gallium: add PIPE_SHADER_CAP_TGSI_SKIP_MERGE_REGISTERS

[mesa.git] / src / gallium / drivers / nouveau / nv50 / nv50_vbo.c

diff --git a/src/gallium/drivers/nouveau/nv50/nv50_vbo.c b/src/gallium/drivers/nouveau/nv50/nv50_vbo.c
index 85878d5fcc70875838e7009bc641b3c80f92ae5f..227038e95a5c68ee68763c3400bf13b8b9dbc38e 100644 (file)
--- a/src/gallium/drivers/nouveau/nv50/nv50_vbo.c
+++ b/src/gallium/drivers/nouveau/nv50/nv50_vbo.c
@@ -76,7 +76,7 @@ nv50_vertex_state_create(struct pipe_context *pipe,
         enum pipe_format fmt = ve->src_format;
 
         so->element[i].pipe = elements[i];
-        so->element[i].state = nv50_format_table[fmt].vtx;
+        so->element[i].state = nv50_vertex_format[fmt].vtx;
 
         if (!so->element[i].state) {
             switch (util_format_get_nr_components(fmt)) {
@@ -89,8 +89,11 @@ nv50_vertex_state_create(struct pipe_context *pipe,
                 FREE(so);
                 return NULL;
             }
-            so->element[i].state = nv50_format_table[fmt].vtx;
+            so->element[i].state = nv50_vertex_format[fmt].vtx;
             so->need_conversion = true;
+            pipe_debug_message(&nouveau_context(pipe)->debug, FALLBACK,
+                               "Converting vertex element %d, no hw format %s",
+                               i, util_format_name(ve->src_format));
         }
         so->element[i].state |= i;
 
@@ -227,7 +230,7 @@ nv50_upload_user_buffers(struct nv50_context *nv50,
       addrs[b] = nouveau_scratch_data(&nv50->base, vb->user_buffer, base, size,
                                       &bo);
       if (addrs[b])
-         BCTX_REFN_bo(nv50->bufctx_3d, VERTEX_TMP, NOUVEAU_BO_GART |
+         BCTX_REFN_bo(nv50->bufctx_3d, 3D_VERTEX_TMP, NOUVEAU_BO_GART |
                       NOUVEAU_BO_RD, bo);
    }
    nv50->base.vbo_dirty = true;
@@ -266,7 +269,7 @@ nv50_update_user_vbufs(struct nv50_context *nv50)
          address[b] = nouveau_scratch_data(&nv50->base, vb->user_buffer,
                                            base, size, &bo);
          if (address[b])
-            BCTX_REFN_bo(nv50->bufctx_3d, VERTEX_TMP, bo_flags, bo);
+            BCTX_REFN_bo(nv50->bufctx_3d, 3D_VERTEX_TMP, bo_flags, bo);
       }
 
       BEGIN_NV04(push, NV50_3D(VERTEX_ARRAY_LIMIT_HIGH(i)), 2);
@@ -283,7 +286,7 @@ static inline void
 nv50_release_user_vbufs(struct nv50_context *nv50)
 {
    if (nv50->vbo_user) {
-      nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_VERTEX_TMP);
+      nouveau_bufctx_reset(nv50->bufctx_3d, NV50_BIND_3D_VERTEX_TMP);
       nouveau_scratch_done(&nv50->base);
    }
 }
@@ -391,7 +394,7 @@ nv50_vertex_arrays_validate(struct nv50_context *nv50)
          struct nv04_resource *buf = nv04_resource(vb->buffer);
          if (!(refd & (1 << b))) {
             refd |= 1 << b;
-            BCTX_REFN(nv50->bufctx_3d, VERTEX, buf, RD);
+            BCTX_REFN(nv50->bufctx_3d, 3D_VERTEX, buf, RD);
          }
          address = buf->address + vb->buffer_offset + ve->pipe.src_offset;
          limit = buf->address + buf->base.width0 - 1;
@@ -633,8 +636,8 @@ nv50_draw_elements(struct nv50_context *nv50, bool shorten,
          BEGIN_NV04(push, NV50_3D(VERTEX_BEGIN_GL), 1);
          PUSH_DATA (push, prim);
 
+         nouveau_pushbuf_space(push, 16, 0, 1);
          PUSH_REFN(push, buf->bo, NOUVEAU_BO_RD | buf->domain);
-         nouveau_pushbuf_space(push, 8, 0, 1);
 
          switch (index_size) {
          case 4:
@@ -699,6 +702,7 @@ nv50_draw_elements(struct nv50_context *nv50, bool shorten,
          prim |= NV50_3D_VERTEX_BEGIN_GL_INSTANCE_NEXT;
       }
    }
+   NOUVEAU_DRV_STAT(&nv50->screen->base, draw_calls_indexed, 1);
 }
 
 static void
@@ -762,7 +766,7 @@ nv50_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    struct nv50_context *nv50 = nv50_context(pipe);
    struct nouveau_pushbuf *push = nv50->base.pushbuf;
    bool tex_dirty = false;
-   int i, s;
+   int s;
 
    /* NOTE: caller must ensure that (min_index + index_bias) is >= 0 */
    nv50->vb_elt_first = info->min_index + info->index_bias;
@@ -776,9 +780,9 @@ nv50_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    nv50->vbo_push_hint = /* the 64 is heuristic */
       !(info->indexed && ((nv50->vb_elt_limit + 64) < info->count));
 
-   if (nv50->vbo_user && !(nv50->dirty & (NV50_NEW_ARRAYS | NV50_NEW_VERTEX))) {
+   if (nv50->vbo_user && !(nv50->dirty_3d & (NV50_NEW_3D_ARRAYS | NV50_NEW_3D_VERTEX))) {
       if (!!nv50->vbo_fifo != nv50->vbo_push_hint)
-         nv50->dirty |= NV50_NEW_ARRAYS;
+         nv50->dirty_3d |= NV50_NEW_3D_ARRAYS;
       else
       if (!nv50->vbo_fifo)
          nv50_update_user_vbufs(nv50);
@@ -787,31 +791,13 @@ nv50_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    if (unlikely(nv50->num_so_targets && !nv50->gmtyprog))
       nv50->state.prim_size = nv50_pipe_prim_to_prim_size[info->mode];
 
-   nv50_state_validate(nv50, ~0, 8); /* 8 as minimum, we use flush_notify */
+   nv50_state_validate_3d(nv50, ~0);
 
    push->kick_notify = nv50_draw_vbo_kick_notify;
 
-   /* TODO: Instead of iterating over all the buffer resources looking for
-    * coherent buffers, keep track of a context-wide count.
-    */
    for (s = 0; s < 3 && !nv50->cb_dirty; ++s) {
-      uint32_t valid = nv50->constbuf_valid[s];
-
-      while (valid && !nv50->cb_dirty) {
-         const unsigned i = ffs(valid) - 1;
-         struct pipe_resource *res;
-
-         valid &= ~(1 << i);
-         if (nv50->constbuf[s][i].user)
-            continue;
-
-         res = nv50->constbuf[s][i].u.buf;
-         if (!res)
-            continue;
-
-         if (res->flags & PIPE_RESOURCE_FLAG_MAP_COHERENT)
-            nv50->cb_dirty = true;
-      }
+      if (nv50->constbuf_coherent[s])
+         nv50->cb_dirty = true;
    }
 
    /* If there are any coherent constbufs, flush the cache */
@@ -822,20 +808,28 @@ nv50_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    }
 
    for (s = 0; s < 3 && !tex_dirty; ++s) {
-      for (i = 0; i < nv50->num_textures[s] && !tex_dirty; ++i) {
-         if (!nv50->textures[s][i] ||
-             nv50->textures[s][i]->texture->target != PIPE_BUFFER)
-            continue;
-         if (nv50->textures[s][i]->texture->flags &
-             PIPE_RESOURCE_FLAG_MAP_COHERENT)
-            tex_dirty = true;
-      }
+      if (nv50->textures_coherent[s])
+         tex_dirty = true;
    }
+
    if (tex_dirty) {
       BEGIN_NV04(push, NV50_3D(TEX_CACHE_CTL), 1);
       PUSH_DATA (push, 0x20);
    }
 
+   if (nv50->screen->base.class_3d >= NVA0_3D_CLASS &&
+       nv50->seamless_cube_map != nv50->state.seamless_cube_map) {
+      nv50->state.seamless_cube_map = nv50->seamless_cube_map;
+      BEGIN_NV04(push, SUBC_3D(NVA0_3D_TEX_MISC), 1);
+      PUSH_DATA (push, nv50->seamless_cube_map ? NVA0_3D_TEX_MISC_SEAMLESS_CUBE_MAP : 0);
+   }
+
+   if (nv50->vertprog->mul_zero_wins != nv50->state.mul_zero_wins) {
+      nv50->state.mul_zero_wins = nv50->vertprog->mul_zero_wins;
+      BEGIN_NV04(push, NV50_3D(UNK1690), 1);
+      PUSH_DATA (push, 0x00010000 * !!nv50->state.mul_zero_wins);
+   }
+
    if (nv50->vbo_fifo) {
       nv50_push_vbo(nv50, info);
       push->kick_notify = nv50_default_kick_notify;
@@ -850,12 +844,7 @@ nv50_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
       PUSH_DATA (push, info->start_instance);
    }
 
-   for (i = 0; i < nv50->num_vtxbufs && !nv50->base.vbo_dirty; ++i) {
-      if (!nv50->vtxbuf[i].buffer)
-         continue;
-      if (nv50->vtxbuf[i].buffer->flags & PIPE_RESOURCE_FLAG_MAP_COHERENT)
-         nv50->base.vbo_dirty = true;
-   }
+   nv50->base.vbo_dirty |= !!nv50->vtxbufs_coherent;
 
    if (nv50->base.vbo_dirty) {
       BEGIN_NV04(push, NV50_3D(VERTEX_ARRAY_FLUSH), 1);