svga: add support for FS sample mask output

This, with the previous work for sample position/id query, allows
us to enable per-sample shading for VGPU 10.1.

Note that quite a few Piglit arb_sample_shading tests still do not
pass, but many do.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c
index 1febe8733765af1f3d5f3bdbd9dae09ec6989b85..b66b560fcc0bfca4935fb1d4f41bbf96430f1d63 100644 (file)
--- a/src/gallium/drivers/svga/svga_screen.c
+++ b/src/gallium/drivers/svga/svga_screen.c
@@ -345,6 +345,8 @@ svga_get_param(struct pipe_screen *screen, enum pipe_cap param)
 
    case PIPE_CAP_CUBE_MAP_ARRAY:
    case PIPE_CAP_INDEP_BLEND_FUNC:
+   case PIPE_CAP_SAMPLE_SHADING:
+   case PIPE_CAP_FORCE_PERSAMPLE_INTERP:
       return sws->have_sm4_1;
 
    /* Unsupported features */
@@ -364,7 +366,6 @@ svga_get_param(struct pipe_screen *screen, enum pipe_cap param)
    case PIPE_CAP_TEXTURE_GATHER_SM5:
    case PIPE_CAP_BUFFER_MAP_PERSISTENT_COHERENT:
    case PIPE_CAP_TEXTURE_QUERY_LOD:
-   case PIPE_CAP_SAMPLE_SHADING:
    case PIPE_CAP_TEXTURE_GATHER_OFFSETS:
    case PIPE_CAP_TGSI_VS_WINDOW_SPACE_POSITION:
    case PIPE_CAP_DRAW_INDIRECT:
@@ -430,7 +431,6 @@ svga_get_param(struct pipe_screen *screen, enum pipe_cap param)
    case PIPE_CAP_TEXTURE_HALF_FLOAT_LINEAR:
    case PIPE_CAP_DEPTH_BOUNDS_TEST:
    case PIPE_CAP_TGSI_TXQS:
-   case PIPE_CAP_FORCE_PERSAMPLE_INTERP:
    case PIPE_CAP_SHAREABLE_SHADERS:
    case PIPE_CAP_DRAW_PARAMETERS:
    case PIPE_CAP_TGSI_FS_POSITION_IS_SYSVAL:

diff --git a/src/gallium/drivers/svga/svga_tgsi_vgpu10.c b/src/gallium/drivers/svga/svga_tgsi_vgpu10.c
index 44e451332eee374c3c4f101c9d02d93bbcdf20bd..756b67316a101ceb8bf34e1a2ca048135589c804 100644 (file)
--- a/src/gallium/drivers/svga/svga_tgsi_vgpu10.c
+++ b/src/gallium/drivers/svga/svga_tgsi_vgpu10.c
@@ -415,6 +415,9 @@ check_register_index(struct svga_shader_emitter_v10 *emit,
          emit->register_overflow = TRUE;
       }
       break;
+   case VGPU10_OPERAND_TYPE_OUTPUT_COVERAGE_MASK:
+      /* nothing */
+      break;
    default:
       assert(0);
       ; /* nothing */
@@ -921,6 +924,15 @@ emit_dst_register(struct svga_shader_emitter_v10 *emit,
             emit_dword(emit, operand0.value);
             return;
          }
+         else if (sem_name == TGSI_SEMANTIC_SAMPLEMASK) {
+            /* Fragment sample mask output */
+            operand0.value = 0;
+            operand0.operandType = VGPU10_OPERAND_TYPE_OUTPUT_COVERAGE_MASK;
+            operand0.indexDimension = VGPU10_OPERAND_INDEX_0D;
+            operand0.numComponents = VGPU10_OPERAND_1_COMPONENT;
+            emit_dword(emit, operand0.value);
+            return;
+         }
          else if (index == emit->fs.color_out_index[0] &&
              emit->fs.color_tmp_index != INVALID_INDEX) {
             /* replace OUTPUT[COLOR] with TEMP[COLOR].  We need to store the
@@ -2271,6 +2283,31 @@ emit_fragdepth_output_declaration(struct svga_shader_emitter_v10 *emit)
 }
 
 
+/**
+ * Emit the declaration for the fragment sample mask/coverage output.
+ */
+static void
+emit_samplemask_output_declaration(struct svga_shader_emitter_v10 *emit)
+{
+   VGPU10OpcodeToken0 opcode0;
+   VGPU10OperandToken0 operand0;
+   VGPU10NameToken name_token;
+
+   assert(emit->unit == PIPE_SHADER_FRAGMENT);
+   assert(emit->version >= 41);
+
+   opcode0.value = operand0.value = name_token.value = 0;
+
+   opcode0.opcodeType = VGPU10_OPCODE_DCL_OUTPUT;
+   operand0.operandType = VGPU10_OPERAND_TYPE_OUTPUT_COVERAGE_MASK;
+   operand0.numComponents = VGPU10_OPERAND_0_COMPONENT;
+   operand0.indexDimension = VGPU10_OPERAND_INDEX_0D;
+   operand0.mask = VGPU10_OPERAND_4_COMPONENT_MASK_ALL;
+
+   emit_decl_instruction(emit, opcode0, operand0, name_token, 0, 1);
+}
+
+
 /**
  * Emit the declaration for a system value input/output.
  */
@@ -2671,6 +2708,10 @@ emit_output_declarations(struct svga_shader_emitter_v10 *emit)
             /* Fragment depth output */
             emit_fragdepth_output_declaration(emit);
          }
+         else if (semantic_name == TGSI_SEMANTIC_SAMPLEMASK) {
+            /* Fragment depth output */
+            emit_samplemask_output_declaration(emit);
+         }
          else {
             assert(!"Bad output semantic name");
          }