[Mesa-dev] [PATCH V4 10/13] i965: Emit a second set of SURFACE_STATE for gather4 from textures.

[Chris Forbes]

This allows us to use a different surface format for gather4, which is
required for R32G32_FLOAT to work on Gen7.

V4: - Only emit alternate surface state for shaders which will actually
      use it.
    - Pass a simple 'for_gather' flag rather than a function pointer.
      The callee can decide what w/a to apply.

Signed-off-by: Chris Forbes <chrisf at ijw.co.nz>
---
 src/mesa/drivers/dri/i965/brw_context.h           |  3 +-
 src/mesa/drivers/dri/i965/brw_wm_surface_state.c  | 38 +++++++++++++++++++----
 src/mesa/drivers/dri/i965/gen7_wm_surface_state.c |  6 +++-
 3 files changed, 39 insertions(+), 8 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 3f2f4ea..6e2edc9 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -928,7 +928,8 @@ struct brw_context
 
       void (*update_texture_surface)(struct gl_context *ctx,
                                      unsigned unit,
-                                     uint32_t *surf_offset);
+                                     uint32_t *surf_offset,
+                                     bool for_gather);
       void (*update_renderbuffer_surface)(struct brw_context *brw,
 					  struct gl_renderbuffer *rb,
 					  bool layered,
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index 471fd03..89827c4 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -251,7 +251,8 @@ brw_update_buffer_texture_surface(struct gl_context *ctx,
 static void
 brw_update_texture_surface(struct gl_context *ctx,
                            unsigned unit,
-                           uint32_t *surf_offset)
+                           uint32_t *surf_offset,
+                           bool for_gather)
 {
    struct brw_context *brw = brw_context(ctx);
    struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
@@ -270,6 +271,8 @@ brw_update_texture_surface(struct gl_context *ctx,
    surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE,
 			  6 * 4, 32, surf_offset);
 
+   (void) for_gather;   /* no w/a to apply for this gen */
+
    surf[0] = (translate_tex_target(tObj->Target) << BRW_SURFACE_TYPE_SHIFT |
 	      BRW_SURFACE_MIPMAPLAYOUT_BELOW << BRW_SURFACE_MIPLAYOUT_SHIFT |
 	      BRW_SURFACE_CUBEFACE_ENABLES |
@@ -713,7 +716,8 @@ const struct brw_tracked_state gen6_renderbuffer_surfaces = {
 static void
 update_stage_texture_surfaces(struct brw_context *brw,
                               const struct gl_program *prog,
-                              uint32_t *surf_offset)
+                              uint32_t *surf_offset,
+                              bool for_gather)
 {
    if (!prog)
       return;
@@ -730,7 +734,7 @@ update_stage_texture_surfaces(struct brw_context *brw,
 
          /* _NEW_TEXTURE */
          if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-            brw->vtbl.update_texture_surface(ctx, unit, surf_offset + s);
+            brw->vtbl.update_texture_surface(ctx, unit, surf_offset + s, for_gather);
          }
       }
    }
@@ -755,13 +759,35 @@ brw_update_texture_surfaces(struct brw_context *brw)
    /* _NEW_TEXTURE */
    update_stage_texture_surfaces(brw, vs,
                                  brw->vs.base.surf_offset +
-                                 SURF_INDEX_VEC4_TEXTURE(0));
+                                 SURF_INDEX_VEC4_TEXTURE(0),
+                                 false);
    update_stage_texture_surfaces(brw, gs,
                                  brw->gs.base.surf_offset +
-                                 SURF_INDEX_VEC4_TEXTURE(0));
+                                 SURF_INDEX_VEC4_TEXTURE(0),
+                                 false);
    update_stage_texture_surfaces(brw, fs,
                                  brw->wm.base.surf_offset +
-                                 SURF_INDEX_TEXTURE(0));
+                                 SURF_INDEX_TEXTURE(0),
+                                 false);
+
+   /* emit alternate set of surface state for gather. this
+    * allows the surface format to be overriden for only the
+    * gather4 messages. */
+   if (vs && vs->UsesGather)
+      update_stage_texture_surfaces(brw, vs,
+                                    brw->vs.base.surf_offset +
+                                    SURF_INDEX_VEC4_GATHER_TEXTURE(0),
+                                    true);
+   if (gs && gs->UsesGather)
+      update_stage_texture_surfaces(brw, gs,
+                                    brw->gs.base.surf_offset +
+                                    SURF_INDEX_VEC4_GATHER_TEXTURE(0),
+                                    true);
+   if (fs && fs->UsesGather)
+      update_stage_texture_surfaces(brw, fs,
+                                    brw->wm.base.surf_offset +
+                                    SURF_INDEX_GATHER_TEXTURE(0),
+                                    true);
 
    brw->state.dirty.brw |= BRW_NEW_SURFACES;
 }
diff --git a/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c b/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
index 7571cbf..fc1613f 100644
--- a/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_wm_surface_state.c
@@ -299,7 +299,8 @@ gen7_update_buffer_texture_surface(struct gl_context *ctx,
 static void
 gen7_update_texture_surface(struct gl_context *ctx,
                             unsigned unit,
-                            uint32_t *surf_offset)
+                            uint32_t *surf_offset,
+                            bool for_gather)
 {
    struct brw_context *brw = brw_context(ctx);
    struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
@@ -323,6 +324,9 @@ gen7_update_texture_surface(struct gl_context *ctx,
                                               tObj->DepthMode,
                                               sampler->sRGBDecode);
 
+   if (for_gather && tex_format == BRW_SURFACEFORMAT_R32G32_FLOAT)
+      tex_format = BRW_SURFACEFORMAT_R32G32_FLOAT_LD;
+
    surf[0] = translate_tex_target(tObj->Target) << BRW_SURFACE_TYPE_SHIFT |
              tex_format << BRW_SURFACE_FORMAT_SHIFT |
              gen7_surface_tiling_mode(mt->region->tiling) |
-- 
1.8.4