gallium: add start_slot parameter to set_vertex_buffers

This allows updating only a subrange of buffer bindings.

set_vertex_buffers(pipe, start_slot, count, NULL) unbinds buffers in that
range. Binding NULL resources unbinds buffers too (both buffer and user_buffer
must be NULL).

The meta ops are adapted to only save, change, and restore the single slot
they use. The cso_context can save and restore only one vertex buffer slot.
The clients can query which one it is using cso_get_aux_vertex_buffer_slot.
It's currently set to 0. (the Draw module breaks if it's set to non-zero)

It should decrease the CPU overhead when using a lot of meta ops, but
the drivers must be able to treat each vertex buffer slot as a separate
state (only r600g does so at the moment).

I can imagine this also being useful for optimizing some OpenGL use cases.

Reviewed-by: Brian Paul <brianp@vmware.com>

4 files changed, 20 insertions, 16 deletions

diff --git a/src/gallium/drivers/nv30/nv30_draw.c b/src/gallium/drivers/nv30/nv30_draw.c
index 46e85362815..51fe4f86ed1 100644
--- a/src/gallium/drivers/nv30/nv30_draw.c
+++ b/src/gallium/drivers/nv30/nv30_draw.c
@@ -379,7 +379,7 @@ nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    if (nv30->draw_dirty & NV30_NEW_CLIP)
       draw_set_clip_state(draw, &nv30->clip);
    if (nv30->draw_dirty & NV30_NEW_ARRAYS) {
-      draw_set_vertex_buffers(draw, nv30->num_vtxbufs, nv30->vtxbuf);
+      draw_set_vertex_buffers(draw, 0, nv30->num_vtxbufs, nv30->vtxbuf);
       draw_set_vertex_elements(draw, nv30->vertex->num_elements, nv30->vertex->pipe);
    }
    if (nv30->draw_dirty & NV30_NEW_FRAGPROG) {
@@ -404,10 +404,14 @@ nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
 
    for (i = 0; i < nv30->num_vtxbufs; i++) {
       const void *map = nv30->vtxbuf[i].user_buffer;
-      if (!map)
+      if (!map) {
+         if (!nv30->vtxbuf[i].buffer) {
+            continue;
+         }
          map = pipe_buffer_map(pipe, nv30->vtxbuf[i].buffer,
                                   PIPE_TRANSFER_UNSYNCHRONIZED |
                                   PIPE_TRANSFER_READ, &transfer[i]);
+      }
       draw_set_mapped_vertex_buffer(draw, i, map);
    }
 
diff --git a/src/gallium/drivers/nv30/nv30_miptree.c b/src/gallium/drivers/nv30/nv30_miptree.c
index e89725feb2f..5d941ab771e 100644
--- a/src/gallium/drivers/nv30/nv30_miptree.c
+++ b/src/gallium/drivers/nv30/nv30_miptree.c
@@ -193,9 +193,7 @@ nv30_blit(struct pipe_context *pipe,
 
    /* XXX turn off occlusion queries */
 
-   util_blitter_save_vertex_buffers(nv30->blitter,
-                                    nv30->num_vtxbufs,
-                                    nv30->vtxbuf);
+   util_blitter_save_vertex_buffer_slot(nv30->blitter, nv30->vtxbuf);
    util_blitter_save_vertex_elements(nv30->blitter, nv30->vertex);
    util_blitter_save_vertex_shader(nv30->blitter, nv30->vertprog.program);
    util_blitter_save_rasterizer(nv30->blitter, nv30->rast);
diff --git a/src/gallium/drivers/nv30/nv30_push.c b/src/gallium/drivers/nv30/nv30_push.c
index c5867bfa11a..5ec2d462a68 100644
--- a/src/gallium/drivers/nv30/nv30_push.c
+++ b/src/gallium/drivers/nv30/nv30_push.c
@@ -211,6 +211,10 @@ nv30_push_vbo(struct nv30_context *nv30, const struct pipe_draw_info *info)
       struct pipe_vertex_buffer *vb = &nv30->vtxbuf[i];
       struct nv04_resource *res = nv04_resource(vb->buffer);
 
+      if (!vb->buffer && !vb->user_buffer) {
+         continue;
+      }
+
       data = nouveau_resource_map_offset(&nv30->base, res,
                                          vb->buffer_offset, NOUVEAU_BO_RD);
 
@@ -276,8 +280,11 @@ nv30_push_vbo(struct nv30_context *nv30, const struct pipe_draw_info *info)
    if (info->indexed)
       nouveau_resource_unmap(nv04_resource(nv30->idxbuf.buffer));
 
-   for (i = 0; i < nv30->num_vtxbufs; ++i)
-      nouveau_resource_unmap(nv04_resource(nv30->vtxbuf[i].buffer));
+   for (i = 0; i < nv30->num_vtxbufs; ++i) {
+      if (nv30->vtxbuf[i].buffer) {
+         nouveau_resource_unmap(nv04_resource(nv30->vtxbuf[i].buffer));
+      }
+   }
 
    nv30_state_release(nv30);
 }
diff --git a/src/gallium/drivers/nv30/nv30_state.c b/src/gallium/drivers/nv30/nv30_state.c
index 046ba723dae..af8bb449fce 100644
--- a/src/gallium/drivers/nv30/nv30_state.c
+++ b/src/gallium/drivers/nv30/nv30_state.c
@@ -23,6 +23,7 @@
  *
  */
 
+#include "util/u_helpers.h"
 #include "util/u_inlines.h"
 
 #include "nouveau/nouveau_gldefs.h"
@@ -393,21 +394,15 @@ nv30_set_viewport_state(struct pipe_context *pipe,
 
 static void
 nv30_set_vertex_buffers(struct pipe_context *pipe,
-                        unsigned count,
+                        unsigned start_slot, unsigned count,
                         const struct pipe_vertex_buffer *vb)
 {
     struct nv30_context *nv30 = nv30_context(pipe);
-    unsigned i;
 
     nouveau_bufctx_reset(nv30->bufctx, BUFCTX_VTXBUF);
 
-    for (i = 0; i < count; ++i)
-       pipe_resource_reference(&nv30->vtxbuf[i].buffer, vb[i].buffer);
-    for (; i < nv30->num_vtxbufs; ++i)
-       pipe_resource_reference(&nv30->vtxbuf[i].buffer, NULL);
-
-    memcpy(nv30->vtxbuf, vb, sizeof(*vb) * count);
-    nv30->num_vtxbufs = count;
+    util_set_vertex_buffers_count(nv30->vtxbuf, &nv30->num_vtxbufs,
+                                  vb, start_slot, count);
 
     nv30->dirty |= NV30_NEW_ARRAYS;
 }