - Begin Loop
+1.13.1  BGNLOOP - Begin a Loop
 
   TBD
 
@@ -850,7 +866,7 @@ TGSI Instruction Specification
   TBD
 
 
-1.13.3  ENDLOOP2 - End Loop
+1.13.3  ENDLOOP - End a Loop
 
   TBD
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c
index d272533d63c..010d501c601 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.c
@@ -335,7 +335,10 @@ tgsi_default_full_immediate( void )
    struct tgsi_full_immediate fullimm;
 
    fullimm.Immediate = tgsi_default_immediate();
-   fullimm.u.Pointer = (void *) 0;
+   fullimm.u[0].Float = 0.0f;
+   fullimm.u[1].Float = 0.0f;
+   fullimm.u[2].Float = 0.0f;
+   fullimm.u[3].Float = 0.0f;
 
    return fullimm;
 }
@@ -352,19 +355,19 @@ immediate_grow(
    header_bodysize_grow( header );
 }
 
-struct tgsi_immediate_float32
+union tgsi_immediate_data
 tgsi_build_immediate_float32(
    float value,
    struct tgsi_immediate *immediate,
    struct tgsi_header *header )
 {
-   struct tgsi_immediate_float32 immediate_float32;
+   union tgsi_immediate_data immediate_data;
 
-   immediate_float32.Float = value;
+   immediate_data.Float = value;
 
    immediate_grow( immediate, header );
 
-   return immediate_float32;
+   return immediate_data;
 }
 
 unsigned
@@ -384,16 +387,18 @@ tgsi_build_full_immediate(
 
    *immediate = tgsi_build_immediate( header );
 
+   assert( full_imm->Immediate.NrTokens <= 4 + 1 );
+
    for( i = 0; i < full_imm->Immediate.NrTokens - 1; i++ ) {
-      struct tgsi_immediate_float32 *if32;
+      union tgsi_immediate_data *data;
 
       if( maxsize <= size )
          return  0;
-      if32 = (struct tgsi_immediate_float32 *) &tokens[size];
+      data = (union tgsi_immediate_data *) &tokens[size];
       size++;
 
-      *if32 = tgsi_build_immediate_float32(
-         full_imm->u.ImmediateFloat32[i].Float,
+      *data = tgsi_build_immediate_float32(
+         full_imm->u[i].Float,
          immediate,
          header );
    }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.h b/src/gallium/auxiliary/tgsi/tgsi_build.h
index 9a3a077cf2b..17d977b0597 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.h
@@ -119,7 +119,7 @@ tgsi_build_immediate(
 struct tgsi_full_immediate
 tgsi_default_full_immediate( void );
 
-struct tgsi_immediate_float32
+union tgsi_immediate_data
 tgsi_build_immediate_float32(
    float value,
    struct tgsi_immediate *immediate,
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index a6994ecd48b..f36b1114a95 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -82,7 +82,7 @@ static const char *processor_type_names[] =
    "GEOM"
 };
 
-static const char *file_names[] =
+static const char *file_names[TGSI_FILE_COUNT] =
 {
    "NULL",
    "CONST",
@@ -91,7 +91,8 @@ static const char *file_names[] =
    "TEMP",
    "SAMP",
    "ADDR",
-   "IMM"
+   "IMM",
+   "LOOP"
 };
 
 static const char *interpolate_names[] =
@@ -295,10 +296,12 @@ iter_immediate(
    ENM( imm->Immediate.DataType, immediate_type_names );
 
    TXT( " { " );
+
+   assert( imm->Immediate.NrTokens <= 4 + 1 );
    for (i = 0; i < imm->Immediate.NrTokens - 1; i++) {
       switch (imm->Immediate.DataType) {
       case TGSI_IMM_FLOAT32:
-         FLT( imm->u.ImmediateFloat32[i].Float );
+         FLT( imm->u[i].Float );
          break;
       default:
          assert( 0 );
@@ -470,8 +473,8 @@ iter_instruction(
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_IF:
    case TGSI_OPCODE_ELSE:
-   case TGSI_OPCODE_BGNLOOP2:
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
+   case TGSI_OPCODE_ENDLOOP:
    case TGSI_OPCODE_CAL:
       TXT( " :" );
       UID( inst->InstructionExtLabel.Label );
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
index 3dc61c48ca3..4a9c02b1413 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
@@ -69,7 +69,7 @@ static const char *TGSI_TOKEN_TYPES[] =
    "TOKEN_TYPE_INSTRUCTION"
 };
 
-static const char *TGSI_FILES[] =
+static const char *TGSI_FILES[TGSI_FILE_COUNT] =
 {
    "FILE_NULL",
    "FILE_CONSTANT",
@@ -78,7 +78,8 @@ static const char *TGSI_FILES[] =
    "FILE_TEMPORARY",
    "FILE_SAMPLER",
    "FILE_ADDRESS",
-   "FILE_IMMEDIATE"
+   "FILE_IMMEDIATE",
+   "FILE_LOOP"
 };
 
 static const char *TGSI_INTERPOLATES[] =
@@ -283,12 +284,13 @@ dump_immediate_verbose(
       UIX( imm->Immediate.Padding );
    }
 
+   assert( imm->Immediate.NrTokens <= 4 + 1 );
    for( i = 0; i < imm->Immediate.NrTokens - 1; i++ ) {
       EOL();
       switch( imm->Immediate.DataType ) {
       case TGSI_IMM_FLOAT32:
          TXT( "\nFloat: " );
-         FLT( imm->u.ImmediateFloat32[i].Float );
+         FLT( imm->u[i].Float );
          break;
 
       default:
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index fe571a86bca..951ecfd5528 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -301,14 +301,14 @@ tgsi_exec_machine_bind_shader(
       case TGSI_TOKEN_TYPE_IMMEDIATE:
          {
             uint size = parse.FullToken.FullImmediate.Immediate.NrTokens - 1;
-            assert( size % 4 == 0 );
-            assert( mach->ImmLimit + size / 4 <= TGSI_EXEC_NUM_IMMEDIATES );
+            assert( size <= 4 );
+            assert( mach->ImmLimit + 1 <= TGSI_EXEC_NUM_IMMEDIATES );
 
             for( i = 0; i < size; i++ ) {
-               mach->Imms[mach->ImmLimit + i / 4][i % 4] = 
-		  parse.FullToken.FullImmediate.u.ImmediateFloat32[i].Float;
+               mach->Imms[mach->ImmLimit][i] = 
+		  parse.FullToken.FullImmediate.u[i].Float;
             }
-            mach->ImmLimit += size / 4;
+            mach->ImmLimit += 1;
          }
          break;
 
@@ -375,15 +375,9 @@ tgsi_exec_machine_create( void )
    if (!mach)
       goto fail;
 
-   mach->Addrs = &mach->Temps[TGSI_EXEC_TEMP_ADDR];
+   memset(mach, 0, sizeof(*mach));
 
-   mach->Samplers = NULL;
-   mach->Consts = NULL;
-   mach->Tokens = NULL;
-   mach->Primitives = NULL;
-   mach->InterpCoefs = NULL;
-   mach->Instructions = NULL;
-   mach->Declarations = NULL;
+   mach->Addrs = &mach->Temps[TGSI_EXEC_TEMP_ADDR];
 
    /* Setup constants. */
    for( i = 0; i < 4; i++ ) {
@@ -2020,8 +2014,7 @@ exec_instruction(
 
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_ARL:
-   case TGSI_OPCODE_FLOOR:
-   /* TGSI_OPCODE_FLR */
+   case TGSI_OPCODE_FLR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          micro_flr( &r[0], &r[0] );
@@ -2290,8 +2283,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LERP:
-   /* TGSI_OPCODE_LRP */
+   case TGSI_OPCODE_LRP:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
@@ -2325,8 +2317,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_DOT2ADD:
-   /* TGSI_OPCODE_DP2A */
+   case TGSI_OPCODE_DP2A:
       FETCH( &r[0], 0, CHAN_X );
       FETCH( &r[1], 1, CHAN_X );
       micro_mul( &r[0], &r[0], &r[1] );
@@ -2344,18 +2335,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_INDEX:
-      /* XXX: considered for removal */
-      assert (0);
-      break;
-
-   case TGSI_OPCODE_NEGATE:
-      /* XXX: considered for removal */
-      assert (0);
-      break;
-
-   case TGSI_OPCODE_FRAC:
-   /* TGSI_OPCODE_FRC */
+   case TGSI_OPCODE_FRC:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          micro_frc( &r[0], &r[0] );
@@ -2383,8 +2363,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_EXPBASE2:
-   /* TGSI_OPCODE_EX2 */
+   case TGSI_OPCODE_EX2:
       FETCH(&r[0], 0, CHAN_X);
 
 #if FAST_MATH
@@ -2398,8 +2377,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LOGBASE2:
-   /* TGSI_OPCODE_LG2 */
+   case TGSI_OPCODE_LG2:
       FETCH( &r[0], 0, CHAN_X );
       micro_lg2( &r[0], &r[0] );
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -2407,8 +2385,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_POWER:
-   /* TGSI_OPCODE_POW */
+   case TGSI_OPCODE_POW:
       FETCH(&r[0], 0, CHAN_X);
       FETCH(&r[1], 1, CHAN_X);
 
@@ -2419,8 +2396,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_CROSSPRODUCT:
-   /* TGSI_OPCODE_XPD */
+   case TGSI_OPCODE_XPD:
       FETCH(&r[0], 0, CHAN_Y);
       FETCH(&r[1], 1, CHAN_Z);
 
@@ -2462,11 +2438,6 @@ exec_instruction(
       }
       break;
 
-    case TGSI_OPCODE_MULTIPLYMATRIX:
-       /* XXX: considered for removal */
-       assert (0);
-       break;
-
     case TGSI_OPCODE_ABS:
        FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
           FETCH(&r[0], 0, chan_index);
@@ -3110,9 +3081,9 @@ exec_instruction(
       mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       /* fall-through (for now) */
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
       assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
@@ -3120,9 +3091,9 @@ exec_instruction(
       mach->ContStack[mach->ContStackTop++] = mach->ContMask;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       /* fall-through (for now at least) */
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       /* Restore ContMask, but don't pop */
       assert(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c
index 37f2b66d1f6..ccf4b205ffb 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_info.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_info.c
@@ -26,136 +26,156 @@
  **************************************************************************/
 
 #include "util/u_debug.h"
+#include "util/u_memory.h"
 #include "tgsi_info.h"
 
 static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
 {
-   { 1, 1, 0, 0, "ARL", NULL, NULL },
-   { 1, 1, 0, 0, "MOV", NULL, NULL },
-   { 1, 1, 0, 0, "LIT", NULL, NULL },
-   { 1, 1, 0, 0, "RCP", "RECIP", NULL },
-   { 1, 1, 0, 0, "RSQ", "RECIPSQRT", NULL },
-   { 1, 1, 0, 0, "EXP", "EXPP", NULL },
-   { 1, 1, 0, 0, "LOG", NULL, NULL },
-   { 1, 2, 0, 0, "MUL", NULL, NULL },
-   { 1, 2, 0, 0, "ADD", NULL, NULL },
-   { 1, 2, 0, 0, "DP3", "DOT3", NULL },
-   { 1, 2, 0, 0, "DP4", "DOT4", NULL },
-   { 1, 2, 0, 0, "DST", NULL, NULL },
-   { 1, 2, 0, 0, "MIN", NULL, NULL },
-   { 1, 2, 0, 0, "MAX", NULL, NULL },
-   { 1, 2, 0, 0, "SLT", "SETLT", NULL },
-   { 1, 2, 0, 0, "SGE", "SETGE", NULL },
-   { 1, 3, 0, 0, "MAD", "MADD", NULL },
-   { 1, 2, 0, 0, "SUB", NULL, NULL },
-   { 1, 3, 0, 0, "LRP", "LERP", NULL },
-   { 1, 3, 0, 0, "CND", NULL, NULL },
-   { 1, 3, 0, 0, "CND0", NULL, NULL },
-   { 1, 3, 0, 0, "DP2A", "DP2ADD", "DOT2ADD" },
-   { 1, 2, 0, 0, "INDEX", NULL, NULL },
-   { 1, 1, 0, 0, "NEGATE", NULL, NULL },
-   { 1, 1, 0, 0, "FRC", "FRAC", NULL },
-   { 1, 3, 0, 0, "CLAMP", NULL, NULL },
-   { 1, 1, 0, 0, "FLR", "FLOOR", NULL },
-   { 1, 1, 0, 0, "ROUND", NULL, NULL },
-   { 1, 1, 0, 0, "EX2", "EXPBASE2", NULL },
-   { 1, 1, 0, 0, "LG2", "LOGBASE2", "LOGP" },
-   { 1, 2, 0, 0, "POW", "POWER", NULL },
-   { 1, 2, 0, 0, "XPD", "CRS", "CROSSPRODUCT" },
-   { 1, 2, 0, 0, "M4X4", "MULTIPLYMATRIX", NULL },
-   { 1, 1, 0, 0, "ABS", NULL, NULL },
-   { 1, 1, 0, 0, "RCC", NULL, NULL },
-   { 1, 2, 0, 0, "DPH", NULL, NULL },
-   { 1, 1, 0, 0, "COS", NULL, NULL },
-   { 1, 1, 0, 0, "DDX", "DSX", NULL },
-   { 1, 1, 0, 0, "DDY", "DSY", NULL },
-   { 0, 0, 0, 0, "KILP", NULL, NULL },
-   { 1, 1, 0, 0, "PK2H", NULL, NULL },
-   { 1, 1, 0, 0, "PK2US", NULL, NULL },
-   { 1, 1, 0, 0, "PK4B", NULL, NULL },
-   { 1, 1, 0, 0, "PK4UB", NULL, NULL },
-   { 1, 2, 0, 0, "RFL", NULL, NULL },
-   { 1, 2, 0, 0, "SEQ", NULL, NULL },
-   { 1, 2, 0, 0, "SFL", NULL, NULL },
-   { 1, 2, 0, 0, "SGT", NULL, NULL },
-   { 1, 1, 0, 0, "SIN", NULL, NULL },
-   { 1, 2, 0, 0, "SLE", NULL, NULL },
-   { 1, 2, 0, 0, "SNE", NULL, NULL },
-   { 1, 2, 0, 0, "STR", NULL, NULL },
-   { 1, 2, 1, 0, "TEX", "TEXLD", NULL },
-   { 1, 4, 1, 0, "TXD", "TEXLDD", NULL },
-   { 1, 2, 1, 0, "TXP", NULL, NULL },
-   { 1, 1, 0, 0, "UP2H", NULL, NULL },
-   { 1, 1, 0, 0, "UP2US", NULL, NULL },
-   { 1, 1, 0, 0, "UP4B", NULL, NULL },
-   { 1, 1, 0, 0, "UP4UB", NULL, NULL },
-   { 1, 3, 0, 0, "X2D", NULL, NULL },
-   { 1, 1, 0, 0, "ARA", NULL, NULL },
-   { 1, 1, 0, 0, "ARR", "MOVA", NULL },
-   { 0, 1, 0, 0, "BRA", NULL, NULL },
-   { 0, 0, 0, 1, "CAL", "CALL", NULL },
-   { 0, 0, 0, 0, "RET", NULL, NULL },
-   { 1, 1, 0, 0, "SGN", "SSG", NULL },
-   { 1, 3, 0, 0, "CMP", NULL, NULL },
-   { 1, 1, 0, 0, "SCS", "SINCOS", NULL },
-   { 1, 2, 1, 0, "TXB", "TEXLDB", NULL },
-   { 1, 1, 0, 0, "NRM", NULL, NULL },
-   { 1, 2, 0, 0, "DIV", NULL, NULL },
-   { 1, 2, 0, 0, "DP2", NULL, NULL },
-   { 1, 2, 1, 0, "TXL", NULL, NULL },
-   { 0, 0, 0, 0, "BRK", "BREAK", NULL },
-   { 0, 1, 0, 1, "IF", NULL, NULL },
-   { 0, 0, 0, 0, "LOOP", NULL, NULL },
-   { 0, 1, 0, 0, "REP", NULL, NULL },
-   { 0, 0, 0, 1, "ELSE", NULL, NULL },
-   { 0, 0, 0, 0, "ENDIF", NULL, NULL },
-   { 0, 0, 0, 0, "ENDLOOP", NULL, NULL },
-   { 0, 0, 0, 0, "ENDREP", NULL, NULL },
-   { 0, 1, 0, 0, "PUSHA", NULL, NULL },
-   { 1, 0, 0, 0, "POPA", NULL, NULL },
-   { 1, 1, 0, 0, "CEIL", NULL, NULL },
-   { 1, 1, 0, 0, "I2F", NULL, NULL },
-   { 1, 1, 0, 0, "NOT", NULL, NULL },
-   { 1, 1, 0, 0, "INT", "TRUNC", NULL },
-   { 1, 2, 0, 0, "SHL", NULL, NULL },
-   { 1, 2, 0, 0, "SHR", NULL, NULL },
-   { 1, 2, 0, 0, "AND", NULL, NULL },
-   { 1, 2, 0, 0, "OR", NULL, NULL },
-   { 1, 2, 0, 0, "MOD", NULL, NULL },
-   { 1, 2, 0, 0, "XOR", NULL, NULL },
-   { 1, 3, 0, 0, "SAD", NULL, NULL },
-   { 1, 2, 1, 0, "TXF", NULL, NULL },
-   { 1, 2, 1, 0, "TXQ", NULL, NULL },
-   { 0, 0, 0, 0, "CONT", NULL, NULL },
-   { 0, 0, 0, 0, "EMIT", NULL, NULL },
-   { 0, 0, 0, 0, "ENDPRIM", NULL, NULL },
-   { 0, 0, 0, 1, "BGNLOOP2", NULL, NULL },
-   { 0, 0, 0, 0, "BGNSUB", NULL, NULL },
-   { 0, 0, 0, 1, "ENDLOOP2", NULL, NULL },
-   { 0, 0, 0, 0, "ENDSUB", NULL, NULL },
-   { 1, 1, 0, 0, "NOISE1", NULL, NULL },
-   { 1, 1, 0, 0, "NOISE2", NULL, NULL },
-   { 1, 1, 0, 0, "NOISE3", NULL, NULL },
-   { 1, 1, 0, 0, "NOISE4", NULL, NULL },
-   { 0, 0, 0, 0, "NOP", NULL, NULL },
-   { 1, 2, 0, 0, "M4X3", NULL, NULL },
-   { 1, 2, 0, 0, "M3X4", NULL, NULL },
-   { 1, 2, 0, 0, "M3X3", NULL, NULL },
-   { 1, 2, 0, 0, "M3X2", NULL, NULL },
-   { 1, 1, 0, 0, "NRM4", NULL, NULL },
-   { 0, 1, 0, 0, "CALLNZ", NULL, NULL },
-   { 0, 1, 0, 0, "IFC", NULL, NULL },
-   { 0, 1, 0, 0, "BREAKC", NULL, NULL },
-   { 0, 1, 0, 0, "KIL", "TEXKILL", NULL },
-   { 0, 0, 0, 0, "END", NULL, NULL },
-   { 1, 1, 0, 0, "SWZ", NULL, NULL }
+   { 1, 1, 0, 0, "ARL", TGSI_OPCODE_ARL },
+   { 1, 1, 0, 0, "MOV", TGSI_OPCODE_MOV },
+   { 1, 1, 0, 0, "LIT", TGSI_OPCODE_LIT },
+   { 1, 1, 0, 0, "RCP", TGSI_OPCODE_RCP },
+   { 1, 1, 0, 0, "RSQ", TGSI_OPCODE_RSQ },
+   { 1, 1, 0, 0, "EXP", TGSI_OPCODE_EXP },
+   { 1, 1, 0, 0, "LOG", TGSI_OPCODE_LOG },
+   { 1, 2, 0, 0, "MUL", TGSI_OPCODE_MUL },
+   { 1, 2, 0, 0, "ADD", TGSI_OPCODE_ADD },
+   { 1, 2, 0, 0, "DP3", TGSI_OPCODE_DP3 },
+   { 1, 2, 0, 0, "DP4", TGSI_OPCODE_DP4 },
+   { 1, 2, 0, 0, "DST", TGSI_OPCODE_DST },
+   { 1, 2, 0, 0, "MIN", TGSI_OPCODE_MIN },
+   { 1, 2, 0, 0, "MAX", TGSI_OPCODE_MAX },
+   { 1, 2, 0, 0, "SLT", TGSI_OPCODE_SLT },
+   { 1, 2, 0, 0, "SGE", TGSI_OPCODE_SGE },
+   { 1, 3, 0, 0, "MAD", TGSI_OPCODE_MAD },
+   { 1, 2, 0, 0, "SUB", TGSI_OPCODE_SUB },
+   { 1, 3, 0, 0, "LRP", TGSI_OPCODE_LRP },
+   { 1, 3, 0, 0, "CND", TGSI_OPCODE_CND },
+   { 1, 3, 0, 0, "CND0", TGSI_OPCODE_CND0 },
+   { 1, 3, 0, 0, "DP2A", TGSI_OPCODE_DP2A },
+   { 0, 0, 0, 0, "", 22 },      /* removed */
+   { 0, 0, 0, 0, "", 23 },      /* removed */
+   { 1, 1, 0, 0, "FRC", TGSI_OPCODE_FRC },
+   { 1, 3, 0, 0, "CLAMP", TGSI_OPCODE_CLAMP },
+   { 1, 1, 0, 0, "FLR", TGSI_OPCODE_FLR },
+   { 1, 1, 0, 0, "ROUND", TGSI_OPCODE_ROUND },
+   { 1, 1, 0, 0, "EX2", TGSI_OPCODE_EX2 },
+   { 1, 1, 0, 0, "LG2", TGSI_OPCODE_LG2 },
+   { 1, 2, 0, 0, "POW", TGSI_OPCODE_POW },
+   { 1, 2, 0, 0, "XPD", TGSI_OPCODE_XPD },
+   { 0, 0, 0, 0, "", 32 },      /* removed */
+   { 1, 1, 0, 0, "ABS", TGSI_OPCODE_ABS },
+   { 1, 1, 0, 0, "RCC", TGSI_OPCODE_RCC },
+   { 1, 2, 0, 0, "DPH", TGSI_OPCODE_DPH },
+   { 1, 1, 0, 0, "COS", TGSI_OPCODE_COS },
+   { 1, 1, 0, 0, "DDX", TGSI_OPCODE_DDX },
+   { 1, 1, 0, 0, "DDY", TGSI_OPCODE_DDY },
+   { 0, 0, 0, 0, "KILP", TGSI_OPCODE_KILP },
+   { 1, 1, 0, 0, "PK2H", TGSI_OPCODE_PK2H },
+   { 1, 1, 0, 0, "PK2US", TGSI_OPCODE_PK2US },
+   { 1, 1, 0, 0, "PK4B", TGSI_OPCODE_PK4B },
+   { 1, 1, 0, 0, "PK4UB", TGSI_OPCODE_PK4UB },
+   { 1, 2, 0, 0, "RFL", TGSI_OPCODE_RFL },
+   { 1, 2, 0, 0, "SEQ", TGSI_OPCODE_SEQ },
+   { 1, 2, 0, 0, "SFL", TGSI_OPCODE_SFL },
+   { 1, 2, 0, 0, "SGT", TGSI_OPCODE_SGT },
+   { 1, 1, 0, 0, "SIN", TGSI_OPCODE_SIN },
+   { 1, 2, 0, 0, "SLE", TGSI_OPCODE_SLE },
+   { 1, 2, 0, 0, "SNE", TGSI_OPCODE_SNE },
+   { 1, 2, 0, 0, "STR", TGSI_OPCODE_STR },
+   { 1, 2, 1, 0, "TEX", TGSI_OPCODE_TEX },
+   { 1, 4, 1, 0, "TXD", TGSI_OPCODE_TXD },
+   { 1, 2, 1, 0, "TXP", TGSI_OPCODE_TXP },
+   { 1, 1, 0, 0, "UP2H", TGSI_OPCODE_UP2H },
+   { 1, 1, 0, 0, "UP2US", TGSI_OPCODE_UP2US },
+   { 1, 1, 0, 0, "UP4B", TGSI_OPCODE_UP4B },
+   { 1, 1, 0, 0, "UP4UB", TGSI_OPCODE_UP4UB },
+   { 1, 3, 0, 0, "X2D", TGSI_OPCODE_X2D },
+   { 1, 1, 0, 0, "ARA", TGSI_OPCODE_ARA },
+   { 1, 1, 0, 0, "ARR", TGSI_OPCODE_ARR },
+   { 0, 1, 0, 0, "BRA", TGSI_OPCODE_BRA },
+   { 0, 0, 0, 1, "CAL", TGSI_OPCODE_CAL },
+   { 0, 0, 0, 0, "RET", TGSI_OPCODE_RET },
+   { 1, 1, 0, 0, "SSG", TGSI_OPCODE_SSG },
+   { 1, 3, 0, 0, "CMP", TGSI_OPCODE_CMP },
+   { 1, 1, 0, 0, "SCS", TGSI_OPCODE_SCS },
+   { 1, 2, 1, 0, "TXB", TGSI_OPCODE_TXB },
+   { 1, 1, 0, 0, "NRM", TGSI_OPCODE_NRM },
+   { 1, 2, 0, 0, "DIV", TGSI_OPCODE_DIV },
+   { 1, 2, 0, 0, "DP2", TGSI_OPCODE_DP2 },
+   { 1, 2, 1, 0, "TXL", TGSI_OPCODE_TXL },
+   { 0, 0, 0, 0, "BRK", TGSI_OPCODE_BRK },
+   { 0, 1, 0, 1, "IF", TGSI_OPCODE_IF },
+   { 1, 1, 0, 0, "BGNFOR", TGSI_OPCODE_BGNFOR },
+   { 0, 1, 0, 0, "REP", TGSI_OPCODE_REP },
+   { 0, 0, 0, 1, "ELSE", TGSI_OPCODE_ELSE },
+   { 0, 0, 0, 0, "ENDIF", TGSI_OPCODE_ENDIF },
+   { 1, 0, 0, 0, "ENDFOR", TGSI_OPCODE_ENDFOR },
+   { 0, 0, 0, 0, "ENDREP", TGSI_OPCODE_ENDREP },
+   { 0, 1, 0, 0, "PUSHA", TGSI_OPCODE_PUSHA },
+   { 1, 0, 0, 0, "POPA", TGSI_OPCODE_POPA },
+   { 1, 1, 0, 0, "CEIL", TGSI_OPCODE_CEIL },
+   { 1, 1, 0, 0, "I2F", TGSI_OPCODE_I2F },
+   { 1, 1, 0, 0, "NOT", TGSI_OPCODE_NOT },
+   { 1, 1, 0, 0, "TRUNC", TGSI_OPCODE_TRUNC },
+   { 1, 2, 0, 0, "SHL", TGSI_OPCODE_SHL },
+   { 1, 2, 0, 0, "SHR", TGSI_OPCODE_SHR },
+   { 1, 2, 0, 0, "AND", TGSI_OPCODE_AND },
+   { 1, 2, 0, 0, "OR", TGSI_OPCODE_OR },
+   { 1, 2, 0, 0, "MOD", TGSI_OPCODE_MOD },
+   { 1, 2, 0, 0, "XOR", TGSI_OPCODE_XOR },
+   { 1, 3, 0, 0, "SAD", TGSI_OPCODE_SAD },
+   { 1, 2, 1, 0, "TXF", TGSI_OPCODE_TXF },
+   { 1, 2, 1, 0, "TXQ", TGSI_OPCODE_TXQ },
+   { 0, 0, 0, 0, "CONT", TGSI_OPCODE_CONT },
+   { 0, 0, 0, 0, "EMIT", TGSI_OPCODE_EMIT },
+   { 0, 0, 0, 0, "ENDPRIM", TGSI_OPCODE_ENDPRIM },
+   { 0, 0, 0, 1, "BGNLOOP", TGSI_OPCODE_BGNLOOP },
+   { 0, 0, 0, 0, "BGNSUB", TGSI_OPCODE_BGNSUB },
+   { 0, 0, 0, 1, "ENDLOOP", TGSI_OPCODE_ENDLOOP },
+   { 0, 0, 0, 0, "ENDSUB", TGSI_OPCODE_ENDSUB },
+   { 1, 1, 0, 0, "NOISE1", TGSI_OPCODE_NOISE1 },
+   { 1, 1, 0, 0, "NOISE2", TGSI_OPCODE_NOISE2 },
+   { 1, 1, 0, 0, "NOISE3", TGSI_OPCODE_NOISE3 },
+   { 1, 1, 0, 0, "NOISE4", TGSI_OPCODE_NOISE4 },
+   { 0, 0, 0, 0, "NOP", TGSI_OPCODE_NOP },
+   { 0, 0, 0, 0, "", 108 },     /* removed */
+   { 0, 0, 0, 0, "", 109 },     /* removed */
+   { 0, 0, 0, 0, "", 110 },     /* removed */
+   { 0, 0, 0, 0, "", 111 },     /* removed */
+   { 1, 1, 0, 0, "NRM4", TGSI_OPCODE_NRM4 },
+   { 0, 1, 0, 0, "CALLNZ", TGSI_OPCODE_CALLNZ },
+   { 0, 1, 0, 0, "IFC", TGSI_OPCODE_IFC },
+   { 0, 1, 0, 0, "BREAKC", TGSI_OPCODE_BREAKC },
+   { 0, 1, 0, 0, "KIL", TGSI_OPCODE_KIL },
+   { 0, 0, 0, 0, "END", TGSI_OPCODE_END },
+   { 1, 1, 0, 0, "SWZ", TGSI_OPCODE_SWZ }
 };
 
 const struct tgsi_opcode_info *
 tgsi_get_opcode_info( uint opcode )
 {
+   static boolean firsttime = 1;
+
+   if (firsttime) {
+      unsigned i;
+      firsttime = 0;
+      for (i = 0; i < Elements(opcode_info); i++)
+         assert(opcode_info[i].opcode == i);
+   }
+   
    if (opcode < TGSI_OPCODE_LAST)
       return &opcode_info[opcode];
+
    assert( 0 );
    return NULL;
 }
+
+
+const char *
+tgsi_get_opcode_name( uint opcode )
+{
+   const struct tgsi_opcode_info *info = tgsi_get_opcode_info(opcode);
+   return info->mnemonic;
+}
+
diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.h b/src/gallium/auxiliary/tgsi/tgsi_info.h
index 077e25acd7f..b2375c69710 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_info.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_info.h
@@ -41,13 +41,16 @@ struct tgsi_opcode_info
    boolean is_tex;
    boolean is_branch;
    const char *mnemonic;
-   const char *alt_mnemonic1;
-   const char *alt_mnemonic2;
+   uint opcode;
 };
 
 const struct tgsi_opcode_info *
 tgsi_get_opcode_info( uint opcode );
 
+const char *
+tgsi_get_opcode_name( uint opcode );
+
+
 #if defined __cplusplus
 }
 #endif
diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h
new file mode 100644
index 00000000000..ed594a3e2c7
--- /dev/null
+++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h
@@ -0,0 +1,173 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+#ifndef OP12_TEX
+#define OP12_TEX(a) OP12(a)
+#endif
+
+#ifndef OP14_TEX
+#define OP14_TEX(a) OP14(a)
+#endif
+
+#ifndef OP00_LBL
+#define OP00_LBL(a) OP00(a)
+#endif
+
+#ifndef OP01_LBL
+#define OP01_LBL(a) OP01(a)
+#endif
+
+OP11(ARL)
+OP11(MOV)
+OP11(LIT)
+OP11(RCP)
+OP11(RSQ)
+OP11(EXP)
+OP11(LOG)
+OP12(MUL)
+OP12(ADD)
+OP12(DP3)
+OP12(DP4)
+OP12(DST)
+OP12(MIN)
+OP12(MAX)
+OP12(SLT)
+OP12(SGE)
+OP13(MAD)
+OP12(SUB)
+OP13(LRP)
+OP13(CND)
+OP13(CND0)
+OP13(DP2A)
+OP11(FRC)
+OP13(CLAMP)
+OP11(FLR)
+OP11(ROUND)
+OP11(EX2)
+OP11(LG2)
+OP12(POW)
+OP12(XPD)
+OP11(ABS)
+OP11(RCC)
+OP12(DPH)
+OP11(COS)
+OP11(DDX)
+OP11(DDY)
+OP00(KILP)
+OP11(PK2H)
+OP11(PK2US)
+OP11(PK4B)
+OP11(PK4UB)
+OP12(RFL)
+OP12(SEQ)
+OP12(SFL)
+OP12(SGT)
+OP11(SIN)
+OP12(SLE)
+OP12(SNE)
+OP12(STR)
+OP12_TEX(TEX)
+OP14_TEX(TXD)
+OP12_TEX(TXP)
+OP11(UP2H)
+OP11(UP2US)
+OP11(UP4B)
+OP11(UP4UB)
+OP13(X2D)
+OP11(ARA)
+OP11(ARR)
+OP01(BRA)
+OP00_LBL(CAL)
+OP00(RET)
+OP11(SSG)
+OP13(CMP)
+OP11(SCS)
+OP12_TEX(TXB)
+OP11(NRM)
+OP12(DIV)
+OP12(DP2)
+OP12_TEX(TXL)
+OP00(BRK)
+OP01_LBL(IF)
+OP11(BGNFOR)
+OP01(REP)
+OP00_LBL(ELSE)
+OP00(ENDIF)
+OP10(ENDFOR)
+OP00(ENDREP)
+OP01(PUSHA)
+OP10(POPA)
+OP11(CEIL)
+OP11(I2F)
+OP11(NOT)
+OP11(TRUNC)
+OP12(SHL)
+OP12(SHR)
+OP12(AND)
+OP12(OR)
+OP12(MOD)
+OP12(XOR)
+OP13(SAD)
+OP12_TEX(TXF)
+OP12_TEX(TXQ)
+OP00(CONT)
+OP00(EMIT)
+OP00(ENDPRIM)
+OP00_LBL(BGNLOOP)
+OP00(BGNSUB)
+OP00_LBL(ENDLOOP)
+OP00(ENDSUB)
+OP11(NOISE1)
+OP11(NOISE2)
+OP11(NOISE3)
+OP11(NOISE4)
+OP00(NOP)
+OP11(NRM4)
+OP01(CALLNZ)
+OP01(IFC)
+OP01(BREAKC)
+OP01(KIL)
+OP00(END)
+OP11(SWZ)
+
+
+#undef OP00
+#undef OP01
+#undef OP10
+#undef OP11
+#undef OP12
+#undef OP13
+
+#ifdef OP14
+#undef OP14
+#endif
+
+#undef OP00_LBL
+#undef OP01_LBL
+
+#undef OP12_TEX
+#undef OP14_TEX
+
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.c b/src/gallium/auxiliary/tgsi/tgsi_parse.c
index 7f2cfb7988f..4870f82b6bd 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.c
@@ -42,9 +42,6 @@ void
 tgsi_full_token_free(
    union tgsi_full_token *full_token )
 {
-   if( full_token->Token.Type == TGSI_TOKEN_TYPE_IMMEDIATE ) {
-      FREE( (void *) full_token->FullImmediate.u.Pointer );
-   }
 }
 
 unsigned
@@ -156,14 +153,8 @@ tgsi_parse_token(
       case TGSI_IMM_FLOAT32:
          {
             uint imm_count = imm->Immediate.NrTokens - 1;
-            struct tgsi_immediate_float32 *data;
-
-            data = (struct tgsi_immediate_float32 *) MALLOC(sizeof(struct tgsi_immediate_float32) * imm_count);
-            if (data) {
-               for (i = 0; i < imm_count; i++) {
-                  next_token(ctx, &data[i]);
-               }
-               imm->u.ImmediateFloat32 = data;
+            for (i = 0; i < imm_count; i++) {
+               next_token(ctx, &imm->u[i]);
             }
          }
          break;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.h b/src/gallium/auxiliary/tgsi/tgsi_parse.h
index a289e26e3ac..1035bda1a87 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.h
@@ -73,11 +73,7 @@ struct tgsi_full_declaration
 struct tgsi_full_immediate
 {
    struct tgsi_immediate   Immediate;
-   union
-   {
-      const void                          *Pointer;
-      const struct tgsi_immediate_float32 *ImmediateFloat32;
-   } u;
+   union tgsi_immediate_data u[4];
 };
 
 #define TGSI_FULL_MAX_DST_REGISTERS 2
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ppc.c b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
index 0c64ae57131..2d6ad12ffbd 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ppc.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
@@ -38,6 +38,7 @@
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "util/u_sse.h"
+#include "tgsi/tgsi_info.h"
 #include "tgsi/tgsi_parse.h"
 #include "tgsi/tgsi_util.h"
 #include "tgsi_dump.h"
@@ -619,17 +620,17 @@ emit_unaryop(struct gen_context *gen, struct tgsi_full_instruction *inst)
             ppc_vandc(gen->f, v1, v0, bit31_vec); /* v1 = v0 & ~bit31 */
          }
          break;
-      case TGSI_OPCODE_FLOOR:
+      case TGSI_OPCODE_FLR:
          ppc_vrfim(gen->f, v1, v0);         /* v1 = floor(v0) */
          break;
-      case TGSI_OPCODE_FRAC:
+      case TGSI_OPCODE_FRC:
          ppc_vrfim(gen->f, v1, v0);      /* tmp = floor(v0) */
          ppc_vsubfp(gen->f, v1, v0, v1); /* v1 = v0 - v1 */
          break;
-      case TGSI_OPCODE_EXPBASE2:
+      case TGSI_OPCODE_EX2:
          ppc_vexptefp(gen->f, v1, v0);     /* v1 = 2^v0 */
          break;
-      case TGSI_OPCODE_LOGBASE2:
+      case TGSI_OPCODE_LG2:
          /* XXX this may be broken! */
          ppc_vlogefp(gen->f, v1, v0);      /* v1 = log2(v0) */
          break;
@@ -1111,10 +1112,10 @@ emit_instruction(struct gen_context *gen,
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:
    case TGSI_OPCODE_ABS:
-   case TGSI_OPCODE_FLOOR:
-   case TGSI_OPCODE_FRAC:
-   case TGSI_OPCODE_EXPBASE2:
-   case TGSI_OPCODE_LOGBASE2:
+   case TGSI_OPCODE_FLR:
+   case TGSI_OPCODE_FRC:
+   case TGSI_OPCODE_EX2:
+   case TGSI_OPCODE_LG2:
       emit_unaryop(gen, inst);
       break;
    case TGSI_OPCODE_RSQ:
@@ -1317,8 +1318,10 @@ tgsi_emit_ppc(const struct tgsi_token *tokens,
          ok = emit_instruction(&gen, &parse.FullToken.FullInstruction);
 
 	 if (!ok) {
-	    debug_printf("failed to translate tgsi opcode %d to PPC (%s)\n", 
-			 parse.FullToken.FullInstruction.Instruction.Opcode,
+            uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
+	    debug_printf("failed to translate tgsi opcode %d (%s) to PPC (%s)\n", 
+			 opcode,
+                         tgsi_get_opcode_name(opcode),
                          parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ?
                          "vertex shader" : "fragment shader");
 	 }
@@ -1333,7 +1336,7 @@ tgsi_emit_ppc(const struct tgsi_token *tokens,
             assert(num_immediates < TGSI_EXEC_NUM_IMMEDIATES);
             for (i = 0; i < size; i++) {
                immediates[num_immediates][i] =
-		  parse.FullToken.FullImmediate.u.ImmediateFloat32[i].Float;
+		  parse.FullToken.FullImmediate.u[i].Float;
             }
             num_immediates++;
          }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index 6f1f5c2b4b0..4fe8553c423 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -131,7 +131,7 @@ is_register_used(
    return (ctx->regs_used[file][index / BITS_IN_REG_FLAG] & (1 << (index % BITS_IN_REG_FLAG))) ? TRUE : FALSE;
 }
 
-static const char *file_names[] =
+static const char *file_names[TGSI_FILE_COUNT] =
 {
    "NULL",
    "CONST",
@@ -140,7 +140,8 @@ static const char *file_names[] =
    "TEMP",
    "SAMP",
    "ADDR",
-   "IMM"
+   "IMM",
+   "LOOP"
 };
 
 static boolean
@@ -234,9 +235,29 @@ iter_instruction(
             index,
             "indirect",
             FALSE );
-         if (file != TGSI_FILE_ADDRESS || index != 0)
-            report_warning( ctx, "Indirect register not ADDR[0]" );
+         if (!(file == TGSI_FILE_ADDRESS || file == TGSI_FILE_LOOP) || index != 0) {
+            report_warning(ctx, "Indirect register neither ADDR[0] nor LOOP[0]");
+         }
+      }
+   }
+
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_BGNFOR:
+   case TGSI_OPCODE_ENDFOR:
+      if (inst->FullDstRegisters[0].DstRegister.File != TGSI_FILE_LOOP ||
+          inst->FullDstRegisters[0].DstRegister.Index != 0) {
+         report_error(ctx, "Destination register must be LOOP[0]");
+      }
+      break;
+   }
+
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_BGNFOR:
+      if (inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_CONSTANT &&
+          inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_IMMEDIATE) {
+         report_error(ctx, "Source register file must be either CONST or IMM");
       }
+      break;
    }
 
    ctx->num_instructions++;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 4c3343d26c3..cfec5cfc019 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -36,6 +36,7 @@
 #if defined(PIPE_ARCH_SSE)
 #include "util/u_sse.h"
 #endif
+#include "tgsi/tgsi_info.h"
 #include "tgsi/tgsi_parse.h"
 #include "tgsi/tgsi_util.h"
 #include "tgsi_exec.h"
@@ -1467,15 +1468,15 @@ emit_tex( struct x86_function *func,
 
    switch (inst->InstructionExtTexture.Texture) {
    case TGSI_TEXTURE_1D:
-   case TGSI_TEXTURE_SHADOW1D:
       count = 1;
       break;
    case TGSI_TEXTURE_2D:
    case TGSI_TEXTURE_RECT:
-   case TGSI_TEXTURE_SHADOW2D:
-   case TGSI_TEXTURE_SHADOWRECT:
       count = 2;
       break;
+   case TGSI_TEXTURE_SHADOW1D:
+   case TGSI_TEXTURE_SHADOW2D:
+   case TGSI_TEXTURE_SHADOWRECT:
    case TGSI_TEXTURE_3D:
    case TGSI_TEXTURE_CUBE:
       count = 3;
@@ -2064,8 +2065,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LERP:
-   /* TGSI_OPCODE_LRP */
+   case TGSI_OPCODE_LRP:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
          FETCH( func, *inst, 1, 1, chan_index );
@@ -2085,8 +2085,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_DOT2ADD:
-   /* TGSI_OPCODE_DP2A */
+   case TGSI_OPCODE_DP2A:
       FETCH( func, *inst, 0, 0, CHAN_X );  /* xmm0 = src[0].x */
       FETCH( func, *inst, 1, 1, CHAN_X );  /* xmm1 = src[1].x */
       emit_mul( func, 0, 1 );              /* xmm0 = xmm0 * xmm1 */
@@ -2101,16 +2100,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_INDEX:
-      return 0;
-      break;
-
-   case TGSI_OPCODE_NEGATE:
-      return 0;
-      break;
-
-   case TGSI_OPCODE_FRAC:
-   /* TGSI_OPCODE_FRC */
+   case TGSI_OPCODE_FRC:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
          emit_frc( func, 0, 0 );
@@ -2122,8 +2112,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_FLOOR:
-   /* TGSI_OPCODE_FLR */
+   case TGSI_OPCODE_FLR:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
          emit_flr( func, 0, 0 );
@@ -2139,8 +2128,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_EXPBASE2:
-   /* TGSI_OPCODE_EX2 */
+   case TGSI_OPCODE_EX2:
       FETCH( func, *inst, 0, 0, CHAN_X );
       emit_ex2( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -2148,8 +2136,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LOGBASE2:
-   /* TGSI_OPCODE_LG2 */
+   case TGSI_OPCODE_LG2:
       FETCH( func, *inst, 0, 0, CHAN_X );
       emit_lg2( func, 0, 0 );
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -2157,8 +2144,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_POWER:
-   /* TGSI_OPCODE_POW */
+   case TGSI_OPCODE_POW:
       FETCH( func, *inst, 0, 0, CHAN_X );
       FETCH( func, *inst, 1, 1, CHAN_X );
       emit_pow( func, 0, 0, 0, 1 );
@@ -2167,8 +2153,7 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_CROSSPRODUCT:
-   /* TGSI_OPCODE_XPD */
+   case TGSI_OPCODE_XPD:
       if( IS_DST0_CHANNEL_ENABLED( *inst, CHAN_X ) ||
           IS_DST0_CHANNEL_ENABLED( *inst, CHAN_Y ) ) {
          FETCH( func, *inst, 1, 1, CHAN_Z );
@@ -2214,10 +2199,6 @@ emit_instruction(
       }
       break;
 
-   case TGSI_OPCODE_MULTIPLYMATRIX:
-      return 0;
-      break;
-
    case TGSI_OPCODE_ABS:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( func, *inst, 0, 0, chan_index );
@@ -2551,7 +2532,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       return 0;
       break;
 
@@ -2567,7 +2548,7 @@ emit_instruction(
       return 0;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       return 0;
       break;
 
@@ -2937,8 +2918,10 @@ tgsi_emit_sse2(
             &parse.FullToken.FullInstruction );
 
 	 if (!ok) {
-	    debug_printf("failed to translate tgsi opcode %d to SSE (%s)\n", 
-			 parse.FullToken.FullInstruction.Instruction.Opcode,
+            uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
+	    debug_printf("failed to translate tgsi opcode %d (%s) to SSE (%s)\n", 
+			 opcode,
+                         tgsi_get_opcode_name(opcode),
                          parse.FullHeader.Processor.Processor == TGSI_PROCESSOR_VERTEX ?
                          "vertex shader" : "fragment shader");
 	 }
@@ -2953,7 +2936,7 @@ tgsi_emit_sse2(
             assert(num_immediates < TGSI_EXEC_NUM_IMMEDIATES);
             for( i = 0; i < size; i++ ) {
                immediates[num_immediates][i] =
-		  parse.FullToken.FullImmediate.u.ImmediateFloat32[i].Float;
+		  parse.FullToken.FullImmediate.u[i].Float;
             }
 #if 0
             debug_printf("SSE FS immediate[%d] = %f %f %f %f\n",
diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c b/src/gallium/auxiliary/tgsi/tgsi_text.c
index a76bbc91400..d438450b1e4 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -231,7 +231,8 @@ static const char *file_names[TGSI_FILE_COUNT] =
    "TEMP",
    "SAMP",
    "ADDR",
-   "IMM"
+   "IMM",
+   "LOOP"
 };
 
 static boolean
@@ -789,16 +790,6 @@ match_inst_mnemonic(const char **pcur,
    if (str_match_no_case(pcur, info->mnemonic)) {
       return TRUE;
    }
-   if (info->alt_mnemonic1) {
-      if (str_match_no_case(pcur, info->alt_mnemonic1)) {
-         return TRUE;
-      }
-      if (info->alt_mnemonic2) {
-         if (str_match_no_case(pcur, info->alt_mnemonic2)) {
-            return TRUE;
-         }
-      }
-   }
    return FALSE;
 }
 
@@ -1091,7 +1082,10 @@ static boolean parse_immediate( struct translate_ctx *ctx )
    imm = tgsi_default_full_immediate();
    imm.Immediate.NrTokens += 4;
    imm.Immediate.DataType = TGSI_IMM_FLOAT32;
-   imm.u.Pointer = values;
+   imm.u[0].Float = values[0];
+   imm.u[1].Float = values[1];
+   imm.u[2].Float = values[2];
+   imm.u[3].Float = values[3];
 
    advance = tgsi_build_full_immediate(
       &imm,
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
new file mode 100644
index 00000000000..ba84a82b2b0
--- /dev/null
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -0,0 +1,797 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE, INC AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "pipe/p_context.h"
+#include "pipe/p_state.h"
+#include "tgsi/tgsi_ureg.h"
+#include "tgsi/tgsi_dump.h"
+#include "util/u_memory.h"
+
+union tgsi_any_token {
+   struct tgsi_version version;
+   struct tgsi_header header;
+   struct tgsi_processor processor;
+   struct tgsi_token token;
+   struct tgsi_declaration decl;
+   struct tgsi_declaration_range decl_range;
+   struct tgsi_declaration_semantic decl_semantic;
+   struct tgsi_immediate imm;
+   union  tgsi_immediate_data imm_data;
+   struct tgsi_instruction insn;
+   struct tgsi_instruction_ext_nv insn_ext_nv;
+   struct tgsi_instruction_ext_label insn_ext_label;
+   struct tgsi_instruction_ext_texture insn_ext_texture;
+   struct tgsi_instruction_ext_predicate insn_ext_predicate;
+   struct tgsi_src_register src;
+   struct tgsi_src_register_ext_swz src_ext_swz;
+   struct tgsi_src_register_ext_mod src_ext_mod;
+   struct tgsi_dimension dim;
+   struct tgsi_dst_register dst;
+   struct tgsi_dst_register_ext_concode dst_ext_code;
+   struct tgsi_dst_register_ext_modulate dst_ext_mod;
+   struct tgsi_dst_register_ext_predicate dst_ext_pred;
+   unsigned value;
+};
+
+
+struct ureg_tokens {
+   union tgsi_any_token *tokens;
+   unsigned size;
+   unsigned order;
+   unsigned count;
+};
+
+#define UREG_MAX_INPUT PIPE_MAX_ATTRIBS
+#define UREG_MAX_OUTPUT PIPE_MAX_ATTRIBS
+#define UREG_MAX_IMMEDIATE 32
+#define UREG_MAX_TEMP 256
+
+#define DOMAIN_DECL 0
+#define DOMAIN_INSN 1
+
+struct ureg_program
+{
+   unsigned processor;
+   struct pipe_context *pipe;
+
+   struct {
+      unsigned semantic_name;
+      unsigned semantic_index;
+      unsigned interp;
+   } input[UREG_MAX_INPUT];
+   unsigned nr_inputs;
+
+   struct {
+      unsigned semantic_name;
+      unsigned semantic_index;
+   } output[UREG_MAX_OUTPUT];
+   unsigned nr_outputs;
+
+   struct {
+      float v[4];
+      unsigned nr;
+   } immediate[UREG_MAX_IMMEDIATE];
+   unsigned nr_immediates;
+
+   unsigned temps_active[UREG_MAX_TEMP / 32];
+   unsigned nr_temps;
+
+   unsigned nr_constants;
+   unsigned nr_samplers;
+
+   struct ureg_tokens domain[2];
+};
+
+static union tgsi_any_token error_tokens[32];
+
+static void tokens_error( struct ureg_tokens *tokens )
+{
+   tokens->tokens = error_tokens;
+   tokens->size = Elements(error_tokens);
+   tokens->count = 0;
+}
+
+
+static void tokens_expand( struct ureg_tokens *tokens,
+                           unsigned count )
+{
+   unsigned old_size = tokens->size * sizeof(unsigned);
+
+   if (tokens->tokens == error_tokens)
+      goto fail;
+
+   while (tokens->count + count > tokens->size) {
+      tokens->size = (1 << ++tokens->order);
+   }
+
+   tokens->tokens = REALLOC(tokens->tokens, 
+                            old_size,
+                            tokens->size * sizeof(unsigned));
+   if (tokens->tokens == NULL) 
+      goto fail;
+
+   return;
+          
+fail:
+   tokens_error(tokens);
+}
+
+static void set_bad( struct ureg_program *ureg )
+{
+   tokens_error(&ureg->domain[0]);
+}
+
+
+
+static union tgsi_any_token *get_tokens( struct ureg_program *ureg,
+                                         unsigned domain,
+                                         unsigned count )
+{
+   struct ureg_tokens *tokens = &ureg->domain[domain];
+   union tgsi_any_token *result;
+
+   if (tokens->count + count > tokens->size) 
+      tokens_expand(tokens, count);
+
+   result = &tokens->tokens[tokens->count];
+   tokens->count += count;
+   return result;
+}
+
+
+static union tgsi_any_token *retrieve_token( struct ureg_program *ureg,
+                                            unsigned domain,
+                                            unsigned nr )
+{
+   if (ureg->domain[domain].tokens == error_tokens)
+      return &error_tokens[0];
+
+   return &ureg->domain[domain].tokens[nr];
+}
+
+
+
+static INLINE struct ureg_dst
+ureg_dst_register( unsigned file,
+                   unsigned index )
+{
+   struct ureg_dst dst;
+
+   dst.File      = file;
+   dst.WriteMask = TGSI_WRITEMASK_XYZW;
+   dst.Indirect  = 0;
+   dst.Saturate  = 0;
+   dst.Index     = index;
+   dst.Pad1      = 0;
+   dst.Pad2      = 0;
+
+   return dst;
+}
+
+static INLINE struct ureg_src 
+ureg_src_register( unsigned file,
+                   unsigned index )
+{
+   struct ureg_src src;
+
+   src.File     = file;
+   src.SwizzleX = TGSI_SWIZZLE_X;
+   src.SwizzleY = TGSI_SWIZZLE_Y;
+   src.SwizzleZ = TGSI_SWIZZLE_Z;
+   src.SwizzleW = TGSI_SWIZZLE_W;
+   src.Pad      = 0;
+   src.Indirect = 0;
+   src.Absolute = 0;
+   src.Index    = index;
+   src.Negate   = 0;
+
+   return src;
+}
+
+
+
+
+static struct ureg_src 
+ureg_DECL_input( struct ureg_program *ureg,
+                 unsigned name,
+                 unsigned index,
+                 unsigned interp_mode )
+{
+   unsigned i;
+
+   for (i = 0; i < ureg->nr_inputs; i++) {
+      if (ureg->input[i].semantic_name == name &&
+          ureg->input[i].semantic_index == index) 
+         goto out;
+   }
+
+   if (ureg->nr_inputs < UREG_MAX_INPUT) {
+      ureg->input[i].semantic_name = name;
+      ureg->input[i].semantic_index = index;
+      ureg->input[i].interp = interp_mode;
+      ureg->nr_inputs++;
+   }
+   else {
+      set_bad( ureg );
+   }
+
+out:
+   return ureg_src_register( TGSI_FILE_INPUT, i );
+}
+
+
+
+struct ureg_src 
+ureg_DECL_fs_input( struct ureg_program *ureg,
+                    unsigned name,
+                    unsigned index,
+                    unsigned interp )
+{
+   return ureg_DECL_input( ureg, name, index, interp );
+}
+
+
+struct ureg_src 
+ureg_DECL_vs_input( struct ureg_program *ureg,
+                    unsigned name,
+                    unsigned index )
+{
+   return ureg_DECL_input( ureg, name, index, TGSI_INTERPOLATE_CONSTANT );
+}
+
+
+struct ureg_dst 
+ureg_DECL_output( struct ureg_program *ureg,
+                  unsigned name,
+                  unsigned index )
+{
+   unsigned i;
+
+   for (i = 0; i < ureg->nr_outputs; i++) {
+      if (ureg->output[i].semantic_name == name &&
+          ureg->output[i].semantic_index == index) 
+         goto out;
+   }
+
+   if (ureg->nr_outputs < UREG_MAX_OUTPUT) {
+      ureg->output[i].semantic_name = name;
+      ureg->output[i].semantic_index = index;
+      ureg->nr_outputs++;
+   }
+   else {
+      set_bad( ureg );
+   }
+
+out:
+   return ureg_dst_register( TGSI_FILE_OUTPUT, i );
+}
+
+
+/* Returns a new constant register.  Keep track of which have been
+ * referred to so that we can emit decls later.
+ *
+ * There is nothing in this code to bind this constant to any tracked
+ * value or manage any constant_buffer contents -- that's the
+ * resposibility of the calling code.
+ */
+struct ureg_src ureg_DECL_constant(struct ureg_program *ureg )
+{
+   return ureg_src_register( TGSI_FILE_TEMPORARY, ureg->nr_constants++ );
+}
+
+
+/* Allocate a new temporary.  Temporaries greater than UREG_MAX_TEMP
+ * are legal, but will not be released.
+ */
+struct ureg_dst ureg_DECL_temporary( struct ureg_program *ureg )
+{
+   unsigned i;
+
+   for (i = 0; i < UREG_MAX_TEMP; i += 32) {
+      int bit = ffs(~ureg->temps_active[i/32]);
+      if (bit != 0) {
+         i += bit - 1;
+         goto out;
+      }
+   }
+
+   /* No reusable temps, so allocate a new one:
+    */
+   i = ureg->nr_temps++;
+
+out:
+   if (i < UREG_MAX_TEMP)
+      ureg->temps_active[i/32] |= 1 << (i % 32);
+
+   if (i >= ureg->nr_temps)
+      ureg->nr_temps = i + 1;
+
+   return ureg_dst_register( TGSI_FILE_TEMPORARY, i );
+}
+
+
+void ureg_release_temporary( struct ureg_program *ureg,
+                             struct ureg_dst tmp )
+{
+   if (tmp.Index < UREG_MAX_TEMP)
+      ureg->temps_active[tmp.Index/32] &= ~(1 << (tmp.Index % 32));
+}
+
+
+/* Allocate a new sampler.
+ */
+struct ureg_src ureg_DECL_sampler( struct ureg_program *ureg )
+{
+   return ureg_src_register( TGSI_FILE_SAMPLER, ureg->nr_samplers++ );
+}
+
+
+
+
+static int match_or_expand_immediate( const float *v,
+                                      unsigned nr,
+                                      float *v2,
+                                      unsigned *nr2,
+                                      unsigned *swizzle )
+{
+   unsigned i, j;
+
+   for (i = 0; i < nr; i++) {
+      boolean found = FALSE;
+
+      for (j = 0; j < *nr2 && !found; j++) {
+         if (v[i] == v2[j]) {
+            *swizzle |= j << (i * 2);
+            found = TRUE;
+         }
+      }
+
+      if (!found) {
+         if (*nr2 >= 4) 
+            return FALSE;
+
+         v2[*nr2] = v[i];
+         *swizzle |= *nr2 << (i * 2);
+         (*nr2)++;
+      }
+   }
+
+   return TRUE;
+}
+
+
+
+
+struct ureg_src ureg_DECL_immediate( struct ureg_program *ureg, 
+                                     const float *v,
+                                     unsigned nr )
+{
+   unsigned i;
+   unsigned swizzle;
+
+   /* Could do a first pass where we examine all existing immediates
+    * without expanding.
+    */
+
+   for (i = 0; i < ureg->nr_immediates; i++) {
+      if (match_or_expand_immediate( v, 
+                                     nr,
+                                     ureg->immediate[i].v,
+                                     &ureg->immediate[i].nr, 
+                                     &swizzle ))
+         goto out;
+   }
+
+   if (ureg->nr_immediates < UREG_MAX_IMMEDIATE) {
+      i = ureg->nr_immediates++;
+      if (match_or_expand_immediate( v,
+                                     nr,
+                                     ureg->immediate[i].v,
+                                     &ureg->immediate[i].nr, 
+                                     &swizzle ))
+         goto out;
+   }
+
+   set_bad( ureg );
+
+out:
+   return ureg_swizzle( ureg_src_register( TGSI_FILE_IMMEDIATE, i ),
+                        (swizzle >> 0) & 0x3,
+                        (swizzle >> 2) & 0x3,
+                        (swizzle >> 4) & 0x3,
+                        (swizzle >> 6) & 0x3);
+}
+
+
+void 
+ureg_emit_src( struct ureg_program *ureg,
+               struct ureg_src src )
+{
+   unsigned size = (1 + 
+                    (src.Absolute ? 1 : 0) +
+                    (src.Indirect ? 1 : 0));
+
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_INSN, size );
+   unsigned n = 0;
+
+   out[n].value = 0;
+   out[n].src.File = src.File;
+   out[n].src.SwizzleX = src.SwizzleX;
+   out[n].src.SwizzleY = src.SwizzleY;
+   out[n].src.SwizzleZ = src.SwizzleZ;
+   out[n].src.SwizzleW = src.SwizzleW;
+   out[n].src.Indirect = src.Indirect;
+   out[n].src.Index = src.Index;
+   n++;
+   
+   if (src.Absolute) {
+      out[n].value = 0;
+      out[n].src_ext_mod.Absolute = 1;
+      n++;
+   }
+
+   if (src.Indirect) {
+      out[n].value = 0;
+      out[n].src.File = TGSI_FILE_ADDRESS;
+      out[n].src.SwizzleX = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleY = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleZ = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleW = TGSI_SWIZZLE_X;
+      out[n].src.Indirect = 0;
+      out[n].src.Index = 0;
+      n++;
+   }
+
+   assert(n == size);
+}
+
+
+void 
+ureg_emit_dst( struct ureg_program *ureg,
+               struct ureg_dst dst )
+{
+   unsigned size = (1 + 
+                    (dst.Indirect ? 1 : 0));
+
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_INSN, size );
+   unsigned n = 0;
+
+   out[n].value = 0;
+   out[n].dst.File = dst.File;
+   out[n].dst.WriteMask = dst.WriteMask;
+   out[n].dst.Indirect = dst.Indirect;
+   out[n].dst.Index = dst.Index;
+   n++;
+   
+   if (dst.Indirect) {
+      out[n].value = 0;
+      out[n].src.File = TGSI_FILE_ADDRESS;
+      out[n].src.SwizzleX = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleY = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleZ = TGSI_SWIZZLE_X;
+      out[n].src.SwizzleW = TGSI_SWIZZLE_X;
+      out[n].src.Indirect = 0;
+      out[n].src.Index = 0;
+      n++;
+   }
+
+   assert(n == size);
+}
+
+
+
+unsigned
+ureg_emit_insn(struct ureg_program *ureg,
+               unsigned opcode,
+               boolean saturate,
+               unsigned num_dst,
+               unsigned num_src )
+{
+   union tgsi_any_token *out;
+
+   out = get_tokens( ureg, DOMAIN_INSN, 1 );
+   out[0].value = 0;
+   out[0].insn.Type = TGSI_TOKEN_TYPE_INSTRUCTION;
+   out[0].insn.NrTokens = 0;
+   out[0].insn.Opcode = opcode;
+   out[0].insn.Saturate = saturate;
+   out[0].insn.NrTokens = 0;
+   out[0].insn.NumDstRegs = num_dst;
+   out[0].insn.NumSrcRegs = num_src;
+   out[0].insn.Padding = 0;
+   out[0].insn.Extended = 0;
+
+   return ureg->domain[DOMAIN_INSN].count - 1;
+}
+
+
+void
+ureg_emit_label(struct ureg_program *ureg,
+                unsigned insn_token,
+                unsigned *label_token )
+{
+   union tgsi_any_token *out, *insn;
+
+   out = get_tokens( ureg, DOMAIN_INSN, 1 );
+   insn = retrieve_token( ureg, DOMAIN_INSN, insn_token );
+
+   insn->insn.Extended = 1;
+
+   out[0].value = 0;
+   out[0].insn_ext_label.Type = TGSI_INSTRUCTION_EXT_TYPE_LABEL;
+}
+
+
+void
+ureg_emit_texture(struct ureg_program *ureg,
+                  unsigned insn_token,
+                  unsigned target )
+{
+   union tgsi_any_token *out, *insn;
+
+   out = get_tokens( ureg, DOMAIN_INSN, 1 );
+   insn = retrieve_token( ureg, DOMAIN_INSN, insn_token );
+
+   insn->insn.Extended = 1;
+
+   out[0].value = 0;
+   out[0].insn_ext_texture.Type = TGSI_INSTRUCTION_EXT_TYPE_TEXTURE;
+   out[0].insn_ext_texture.Texture = target;
+}
+
+
+void
+ureg_fixup_insn_size(struct ureg_program *ureg,
+                     unsigned insn )
+{
+   union tgsi_any_token *out = retrieve_token( ureg, DOMAIN_INSN, insn );
+
+   out->insn.NrTokens = ureg->domain[DOMAIN_INSN].count - insn - 1;
+}
+
+
+
+
+
+static void emit_decl( struct ureg_program *ureg,
+                       unsigned file,
+                       unsigned index,
+                       unsigned semantic_name,
+                       unsigned semantic_index,
+                       unsigned interp )
+{
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 3 );
+
+   out[0].value = 0;
+   out[0].decl.Type = TGSI_TOKEN_TYPE_DECLARATION;
+   out[0].decl.NrTokens = 3;
+   out[0].decl.File = file;
+   out[0].decl.UsageMask = TGSI_WRITEMASK_XYZW; /* FIXME! */
+   out[0].decl.Interpolate = interp;
+   out[0].decl.Semantic = 1;
+
+   out[1].value = 0;
+   out[1].decl_range.First = 
+      out[1].decl_range.Last = index;
+
+   out[2].value = 0;
+   out[2].decl_semantic.SemanticName = semantic_name;
+   out[2].decl_semantic.SemanticIndex = semantic_index;
+
+}
+
+
+static void emit_decl_range( struct ureg_program *ureg,
+                             unsigned file,
+                             unsigned first,
+                             unsigned count )
+{
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 2 );
+
+   out[0].value = 0;
+   out[0].decl.Type = TGSI_TOKEN_TYPE_DECLARATION;
+   out[0].decl.NrTokens = 2;
+   out[0].decl.File = file;
+   out[0].decl.UsageMask = 0xf;
+   out[0].decl.Interpolate = TGSI_INTERPOLATE_CONSTANT;
+   out[0].decl.Semantic = 0;
+
+   out[1].value = 0;
+   out[1].decl_range.First = first;
+   out[1].decl_range.Last = first + count - 1;
+}
+
+static void emit_immediate( struct ureg_program *ureg,
+                            const float *v )
+{
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 5 );
+
+   out[0].value = 0;
+   out[0].imm.Type = TGSI_TOKEN_TYPE_IMMEDIATE;
+   out[0].imm.NrTokens = 5;
+   out[0].imm.DataType = TGSI_IMM_FLOAT32;
+   out[0].imm.Padding = 0;
+   out[0].imm.Extended = 0;
+
+   out[1].imm_data.Float = v[0];
+   out[2].imm_data.Float = v[1];
+   out[3].imm_data.Float = v[2];
+   out[4].imm_data.Float = v[3];
+}
+
+
+
+
+static void emit_decls( struct ureg_program *ureg )
+{
+   unsigned i;
+
+   for (i = 0; i < ureg->nr_inputs; i++) {
+      emit_decl( ureg, 
+                 TGSI_FILE_INPUT, 
+                 i,
+                 ureg->input[i].semantic_name,
+                 ureg->input[i].semantic_index,
+                 ureg->input[i].interp );
+   }
+
+   for (i = 0; i < ureg->nr_outputs; i++) {
+      emit_decl( ureg, 
+                 TGSI_FILE_OUTPUT, 
+                 i,
+                 ureg->output[i].semantic_name,
+                 ureg->output[i].semantic_index,
+                 TGSI_INTERPOLATE_CONSTANT );
+   }
+
+   if (ureg->nr_samplers) {
+      emit_decl_range( ureg, 
+                       TGSI_FILE_SAMPLER,
+                       0, ureg->nr_samplers );
+   }
+
+   if (ureg->nr_constants) {
+      emit_decl_range( ureg,
+                       TGSI_FILE_CONSTANT,
+                       0, ureg->nr_constants );
+   }
+
+   if (ureg->nr_temps) {
+      emit_decl_range( ureg,
+                       TGSI_FILE_TEMPORARY,
+                       0, ureg->nr_temps );
+   }
+
+   for (i = 0; i < ureg->nr_immediates; i++) {
+      emit_immediate( ureg,
+                      ureg->immediate[i].v );
+   }
+}
+
+/* Append the instruction tokens onto the declarations to build a
+ * contiguous stream suitable to send to the driver.
+ */
+static void copy_instructions( struct ureg_program *ureg )
+{
+   unsigned nr_tokens = ureg->domain[DOMAIN_INSN].count;
+   union tgsi_any_token *out = get_tokens( ureg, 
+                                           DOMAIN_DECL, 
+                                           nr_tokens );
+
+   memcpy(out, 
+          ureg->domain[DOMAIN_INSN].tokens, 
+          nr_tokens * sizeof out[0] );
+}
+
+
+static void
+fixup_header_size(struct ureg_program *ureg,
+                     unsigned insn )
+{
+   union tgsi_any_token *out = retrieve_token( ureg, DOMAIN_DECL, 1 );
+
+   out->header.BodySize = ureg->domain[DOMAIN_DECL].count - 3;
+}
+
+
+static void
+emit_header( struct ureg_program *ureg )
+{
+   union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 3 );
+
+   out[0].version.MajorVersion = 1;
+   out[0].version.MinorVersion = 1;
+   out[0].version.Padding = 0;
+
+   out[1].header.HeaderSize = 2;
+   out[1].header.BodySize = 0;
+
+   out[2].processor.Processor = ureg->processor;
+   out[2].processor.Padding = 0;
+}
+
+
+void *ureg_create_shader( struct ureg_program *ureg )
+{
+   struct pipe_shader_state state;
+   unsigned insn;
+
+   emit_header( ureg );
+   emit_decls( ureg );
+   copy_instructions( ureg );
+   fixup_header_size( ureg, insn );
+   
+   if (ureg->domain[0].tokens == error_tokens ||
+       ureg->domain[1].tokens == error_tokens) {
+      debug_printf("%s: error in generated shader\n", __FUNCTION__);
+      assert(0);
+      return NULL;
+   }
+
+   state.tokens = (const struct tgsi_token *)ureg->domain[DOMAIN_DECL].tokens;
+
+   if (0) {
+      debug_printf("%s: emitted shader %d tokens:\n", __FUNCTION__, 
+                   ureg->domain[DOMAIN_DECL].count);
+      tgsi_dump( state.tokens, 0 );
+   }
+
+   if (ureg->processor == TGSI_PROCESSOR_VERTEX)
+      return ureg->pipe->create_vs_state( ureg->pipe, &state );
+   else
+      return ureg->pipe->create_fs_state( ureg->pipe, &state );
+}
+
+
+
+
+struct ureg_program *ureg_create( struct pipe_context *pipe,
+                                  unsigned processor )
+{
+   struct ureg_program *ureg = CALLOC_STRUCT( ureg_program );
+   if (ureg == NULL)
+      return NULL;
+
+   ureg->pipe = pipe;
+   ureg->processor = processor;
+   return ureg;
+}
+
+
+void ureg_destroy( struct ureg_program *ureg )
+{
+   unsigned i;
+
+   for (i = 0; i < Elements(ureg->domain); i++) {
+      if (ureg->domain[i].tokens && 
+          ureg->domain[i].tokens != error_tokens)
+         FREE(ureg->domain[i].tokens);
+   }
+   
+   FREE(ureg);
+}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
new file mode 100644
index 00000000000..0a976fd63b7
--- /dev/null
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
@@ -0,0 +1,439 @@
+/**************************************************************************
+ * 
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE, INC AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef TGSI_UREG_H
+#define TGSI_UREG_H
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_shader_tokens.h"
+
+struct ureg_program;
+
+/* Almost a tgsi_src_register, but we need to pull in the Absolute
+ * flag from the _ext token.  Indirect flag always implies ADDR[0].
+ */
+struct ureg_src
+{
+   unsigned File        : 4;  /* TGSI_FILE_ */
+   unsigned SwizzleX    : 2;  /* TGSI_SWIZZLE_ */
+   unsigned SwizzleY    : 2;  /* TGSI_SWIZZLE_ */
+   unsigned SwizzleZ    : 2;  /* TGSI_SWIZZLE_ */
+   unsigned SwizzleW    : 2;  /* TGSI_SWIZZLE_ */
+   unsigned Pad         : 1;  /* BOOL */
+   unsigned Indirect    : 1;  /* BOOL */
+   unsigned Absolute    : 1;  /* BOOL */
+   int      Index       : 16; /* SINT */
+   unsigned Negate      : 1;  /* BOOL */
+};
+
+/* Very similar to a tgsi_dst_register, removing unsupported fields
+ * and adding a Saturate flag.  It's easier to push saturate into the
+ * destination register than to try and create a _SAT varient of each
+ * instruction function.
+ */
+struct ureg_dst
+{
+   unsigned File        : 4;  /* TGSI_FILE_ */
+   unsigned WriteMask   : 4;  /* TGSI_WRITEMASK_ */
+   unsigned Indirect    : 1;  /* BOOL */
+   unsigned Saturate    : 1;  /* BOOL */
+   int      Index       : 16; /* SINT */
+   unsigned Pad1        : 5;
+   unsigned Pad2        : 1;  /* BOOL */
+};
+
+struct pipe_context;
+
+struct ureg_program *
+ureg_create( struct pipe_context *pipe,
+             unsigned processor );
+
+void *
+ureg_create_shader( struct ureg_program * );
+
+void 
+ureg_destroy( struct ureg_program * );
+
+
+/***********************************************************************
+ * Convenience routine:
+ */
+static INLINE void *ureg_create_shader_and_destroy( struct ureg_program *p )
+{
+   void *result = ureg_create_shader( p );
+   ureg_destroy( p );
+   return result;
+}
+
+
+
+/***********************************************************************
+ * Build shader declarations:
+ */
+
+struct ureg_src
+ureg_DECL_fs_input( struct ureg_program *,
+                    unsigned semantic_name,
+                    unsigned semantic_index,
+                    unsigned interp_mode );
+
+struct ureg_src
+ureg_DECL_vs_input( struct ureg_program *,
+                    unsigned semantic_name,
+                    unsigned semantic_index );
+
+struct ureg_dst
+ureg_DECL_output( struct ureg_program *,
+                  unsigned semantic_name,
+                  unsigned semantic_index );
+
+struct ureg_src
+ureg_DECL_immediate( struct ureg_program *,
+                     const float *v,
+                     unsigned nr );
+
+struct ureg_src
+ureg_DECL_constant( struct ureg_program * );
+
+struct ureg_dst
+ureg_DECL_temporary( struct ureg_program * );
+
+void 
+ureg_release_temporary( struct ureg_program *ureg,
+                        struct ureg_dst tmp );
+
+struct ureg_src
+ureg_DECL_sampler( struct ureg_program * );
+
+
+static INLINE struct ureg_src
+ureg_DECL_immediate4f( struct ureg_program *ureg,
+                       float a, float b,
+                       float c, float d)
+{
+   float v[4];
+   v[0] = a;
+   v[1] = b;
+   v[2] = c;
+   v[3] = d;
+   return ureg_DECL_immediate( ureg, v, 4 );
+}
+
+static INLINE struct ureg_src
+ureg_DECL_immediate3f( struct ureg_program *ureg,
+                       float a, float b,
+                       float c)
+{
+   float v[3];
+   v[0] = a;
+   v[1] = b;
+   v[2] = c;
+   return ureg_DECL_immediate( ureg, v, 3 );
+}
+
+static INLINE struct ureg_src
+ureg_DECL_immediate2f( struct ureg_program *ureg,
+                       float a, float b)
+{
+   float v[2];
+   v[0] = a;
+   v[1] = b;
+   return ureg_DECL_immediate( ureg, v, 2 );
+}
+
+static INLINE struct ureg_src
+ureg_DECL_immediate1f( struct ureg_program *ureg,
+                       float a)
+{
+   float v[1];
+   v[0] = a;
+   return ureg_DECL_immediate( ureg, v, 1 );
+}
+
+/***********************************************************************
+ * Internal instruction helpers, don't call these directly:
+ */
+
+unsigned
+ureg_emit_insn(struct ureg_program *ureg,
+               unsigned opcode,
+               boolean saturate,
+               unsigned num_dst,
+               unsigned num_src );
+
+void
+ureg_emit_label(struct ureg_program *ureg,
+                unsigned insn_token,
+                unsigned *label_token );
+
+void
+ureg_emit_texture(struct ureg_program *ureg,
+                  unsigned insn_token,
+                  unsigned target );
+
+void 
+ureg_emit_dst( struct ureg_program *ureg,
+               struct ureg_dst dst );
+
+void 
+ureg_emit_src( struct ureg_program *ureg,
+               struct ureg_src src );
+
+void
+ureg_fixup_insn_size(struct ureg_program *ureg,
+                     unsigned insn );
+
+
+#define OP00( op )                                              \
+static INLINE void ureg_##op( struct ureg_program *ureg )       \
+{                                                               \
+   unsigned opcode = TGSI_OPCODE_##op;                          \
+   unsigned insn = ureg_emit_insn( ureg, opcode, FALSE, 0, 0 ); \
+   ureg_fixup_insn_size( ureg, insn );                          \
+}
+
+#define OP01( op )                                              \
+static INLINE void ureg_##op( struct ureg_program *ureg,        \
+                              struct ureg_src src )             \
+{                                                               \
+   unsigned opcode = TGSI_OPCODE_##op;                          \
+   unsigned insn = ureg_emit_insn( ureg, opcode, FALSE, 0, 1 ); \
+   ureg_emit_src( ureg, src );                                  \
+   ureg_fixup_insn_size( ureg, insn );                          \
+}
+
+#define OP00_LBL( op )                                          \
+static INLINE void ureg_##op( struct ureg_program *ureg,        \
+                              unsigned *label_token )           \
+{                                                               \
+   unsigned opcode = TGSI_OPCODE_##op;                          \
+   unsigned insn = ureg_emit_insn( ureg, opcode, FALSE, 0, 0 ); \
+   ureg_emit_label( ureg, insn, label_token );                  \
+   ureg_fixup_insn_size( ureg, insn );                          \
+}
+
+#define OP01_LBL( op )                                          \
+static INLINE void ureg_##op( struct ureg_program *ureg,        \
+                              struct ureg_src src,              \
+                              unsigned *label_token )          \
+{                                                               \
+   unsigned opcode = TGSI_OPCODE_##op;                          \
+   unsigned insn = ureg_emit_insn( ureg, opcode, FALSE, 0, 1 ); \
+   ureg_emit_label( ureg, insn, label_token );                  \
+   ureg_emit_src( ureg, src );                                  \
+   ureg_fixup_insn_size( ureg, insn );                          \
+}
+
+#define OP10( op )                                                      \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst )                     \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 0 );  \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+
+#define OP11( op )                                                      \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst,                      \
+                              struct ureg_src src )                     \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 1 );  \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_emit_src( ureg, src );                                          \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+#define OP12( op )                                                      \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst,                      \
+                              struct ureg_src src0,                     \
+                              struct ureg_src src1 )                    \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 2 );  \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_emit_src( ureg, src0 );                                         \
+   ureg_emit_src( ureg, src1 );                                         \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+#define OP12_TEX( op )                                                  \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst,                      \
+                              unsigned target,                          \
+                              struct ureg_src src0,                     \
+                              struct ureg_src src1 )                    \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 2 );  \
+   ureg_emit_texture( ureg, insn, target );                             \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_emit_src( ureg, src0 );                                         \
+   ureg_emit_src( ureg, src1 );                                         \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+#define OP13( op )                                                      \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst,                      \
+                              struct ureg_src src0,                     \
+                              struct ureg_src src1,                     \
+                              struct ureg_src src2 )                    \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 3 );  \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_emit_src( ureg, src0 );                                         \
+   ureg_emit_src( ureg, src1 );                                         \
+   ureg_emit_src( ureg, src2 );                                         \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+#define OP14_TEX( op )                                                  \
+static INLINE void ureg_##op( struct ureg_program *ureg,                \
+                              struct ureg_dst dst,                      \
+                              unsigned target,                          \
+                              struct ureg_src src0,                     \
+                              struct ureg_src src1,                     \
+                              struct ureg_src src2,                     \
+                              struct ureg_src src3 )                    \
+{                                                                       \
+   unsigned opcode = TGSI_OPCODE_##op;                                  \
+   unsigned insn = ureg_emit_insn( ureg, opcode, dst.Saturate, 1, 4 );  \
+   ureg_emit_texture( ureg, insn, target );                             \
+   ureg_emit_dst( ureg, dst );                                          \
+   ureg_emit_src( ureg, src0 );                                         \
+   ureg_emit_src( ureg, src1 );                                         \
+   ureg_emit_src( ureg, src2 );                                         \
+   ureg_emit_src( ureg, src3 );                                         \
+   ureg_fixup_insn_size( ureg, insn );                                  \
+}
+
+
+/* Use a template include to generate a correctly-typed ureg_OP()
+ * function for each TGSI opcode:
+ */
+#include "tgsi_opcode_tmp.h"
+
+
+/***********************************************************************
+ * Inline helpers for manipulating register structs:
+ */
+static INLINE struct ureg_src 
+ureg_negate( struct ureg_src reg )
+{
+   reg.Negate ^= 1;
+   return reg;
+}
+
+static INLINE struct ureg_src
+ureg_abs( struct ureg_src reg )
+{
+   reg.Absolute = 1;
+   reg.Negate = 0;
+   return reg;
+}
+
+static INLINE struct ureg_src 
+ureg_swizzle( struct ureg_src reg, 
+              int x, int y, int z, int w )
+{
+   unsigned swz = ( (reg.SwizzleX << 0) |
+                    (reg.SwizzleY << 2) |
+                    (reg.SwizzleZ << 4) |
+                    (reg.SwizzleW << 6));
+
+   reg.SwizzleX = (swz >> (x*2)) & 0x3;
+   reg.SwizzleY = (swz >> (y*2)) & 0x3;
+   reg.SwizzleZ = (swz >> (z*2)) & 0x3;
+   reg.SwizzleW = (swz >> (w*2)) & 0x3;
+   return reg;
+}
+
+static INLINE struct ureg_src
+ureg_scalar( struct ureg_src reg, int x )
+{
+   return ureg_swizzle(reg, x, x, x, x);
+}
+
+static INLINE struct ureg_dst 
+ureg_writemask( struct ureg_dst reg,
+                unsigned writemask )
+{
+   reg.WriteMask &= writemask;
+   return reg;
+}
+
+static INLINE struct ureg_dst 
+ureg_saturate( struct ureg_dst reg )
+{
+   reg.Saturate = 1;
+   return reg;
+}
+
+static INLINE struct ureg_dst
+ureg_dst( struct ureg_src src )
+{
+   struct ureg_dst dst;
+
+   dst.File      = src.File;
+   dst.WriteMask = TGSI_WRITEMASK_XYZW;
+   dst.Indirect  = src.Indirect;
+   dst.Saturate  = 0;
+   dst.Index     = src.Index;
+   dst.Pad1      = 0;
+   dst.Pad2      = 0;
+
+   return dst;
+}
+
+static INLINE struct ureg_src
+ureg_src( struct ureg_dst dst )
+{
+   struct ureg_src src;
+
+   src.File      = dst.File;
+   src.SwizzleX  = TGSI_SWIZZLE_X;
+   src.SwizzleY  = TGSI_SWIZZLE_Y;
+   src.SwizzleZ  = TGSI_SWIZZLE_Z;
+   src.SwizzleW  = TGSI_SWIZZLE_W;
+   src.Pad       = 0;
+   src.Indirect  = dst.Indirect;
+   src.Absolute  = 0;
+   src.Index     = dst.Index;
+   src.Negate    = 0;
+
+   return src;
+}
+
+
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index 414cf910254..cda6dbd46d7 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -45,6 +45,7 @@
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
+#include "util/u_surface.h"
 
 #include "cso_cache/cso_context.h"
 
@@ -155,7 +156,11 @@ util_destroy_blit(struct blit_state *ctx)
 }
 
 
-static unsigned get_next_slot( struct blit_state *ctx )
+/**
+ * Get offset of next free slot in vertex buffer for quad vertices.
+ */
+static unsigned
+get_next_slot( struct blit_state *ctx )
 {
    const unsigned max_slots = 4096 / sizeof ctx->vertices;
 
@@ -173,7 +178,6 @@ static unsigned get_next_slot( struct blit_state *ctx )
 }
                                
 
-
 /**
  * Setup vertex data for the textured quad we'll draw.
  * Note: y=0=top
@@ -260,9 +264,38 @@ setup_vertex_data_tex(struct blit_state *ctx,
 
    return offset;
 }
+
+
+/**
+ * \return TRUE if two regions overlap, FALSE otherwise
+ */
+static boolean
+regions_overlap(int srcX0, int srcY0,
+                int srcX1, int srcY1,
+                int dstX0, int dstY0,
+                int dstX1, int dstY1)
+{
+   if (MAX2(srcX0, srcX1) < MIN2(dstX0, dstX1))
+      return FALSE; /* src completely left of dst */
+
+   if (MAX2(dstX0, dstX1) < MIN2(srcX0, srcX1))
+      return FALSE; /* dst completely left of src */
+
+   if (MAX2(srcY0, srcY1) < MIN2(dstY0, dstY1))
+      return FALSE; /* src completely above dst */
+
+   if (MAX2(dstY0, dstY1) < MIN2(srcY0, srcY1))
+      return FALSE; /* dst completely above src */
+
+   return TRUE; /* some overlap */
+}
+
+
 /**
  * Copy pixel block from src surface to dst surface.
  * Overlapping regions are acceptable.
+ * Flipping and stretching are supported.
+ * XXX what about clipping???
  * XXX need some control over blitting Z and/or stencil.
  */
 void
@@ -285,10 +318,41 @@ util_blit_pixels(struct blit_state *ctx,
    const int srcLeft = MIN2(srcX0, srcX1);
    const int srcTop = MIN2(srcY0, srcY1);
    unsigned offset;
+   boolean overlap;
 
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
 
+   assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE_2D,
+                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
+   assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
+                                      PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
+
+   /* do the regions overlap? */
+   overlap = util_same_surface(src, dst) &&
+      regions_overlap(srcX0, srcY0, srcX1, srcY1,
+                      dstX0, dstY0, dstX1, dstY1);
+
+   /*
+    * Check for simple case:  no format conversion, no flipping, no stretching,
+    * no overlapping.
+    * Filter mode should not matter since there's no stretching.
+    */
+   if (dst->format == src->format &&
+       srcX0 < srcX1 &&
+       dstX0 < dstX1 &&
+       srcY0 < srcY1 &&
+       dstY0 < dstY1 &&
+       (dstX1 - dstX0) == (srcX1 - srcX0) &&
+       (dstY1 - dstY0) == (srcY1 - srcY0) &&
+       !overlap) {
+      pipe->surface_copy(pipe,
+			 dst, dstX0, dstY0, /* dest */
+			 src, srcX0, srcY0, /* src */
+			 srcW, srcH);       /* size */
+      return;
+   }
+   
    if (srcLeft != srcX0) {
       /* left-right flip */
       int tmp = dstX0;
@@ -303,20 +367,6 @@ util_blit_pixels(struct blit_state *ctx,
       dstY1 = tmp;
    }
 
-   assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
-   assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
-
-   if(dst->format == src->format && (dstX1 - dstX0) == srcW && (dstY1 - dstY0) == srcH) {
-      /* FIXME: this will most surely fail for overlapping rectangles */
-      pipe->surface_copy(pipe,
-			 dst, dstX0, dstY0,   /* dest */
-			 src, srcX0, srcY0, /* src */
-			 srcW, srcH);     /* size */
-      return;
-   }
-   
    assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
                                       PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
 
diff --git a/src/gallium/auxiliary/util/u_debug.c b/src/gallium/auxiliary/util/u_debug.c
index a5ca0b72bd7..96d400c839b 100644
--- a/src/gallium/auxiliary/util/u_debug.c
+++ b/src/gallium/auxiliary/util/u_debug.c
@@ -143,11 +143,9 @@ void _debug_vprintf(const char *format, va_list ap)
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
    /* TODO */
 #else /* !PIPE_SUBSYSTEM_WINDOWS */
-#ifdef DEBUG
    fflush(stdout);
    vfprintf(stderr, format, ap);
 #endif
-#endif
 }
 
 
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c
index ca797486a0e..edc37561ab1 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.c
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.c
@@ -46,10 +46,6 @@
 #include "util/u_gen_mipmap.h"
 #include "util/u_simple_shaders.h"
 
-#include "tgsi/tgsi_build.h"
-#include "tgsi/tgsi_dump.h"
-#include "tgsi/tgsi_parse.h"
-
 #include "cso_cache/cso_context.h"
 
 
diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index e5003af01d8..57410e78b02 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -53,11 +53,11 @@ __inline double ceil(double val)
 {
    double ceil_val;
 
-   if((val - (long) val) == 0) {
+   if ((val - (long) val) == 0) {
       ceil_val = val;
    }
    else {
-      if(val > 0) {
+      if (val > 0) {
          ceil_val = (long) val + 1;
       }
       else {
@@ -73,11 +73,11 @@ __inline double floor(double val)
 {
    double floor_val;
 
-   if((val - (long) val) == 0) {
+   if ((val - (long) val) == 0) {
       floor_val = val;
    }
    else {
-      if(val > 0) {
+      if (val > 0) {
          floor_val = (long) val;
       }
       else {
@@ -189,7 +189,10 @@ static INLINE double log2( double x )
 extern float pow2_table[POW2_TABLE_SIZE];
 
 
-
+/**
+ * Initialize math module.  This should be called before using any
+ * other functions in this module.
+ */
 extern void
 util_init_math(void);
 
@@ -216,23 +219,24 @@ util_fast_exp2(float x)
    int32_t ipart;
    float fpart, mpart;
    union fi epart;
-   
+
    if(x > 129.00000f)
       return 3.402823466e+38f;
-   
-   if(x < -126.99999f)
+
+   if (x < -126.99999f)
       return 0.0f;
 
    ipart = (int32_t) x;
    fpart = x - (float) ipart;
-   
+
    /* same as
     *   epart.f = (float) (1 << ipart)
-    * but faster and without integer overflow for ipart > 31 */
+    * but faster and without integer overflow for ipart > 31
+    */
    epart.i = (ipart + 127 ) << 23;
-   
+
    mpart = pow2_table[POW2_TABLE_OFFSET + (int)(fpart * POW2_TABLE_SCALE)];
-   
+
    return epart.f * mpart;
 }
 
@@ -254,6 +258,9 @@ util_fast_exp(float x)
 extern float log2_table[LOG2_TABLE_SIZE];
 
 
+/**
+ * Fast approximation to log2(x).
+ */
 static INLINE float
 util_fast_log2(float x)
 {
@@ -267,6 +274,9 @@ util_fast_log2(float x)
 }
 
 
+/**
+ * Fast approximation to x^y.
+ */
 static INLINE float
 util_fast_pow(float x, float y)
 {
@@ -274,7 +284,6 @@ util_fast_pow(float x, float y)
 }
 
 
-
 /**
  * Floor(x), returned as int.
  */
@@ -284,8 +293,8 @@ util_ifloor(float f)
    int ai, bi;
    double af, bf;
    union fi u;
-   af = (3 << 22) + 0.5 + (double)f;
-   bf = (3 << 22) + 0.5 - (double)f;
+   af = (3 << 22) + 0.5 + (double) f;
+   bf = (3 << 22) + 0.5 - (double) f;
    u.f = (float) af;  ai = u.i;
    u.f = (float) bf;  bi = u.i;
    return (ai - bi) >> 1;
@@ -305,9 +314,9 @@ util_iround(float f)
 #elif defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
    int r;
    _asm {
-	 fld f
-	 fistp r
-	}
+      fld f
+      fistp r
+   }
    return r;
 #else
    if (f >= 0.0f)
@@ -340,7 +349,7 @@ static INLINE
 unsigned long ffs( unsigned long u )
 {
    unsigned long i;
-   if(_BitScanForward(&i, u))
+   if (_BitScanForward(&i, u))
       return i + 1;
    else
       return 0;
@@ -351,7 +360,7 @@ unsigned ffs( unsigned u )
 {
    unsigned i;
 
-   if( u == 0 ) {
+   if (u == 0) {
       return 0;
    }
 
@@ -378,7 +387,10 @@ fui( float f )
 }
 
 
-
+/**
+ * Convert ubyte to float in [0, 1].
+ * XXX a 256-entry lookup table would be slightly faster.
+ */
 static INLINE float
 ubyte_to_float(ubyte ub)
 {
@@ -409,7 +421,23 @@ float_to_ubyte(float f)
 }
 
 
+/**
+ * Calc log base 2
+ */
+static INLINE unsigned
+util_logbase2(unsigned n)
+{
+   unsigned log2 = 0;
+   while (n >>= 1)
+      ++log2;
+   return log2;
+}
+
 
+/**
+ * Clamp X to [MIN, MAX].
+ * This is a macro to allow float, int, uint, etc. types.
+ */
 #define CLAMP( X, MIN, MAX )  ( (X)<(MIN) ? (MIN) : ((X)>(MAX) ? (MAX) : (X)) )
 
 #define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
@@ -422,6 +450,11 @@ align(int value, int alignment)
    return (value + alignment - 1) & ~(alignment - 1);
 }
 
+static INLINE unsigned
+minify(unsigned value)
+{
+    return MAX2(1, value >> 1);
+}
 
 #ifndef COPY_4V
 #define COPY_4V( DST, SRC )         \
diff --git a/src/gallium/auxiliary/util/u_memory.h b/src/gallium/auxiliary/util/u_memory.h
index 0b18d043adb..c3f8c918338 100644
--- a/src/gallium/auxiliary/util/u_memory.h
+++ b/src/gallium/auxiliary/util/u_memory.h
@@ -100,8 +100,14 @@ ExFreePool(void *P);
 #define MALLOC( SIZE )  malloc( SIZE )
 #define CALLOC( COUNT, SIZE )   calloc( COUNT, SIZE )
 #define FREE( PTR )  free( PTR )
-#define REALLOC( OLDPTR, OLDSIZE, NEWSIZE )  realloc( OLDPTR, NEWSIZE )
 
+static INLINE void *
+_REALLOC( void *old_ptr, unsigned old_size, unsigned new_size )
+{
+   (void) old_size;
+   return realloc(old_ptr, new_size);
+}
+#define REALLOC( a, b, c ) _REALLOC( a, b, c )
 #endif
 
 
diff --git a/src/gallium/auxiliary/util/u_mm.c b/src/gallium/auxiliary/util/u_mm.c
index 151a480d34d..4b75d4ba1d0 100644
--- a/src/gallium/auxiliary/util/u_mm.c
+++ b/src/gallium/auxiliary/util/u_mm.c
@@ -33,30 +33,32 @@
 void
 u_mmDumpMemInfo(const struct mem_block *heap)
 {
-   debug_printf("Memory heap %p:\n", (void *)heap);
+   debug_printf("Memory heap %p:\n", (void *) heap);
    if (heap == 0) {
       debug_printf("  heap == 0\n");
-   } else {
+   }
+   else {
       const struct mem_block *p;
 
-      for(p = heap->next; p != heap; p = p->next) {
-	 debug_printf("  Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size,
-		 p->free ? 'F':'.',
-		 p->reserved ? 'R':'.');
+      for (p = heap->next; p != heap; p = p->next) {
+	 debug_printf("  Offset:%08x, Size:%08x, %c%c\n", p->ofs, p->size,
+                      p->free ? 'F':'.',
+                      p->reserved ? 'R':'.');
       }
 
       debug_printf("\nFree list:\n");
 
-      for(p = heap->next_free; p != heap; p = p->next_free) {
-	 debug_printf(" FREE Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size,
-		 p->free ? 'F':'.',
-		 p->reserved ? 'R':'.');
+      for (p = heap->next_free; p != heap; p = p->next_free) {
+	 debug_printf(" FREE Offset:%08x, Size:%08x, %c%c\n", p->ofs, p->size,
+                      p->free ? 'F':'.',
+                      p->reserved ? 'R':'.');
       }
 
    }
    debug_printf("End of memory blocks\n");
 }
 
+
 struct mem_block *
 u_mmInit(int ofs, int size)
 {
diff --git a/src/gallium/auxiliary/util/u_mm.h b/src/gallium/auxiliary/util/u_mm.h
index ce20e487635..6b158aae6e4 100644
--- a/src/gallium/auxiliary/util/u_mm.h
+++ b/src/gallium/auxiliary/util/u_mm.h
@@ -84,7 +84,7 @@ extern struct mem_block *u_mmFindBlock(struct mem_block *heap, int start);
 extern void u_mmDestroy(struct mem_block *mmInit);
 
 /**
- * For debuging purpose.
+ * For debugging purposes.
  */
 extern void u_mmDumpMemInfo(const struct mem_block *mmInit);
 
diff --git a/src/gallium/auxiliary/util/u_simple_screen.c b/src/gallium/auxiliary/util/u_simple_screen.c
index 8114b53cd0d..f01296b40fc 100644
--- a/src/gallium/auxiliary/util/u_simple_screen.c
+++ b/src/gallium/auxiliary/util/u_simple_screen.c
@@ -65,12 +65,13 @@ pass_surface_buffer_create(struct pipe_screen *screen,
                            unsigned width, unsigned height,
                            enum pipe_format format,
                            unsigned usage,
+                           unsigned tex_usage,
                            unsigned *stride)
 {
    struct pipe_buffer *buffer =
       screen->winsys->surface_buffer_create(screen->winsys,
                                                 width, height,
-                                                format, usage, stride);
+                                                format, usage, tex_usage, stride);
 
    buffer->screen = screen;
 
diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c
index e519c354d25..1152d62e73e 100644
--- a/src/gallium/auxiliary/util/u_simple_shaders.c
+++ b/src/gallium/auxiliary/util/u_simple_shaders.c
@@ -42,9 +42,7 @@
 #include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
 
-#include "tgsi/tgsi_build.h"
-#include "tgsi/tgsi_dump.h"
-#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_ureg.h"
 
 
 
@@ -58,93 +56,31 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
                                     const uint *semantic_indexes)
                                     
 {
-   struct pipe_shader_state shader;
-   struct tgsi_token tokens[100];
-   struct tgsi_header *header;
-   struct tgsi_processor *processor;
-   struct tgsi_full_declaration decl;
-   struct tgsi_full_instruction inst;
-   const uint procType = TGSI_PROCESSOR_VERTEX;
-   uint ti, i;
+   struct ureg_program *ureg;
+   uint i;
 
-   /* shader header
-    */
-   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
+   ureg = ureg_create( pipe, TGSI_PROCESSOR_VERTEX );
+   if (ureg == NULL)
+      return NULL;
 
-   header = (struct tgsi_header *) &tokens[1];
-   *header = tgsi_build_header();
-
-   processor = (struct tgsi_processor *) &tokens[2];
-   *processor = tgsi_build_processor( procType, header );
-
-   ti = 3;
-
-   /* declare inputs */
-   for (i = 0; i < num_attribs; i++) {
-      decl = tgsi_default_full_declaration();
-      decl.Declaration.File = TGSI_FILE_INPUT;
-
-      decl.Declaration.Semantic = 1;
-      decl.Semantic.SemanticName = semantic_names[i];
-      decl.Semantic.SemanticIndex = semantic_indexes[i];
-
-      decl.DeclarationRange.First = 
-      decl.DeclarationRange.Last = i;
-      ti += tgsi_build_full_declaration(&decl,
-                                        &tokens[ti],
-                                        header,
-                                        Elements(tokens) - ti);
-   }
-
-   /* declare outputs */
    for (i = 0; i < num_attribs; i++) {
-      decl = tgsi_default_full_declaration();
-      decl.Declaration.File = TGSI_FILE_OUTPUT;
-      decl.Declaration.Semantic = 1;
-      decl.Semantic.SemanticName = semantic_names[i];
-      decl.Semantic.SemanticIndex = semantic_indexes[i];
-      decl.DeclarationRange.First = 
-         decl.DeclarationRange.Last = i;
-      ti += tgsi_build_full_declaration(&decl,
-                                        &tokens[ti],
-                                        header,
-                                        Elements(tokens) - ti);
+      struct ureg_src src;
+      struct ureg_dst dst;
+
+      src = ureg_DECL_vs_input( ureg,
+                                semantic_names[i],
+                                semantic_indexes[i]);
+      
+      dst = ureg_DECL_output( ureg,
+                              semantic_names[i],
+                              semantic_indexes[i]);
+      
+      ureg_MOV( ureg, dst, src );
    }
 
-   /* emit MOV instructions */
-   for (i = 0; i < num_attribs; i++) {
-      /* MOVE out[i], in[i]; */
-      inst = tgsi_default_full_instruction();
-      inst.Instruction.Opcode = TGSI_OPCODE_MOV;
-      inst.Instruction.NumDstRegs = 1;
-      inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
-      inst.FullDstRegisters[0].DstRegister.Index = i;
-      inst.Instruction.NumSrcRegs = 1;
-      inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
-      inst.FullSrcRegisters[0].SrcRegister.Index = i;
-      ti += tgsi_build_full_instruction(&inst,
-                                        &tokens[ti],
-                                        header,
-                                        Elements(tokens) - ti );
-   }
-
-   /* END instruction */
-   inst = tgsi_default_full_instruction();
-   inst.Instruction.Opcode = TGSI_OPCODE_END;
-   inst.Instruction.NumDstRegs = 0;
-   inst.Instruction.NumSrcRegs = 0;
-   ti += tgsi_build_full_instruction(&inst,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti );
-
-#if 0 /*debug*/
-   tgsi_dump(tokens, 0);
-#endif
-
-   shader.tokens = tokens;
+   ureg_END( ureg );
 
-   return pipe->create_vs_state(pipe, &shader);
+   return ureg_create_shader_and_destroy( ureg );
 }
 
 
@@ -158,99 +94,29 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
 void *
 util_make_fragment_tex_shader(struct pipe_context *pipe)
 {
-   struct pipe_shader_state shader;
-   struct tgsi_token tokens[100];
-   struct tgsi_header *header;
-   struct tgsi_processor *processor;
-   struct tgsi_full_declaration decl;
-   struct tgsi_full_instruction inst;
-   const uint procType = TGSI_PROCESSOR_FRAGMENT;
-   uint ti;
-
-   /* shader header
-    */
-   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
-
-   header = (struct tgsi_header *) &tokens[1];
-   *header = tgsi_build_header();
-
-   processor = (struct tgsi_processor *) &tokens[2];
-   *processor = tgsi_build_processor( procType, header );
-
-   ti = 3;
-
-   /* declare TEX[0] input */
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_INPUT;
-   /* XXX this could be linear... */
-   decl.Declaration.Interpolate = TGSI_INTERPOLATE_PERSPECTIVE;
-   decl.Declaration.Semantic = 1;
-   decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC;
-   decl.Semantic.SemanticIndex = 0;
-   decl.DeclarationRange.First = 
-   decl.DeclarationRange.Last = 0;
-   ti += tgsi_build_full_declaration(&decl,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti);
-
-   /* declare color[0] output */
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_OUTPUT;
-   decl.Declaration.Semantic = 1;
-   decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR;
-   decl.Semantic.SemanticIndex = 0;
-   decl.DeclarationRange.First = 
-   decl.DeclarationRange.Last = 0;
-   ti += tgsi_build_full_declaration(&decl,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti);
-
-   /* declare sampler */
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_SAMPLER;
-   decl.DeclarationRange.First = 
-   decl.DeclarationRange.Last = 0;
-   ti += tgsi_build_full_declaration(&decl,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti);
-
-   /* TEX instruction */
-   inst = tgsi_default_full_instruction();
-   inst.Instruction.Opcode = TGSI_OPCODE_TEX;
-   inst.Instruction.NumDstRegs = 1;
-   inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
-   inst.FullDstRegisters[0].DstRegister.Index = 0;
-   inst.Instruction.NumSrcRegs = 2;
-   inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D;
-   inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
-   inst.FullSrcRegisters[0].SrcRegister.Index = 0;
-   inst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
-   inst.FullSrcRegisters[1].SrcRegister.Index = 0;
-   ti += tgsi_build_full_instruction(&inst,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti );
-
-   /* END instruction */
-   inst = tgsi_default_full_instruction();
-   inst.Instruction.Opcode = TGSI_OPCODE_END;
-   inst.Instruction.NumDstRegs = 0;
-   inst.Instruction.NumSrcRegs = 0;
-   ti += tgsi_build_full_instruction(&inst,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti );
-
-#if 0 /*debug*/
-   tgsi_dump(tokens, 0);
-#endif
-
-   shader.tokens = tokens;
-
-   return pipe->create_fs_state(pipe, &shader);
+   struct ureg_program *ureg;
+   struct ureg_src sampler;
+   struct ureg_src tex;
+   struct ureg_dst out;
+
+   ureg = ureg_create( pipe, TGSI_PROCESSOR_FRAGMENT );
+   if (ureg == NULL)
+      return NULL;
+   
+   sampler = ureg_DECL_sampler( ureg );
+
+   tex = ureg_DECL_fs_input( ureg, 
+                             TGSI_SEMANTIC_GENERIC, 0, 
+                             TGSI_INTERPOLATE_PERSPECTIVE );
+
+   out = ureg_DECL_output( ureg, 
+                           TGSI_SEMANTIC_COLOR,
+                           0 );
+
+   ureg_TEX( ureg, out, TGSI_TEXTURE_2D, tex, sampler );
+   ureg_END( ureg );
+
+   return ureg_create_shader_and_destroy( ureg );
 }
 
 
@@ -263,87 +129,23 @@ util_make_fragment_tex_shader(struct pipe_context *pipe)
 void *
 util_make_fragment_passthrough_shader(struct pipe_context *pipe)
 {
-   struct pipe_shader_state shader;
-   struct tgsi_token tokens[40];
-   struct tgsi_header *header;
-   struct tgsi_processor *processor;
-   struct tgsi_full_declaration decl;
-   struct tgsi_full_instruction inst;
-   const uint procType = TGSI_PROCESSOR_FRAGMENT;
-   uint ti;
-
-   /* shader header
-    */
-   *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
-
-   header = (struct tgsi_header *) &tokens[1];
-   *header = tgsi_build_header();
-
-   processor = (struct tgsi_processor *) &tokens[2];
-   *processor = tgsi_build_processor( procType, header );
-
-   ti = 3;
-
-   /* declare input */
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_INPUT;
-   decl.Declaration.Semantic = 1;
-   decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR;
-   decl.Semantic.SemanticIndex = 0;
-   decl.DeclarationRange.First = 
-      decl.DeclarationRange.Last = 0;
-   ti += tgsi_build_full_declaration(&decl,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti);
-
-   /* declare output */
-   decl = tgsi_default_full_declaration();
-   decl.Declaration.File = TGSI_FILE_OUTPUT;
-   decl.Declaration.Semantic = 1;
-   decl.Semantic.SemanticName = TGSI_SEMANTIC_COLOR;
-   decl.Semantic.SemanticIndex = 0;
-   decl.DeclarationRange.First = 
-      decl.DeclarationRange.Last = 0;
-   ti += tgsi_build_full_declaration(&decl,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti);
-
-
-   /* MOVE out[0], in[0]; */
-   inst = tgsi_default_full_instruction();
-   inst.Instruction.Opcode = TGSI_OPCODE_MOV;
-   inst.Instruction.NumDstRegs = 1;
-   inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
-   inst.FullDstRegisters[0].DstRegister.Index = 0;
-   inst.Instruction.NumSrcRegs = 1;
-   inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
-   inst.FullSrcRegisters[0].SrcRegister.Index = 0;
-   ti += tgsi_build_full_instruction(&inst,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti );
-
-   /* END instruction */
-   inst = tgsi_default_full_instruction();
-   inst.Instruction.Opcode = TGSI_OPCODE_END;
-   inst.Instruction.NumDstRegs = 0;
-   inst.Instruction.NumSrcRegs = 0;
-   ti += tgsi_build_full_instruction(&inst,
-                                     &tokens[ti],
-                                     header,
-                                     Elements(tokens) - ti );
-
-   assert(ti < Elements(tokens));
-
-#if 0 /*debug*/
-   tgsi_dump(tokens, 0);
-#endif
-
-   shader.tokens = tokens;
-
-   return pipe->create_fs_state(pipe, &shader);
+   struct ureg_program *ureg;
+   struct ureg_src src;
+   struct ureg_dst dst;
+
+   ureg = ureg_create( pipe, TGSI_PROCESSOR_FRAGMENT );
+   if (ureg == NULL)
+      return NULL;
+
+   src = ureg_DECL_fs_input( ureg, TGSI_SEMANTIC_COLOR, 0, 
+                             TGSI_INTERPOLATE_PERSPECTIVE );
+
+   dst = ureg_DECL_output( ureg, TGSI_SEMANTIC_COLOR, 0 );
+
+   ureg_MOV( ureg, dst, src );
+   ureg_END( ureg );
+
+   return ureg_create_shader_and_destroy( ureg );
 }
 
 
diff --git a/src/gallium/auxiliary/util/u_surface.h b/src/gallium/auxiliary/util/u_surface.h
index a5b73cfc20a..ce84ed7ad06 100644
--- a/src/gallium/auxiliary/util/u_surface.h
+++ b/src/gallium/auxiliary/util/u_surface.h
@@ -37,6 +37,23 @@ struct pipe_texture;
 struct pipe_surface;
 
 
+/**
+ * Are s1 and s2 the same surface?
+ * Surfaces are basically views into textures so check if the two surfaces
+ * name the same part of the same texture.
+ */
+static INLINE boolean
+util_same_surface(const struct pipe_surface *s1, const struct pipe_surface *s2)
+{
+   return (s1->texture == s2->texture &&
+           s1->face == s2->face &&
+           s1->level == s2->level &&
+           s1->zslice == s2->zslice);
+}
+
+
+
+
 extern boolean
 util_create_rgba_surface(struct pipe_screen *screen,
                          uint width, uint height,
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index a0c8ed88f74..422bc76003a 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -153,7 +153,7 @@ a8r8g8b8_put_tile_rgba(unsigned *dst,
 }
 
 
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
+/*** PIPE_FORMAT_X8R8G8B8_UNORM ***/
 
 static void
 x8r8g8b8_get_tile_rgba(const unsigned *src,
diff --git a/src/gallium/auxiliary/util/u_time.c b/src/gallium/auxiliary/util/u_time.c
index 5268cbf79ce..c16cdd0b226 100644
--- a/src/gallium/auxiliary/util/u_time.c
+++ b/src/gallium/auxiliary/util/u_time.c
@@ -35,7 +35,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
 #include <sys/time.h>
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
 #include <windows.h>
@@ -77,7 +77,7 @@ util_time_get_frequency(void)
 void 
 util_time_get(struct util_time *t)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    gettimeofday(&t->tv, NULL);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
    LONGLONG temp;
@@ -102,7 +102,7 @@ util_time_add(const struct util_time *t1,
               int64_t usecs,
               struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    t2->tv.tv_sec = t1->tv.tv_sec + usecs / 1000000;
    t2->tv.tv_usec = t1->tv.tv_usec + usecs % 1000000;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
@@ -124,7 +124,7 @@ int64_t
 util_time_diff(const struct util_time *t1, 
                const struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    return (t2->tv.tv_usec - t1->tv.tv_usec) + 
           (t2->tv.tv_sec - t1->tv.tv_sec)*1000000;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
@@ -144,7 +144,7 @@ util_time_micros( void )
    
    util_time_get(&t1);
    
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    return t1.tv.tv_usec + t1.tv.tv_sec*1000000LL;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_USER) || defined(PIPE_SUBSYSTEM_WINDOWS_CE)
    util_time_get_frequency();
@@ -166,7 +166,7 @@ static INLINE int
 util_time_compare(const struct util_time *t1, 
                   const struct util_time *t2)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    if (t1->tv.tv_sec < t2->tv.tv_sec)
       return -1;
    else if(t1->tv.tv_sec > t2->tv.tv_sec)
diff --git a/src/gallium/auxiliary/util/u_time.h b/src/gallium/auxiliary/util/u_time.h
index 6bca6077a2a..7a5c54d9b23 100644
--- a/src/gallium/auxiliary/util/u_time.h
+++ b/src/gallium/auxiliary/util/u_time.h
@@ -38,7 +38,7 @@
 
 #include "pipe/p_config.h"
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
 #include <time.h> /* timeval */
 #include <unistd.h> /* usleep */
 #endif
@@ -58,7 +58,7 @@ extern "C" {
  */
 struct util_time 
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
    struct timeval tv;
 #else
    int64_t counter;
@@ -89,7 +89,7 @@ util_time_timeout(const struct util_time *start,
                   const struct util_time *end,
                   const struct util_time *curr);
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE)
 #define util_time_sleep usleep
 #else
 void
diff --git a/src/gallium/auxiliary/util/u_timed_winsys.c b/src/gallium/auxiliary/util/u_timed_winsys.c
index 77b2a3a1c87..178acdca4df 100644
--- a/src/gallium/auxiliary/util/u_timed_winsys.c
+++ b/src/gallium/auxiliary/util/u_timed_winsys.c
@@ -212,13 +212,14 @@ timed_surface_buffer_create(struct pipe_winsys *winsys,
                               unsigned width, unsigned height,
                               enum pipe_format format, 
                               unsigned usage,
+                              unsigned tex_usage,
                               unsigned *stride)
 {
    struct pipe_winsys *backend = timed_winsys(winsys)->backend;
    uint64_t start = time_start();
 
    struct pipe_buffer *ret = backend->surface_buffer_create( backend, width, height, 
-                                                             format, usage, stride );
+                                                             format, usage, tex_usage, stride );
 
    time_finish(winsys, start, 7, __FUNCTION__);
    
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
index 5a889a6119d..58a8b5d0b0f 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
@@ -1834,9 +1834,9 @@ emit_instruction(struct codegen *gen,
    case TGSI_OPCODE_ENDIF:
       return emit_ENDIF(gen, inst);
 
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       return emit_BGNLOOP(gen, inst);
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       return emit_ENDLOOP(gen, inst);
    case TGSI_OPCODE_BRK:
       return emit_BRK(gen, inst);
@@ -1875,9 +1875,9 @@ emit_immediate(struct codegen *gen, const struct tgsi_full_immediate *immed)
    assert(gen->num_imm < MAX_TEMPS);
 
    for (ch = 0; ch < 4; ch++) {
-      float val = immed->u.ImmediateFloat32[ch].Float;
+      float val = immed->u[ch].Float;
 
-      if (ch > 0 && val == immed->u.ImmediateFloat32[ch - 1].Float) {
+      if (ch > 0 && val == immed->u[ch - 1].Float) {
          /* re-use previous register */
          gen->imm_regs[gen->num_imm][ch] = gen->imm_regs[gen->num_imm][ch - 1];
       }
diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c
index e26594448f0..6a63a0e6ced 100644
--- a/src/gallium/drivers/cell/ppu/cell_texture.c
+++ b/src/gallium/drivers/cell/ppu/cell_texture.c
@@ -44,13 +44,6 @@
 
 
 
-static unsigned
-minify(unsigned d)
-{
-   return MAX2(1, d>>1);
-}
-
-
 static void
 cell_texture_layout(struct cell_texture *ct)
 {
@@ -424,7 +417,8 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer)
    if (!ctrans->map)
       return NULL; /* out of memory */
 
-   if (transfer->usage & PIPE_TRANSFER_READ) {
+   if (transfer->usage == PIPE_TRANSFER_READ ||
+       transfer->usage == PIPE_TRANSFER_READ_WRITE) {
       /* need to untwiddle the texture to make a linear version */
       const uint bpp = pf_get_size(ct->base.format);
       if (bpp == 4) {
@@ -465,7 +459,8 @@ cell_transfer_unmap(struct pipe_screen *screen,
                                    PIPE_BUFFER_USAGE_CPU_READ);
    }
 
-   if (transfer->usage & PIPE_TRANSFER_WRITE) {
+   if (transfer->usage == PIPE_TRANSFER_WRITE ||
+       transfer->usage == PIPE_TRANSFER_READ_WRITE) {
       /* The user wrote new texture data into the mapped buffer.
        * We need to convert the new linear data into the twiddled/tiled format.
        */
diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c
index e27df2dfb38..6db9501128c 100644
--- a/src/gallium/drivers/cell/spu/spu_exec.c
+++ b/src/gallium/drivers/cell/spu/spu_exec.c
@@ -952,7 +952,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_RCP:
-   /* TGSI_OPCODE_RECIP */
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q);
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -961,7 +960,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_RSQ:
-   /* TGSI_OPCODE_RECIPSQRT */
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_sqrt(r[0].q);
       r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q);
@@ -1115,7 +1113,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_MAD:
-   /* TGSI_OPCODE_MADD */
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
@@ -1136,8 +1133,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LERP:
-   /* TGSI_OPCODE_LRP */
+   case TGSI_OPCODE_LRP:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
@@ -1158,21 +1154,11 @@ exec_instruction(
       ASSERT (0);
       break;
 
-   case TGSI_OPCODE_DOT2ADD:
-      /* TGSI_OPCODE_DP2A */
+   case TGSI_OPCODE_DP2A:
       ASSERT (0);
       break;
 
-   case TGSI_OPCODE_INDEX:
-      ASSERT (0);
-      break;
-
-   case TGSI_OPCODE_NEGATE:
-      ASSERT (0);
-      break;
-
-   case TGSI_OPCODE_FRAC:
-   /* TGSI_OPCODE_FRC */
+   case TGSI_OPCODE_FRC:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          r[0].q = micro_frc(r[0].q);
@@ -1184,8 +1170,7 @@ exec_instruction(
       ASSERT (0);
       break;
 
-   case TGSI_OPCODE_FLOOR:
-   /* TGSI_OPCODE_FLR */
+   case TGSI_OPCODE_FLR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          r[0].q = micro_flr(r[0].q);
@@ -1201,8 +1186,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_EXPBASE2:
-    /* TGSI_OPCODE_EX2 */
+   case TGSI_OPCODE_EX2:
       FETCH(&r[0], 0, CHAN_X);
 
       r[0].q = micro_pow(mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].q, r[0].q);
@@ -1212,8 +1196,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LOGBASE2:
-   /* TGSI_OPCODE_LG2 */
+   case TGSI_OPCODE_LG2:
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_lg2(r[0].q);
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -1221,8 +1204,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_POWER:
-      /* TGSI_OPCODE_POW */
+   case TGSI_OPCODE_POW:
       FETCH(&r[0], 0, CHAN_X);
       FETCH(&r[1], 1, CHAN_X);
 
@@ -1233,7 +1215,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_CROSSPRODUCT:
+   case TGSI_OPCODE_XPD:
       /* TGSI_OPCODE_XPD */
       FETCH(&r[0], 0, CHAN_Y);
       FETCH(&r[1], 1, CHAN_Z);
@@ -1275,10 +1257,6 @@ exec_instruction(
       }
       break;
 
-    case TGSI_OPCODE_MULTIPLYMATRIX:
-       ASSERT (0);
-       break;
-
     case TGSI_OPCODE_ABS:
        FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
           FETCH(&r[0], 0, chan_index);
@@ -1780,9 +1758,9 @@ exec_instruction(
       mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       /* fall-through (for now) */
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
       ASSERT(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
@@ -1790,9 +1768,9 @@ exec_instruction(
       mach->ContStack[mach->ContStackTop++] = mach->ContMask;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       /* fall-through (for now at least) */
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       /* Restore ContMask, but don't pop */
       ASSERT(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
diff --git a/src/gallium/drivers/i915simple/i915_batch.h b/src/gallium/drivers/i915simple/i915_batch.h
index a433cf054de..c6e68ea38a2 100644
--- a/src/gallium/drivers/i915simple/i915_batch.h
+++ b/src/gallium/drivers/i915simple/i915_batch.h
@@ -50,8 +50,8 @@ i915_batchbuffer_check( struct i915_batchbuffer *batch,
 			size_t dwords,
 			size_t relocs )
 {
-   /** TODO JB: Check relocs */
-   return dwords * 4 <= batch->size - (batch->ptr - batch->map);
+   return dwords * 4 <= batch->size - (batch->ptr - batch->map) &&
+          relocs <= (batch->max_relocs - batch->relocs);
 }
 
 static INLINE size_t
diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915simple/i915_context.c
index ccf9bb31fb0..bf69c8e9f53 100644
--- a/src/gallium/drivers/i915simple/i915_context.c
+++ b/src/gallium/drivers/i915simple/i915_context.c
@@ -142,10 +142,14 @@ i915_is_texture_referenced( struct pipe_context *pipe,
 			    unsigned face, unsigned level)
 {
    /**
-    * FIXME: Optimize.
+    * FIXME: Return the corrent result. We can't alays return referenced
+    *        since it causes a double flush within the vbo module.
     */
-
+#if 0
    return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
+#else
+   return 0;
+#endif
 }
 
 static unsigned int
@@ -153,10 +157,14 @@ i915_is_buffer_referenced( struct pipe_context *pipe,
 			   struct pipe_buffer *buf)
 {
    /**
-    * FIXME: Optimize.
+    * FIXME: Return the corrent result. We can't alays return referenced
+    *        since it causes a double flush within the vbo module.
     */
-
+#if 0
    return PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE;
+#else
+   return 0;
+#endif
 }
 
 
diff --git a/src/gallium/drivers/i915simple/i915_fpc_translate.c b/src/gallium/drivers/i915simple/i915_fpc_translate.c
index 961c1bf2134..89504ced276 100644
--- a/src/gallium/drivers/i915simple/i915_fpc_translate.c
+++ b/src/gallium/drivers/i915simple/i915_fpc_translate.c
@@ -975,8 +975,9 @@ i915_translate_instructions(struct i915_fp_compile *p,
                = &parse.FullToken.FullImmediate;
             const uint pos = p->num_immediates++;
             uint j;
+            assert( imm->Immediate.NrTokens <= 4 + 1 );
             for (j = 0; j < imm->Immediate.NrTokens - 1; j++) {
-               p->