[Mesa-dev] [PATCH 35/37] i965/gen6/gs: Use a specific implementation of geometry shaders for gen6.

[Iago Toral Quiroga]

In gen6 we will use the geometry shader implementation from gen6_gs_visitor.cpp
and keep the implementation in brw_vec4_gs_visitor.cpp for gen7+. Notice that
gen6_gs_visitor inherits from brw_vec4_gs_visitor so it is not a completely
seprate implementation of geometry shaders.

Also, gen6 does not support multiple dispatch modes, its default operation mode
is equivalent to gen7's SINGLE mode, so select that in gen6 for consistency.
---
 src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp | 56 ++++++++++++++---------
 1 file changed, 35 insertions(+), 21 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp
index c2a4892..d9f658e 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_gs_visitor.cpp
@@ -28,6 +28,7 @@
  */
 
 #include "brw_vec4_gs_visitor.h"
+#include "gen6_gs_visitor.h"
 
 const unsigned MAX_GS_INPUT_VERTICES = 6;
 
@@ -634,19 +635,21 @@ brw_gs_emit(struct brw_context *brw,
       brw_dump_ir(brw, "geometry", prog, &shader->base, NULL);
    }
 
-   /* Compile the geometry shader in DUAL_OBJECT dispatch mode, if we can do
-    * so without spilling. If the GS invocations count > 1, then we can't use
-    * dual object mode.
-    */
-   if (c->prog_data.invocations <= 1 &&
-       likely(!(INTEL_DEBUG & DEBUG_NO_DUAL_OBJECT_GS))) {
-      c->prog_data.dispatch_mode = GEN7_GS_DISPATCH_MODE_DUAL_OBJECT;
-
-      vec4_gs_visitor v(brw, c, prog, mem_ctx, true /* no_spills */);
-      if (v.run()) {
-         return generate_assembly(brw, prog, &c->gp->program.Base,
-                                  &c->prog_data.base, mem_ctx, &v.instructions,
-                                  final_assembly_size);
+   if (brw->gen >= 7) {
+      /* Compile the geometry shader in DUAL_OBJECT dispatch mode, if we can do
+       * so without spilling. If the GS invocations count > 1, then we can't use
+       * dual object mode.
+       */
+      if (c->prog_data.invocations <= 1 &&
+          likely(!(INTEL_DEBUG & DEBUG_NO_DUAL_OBJECT_GS))) {
+         c->prog_data.dispatch_mode = GEN7_GS_DISPATCH_MODE_DUAL_OBJECT;
+
+         vec4_gs_visitor v(brw, c, prog, mem_ctx, true /* no_spills */);
+         if (v.run()) {
+            return generate_assembly(brw, prog, &c->gp->program.Base,
+                                     &c->prog_data.base, mem_ctx,
+                                     &v.instructions, final_assembly_size);
+         }
       }
    }
 
@@ -655,22 +658,33 @@ brw_gs_emit(struct brw_context *brw,
     * back to DUAL_INSTANCED or SINGLE mode, which consumes fewer registers.
     *
     * SINGLE mode is more performant when invocations == 1 and DUAL_INSTANCE
-    * mode is more performant when invocations > 1.
+    * mode is more performant when invocations > 1. Gen6 only supports
+    * SINGLE mode.
     */
-   if (c->prog_data.invocations <= 1)
+   if (c->prog_data.invocations <= 1 || brw->gen < 7)
       c->prog_data.dispatch_mode = GEN7_GS_DISPATCH_MODE_SINGLE;
    else
       c->prog_data.dispatch_mode = GEN7_GS_DISPATCH_MODE_DUAL_INSTANCE;
 
-   vec4_gs_visitor v(brw, c, prog, mem_ctx, false /* no_spills */);
-   if (!v.run()) {
+   vec4_gs_visitor *gs = NULL;
+   const unsigned *ret = NULL;
+
+   if (brw->gen >= 7)
+      gs = new vec4_gs_visitor(brw, c, prog, mem_ctx, false /* no_spills */);
+   else
+      gs = new gen6_gs_visitor(brw, c, prog, mem_ctx, false /* no_spills */);
+
+   if (!gs->run()) {
       prog->LinkStatus = false;
-      ralloc_strcat(&prog->InfoLog, v.fail_msg);
-      return NULL;
+      ralloc_strcat(&prog->InfoLog, gs->fail_msg);
+   } else {
+      ret = generate_assembly(brw, prog, &c->gp->program.Base,
+                              &c->prog_data.base, mem_ctx, &gs->instructions,
+                              final_assembly_size);
    }
 
-   return generate_assembly(brw, prog, &c->gp->program.Base, &c->prog_data.base,
-                            mem_ctx, &v.instructions, final_assembly_size);
+   delete gs;
+   return ret;
 }
 
 
-- 
1.9.1