[Mesa-dev] [PATCH 3/3] ac/radv: mark all tess output for an indirect access.

[Dave Airlie]

From: Dave Airlie <airlied at redhat.com>

If a shader does a tcs store with an indirect access, we
were only marking the first spot as used. For indirect access
we always now mark all slots used by the variable.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105464
Fixes: 94f9591995 (radv/ac: add support for TCS/TES inputs/outputs.)
Signed-off-by: Dave Airlie <airlied at redhat.com>
---
 src/amd/common/ac_nir_to_llvm.c | 23 +++++++++++++++--------
 1 file changed, 15 insertions(+), 8 deletions(-)

diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 600b6f72167..7b9b37ce163 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -2773,13 +2773,14 @@ static LLVMValueRef get_tcs_tes_buffer_address_params(struct radv_shader_context
 
 static void
 mark_tess_output(struct radv_shader_context *ctx,
-		 bool is_patch, uint32_t param)
+		 bool is_patch, uint32_t param, int num_slots)
 
 {
+	uint64_t slot_mask = (1ull << num_slots) - 1;
 	if (is_patch) {
-		ctx->tess_patch_outputs_written |= (1ull << param);
+		ctx->tess_patch_outputs_written |= (slot_mask << param);
 	} else
-		ctx->tess_outputs_written |= (1ull << param);
+		ctx->tess_outputs_written |= (slot_mask << param);
 }
 
 static LLVMValueRef
@@ -2876,6 +2877,7 @@ store_tcs_output(struct ac_shader_abi *abi,
 	const unsigned component = var->data.location_frac;
 	const bool is_patch = var->data.patch;
 	const bool is_compact = var->data.compact;
+	const unsigned count = glsl_count_attribute_slots(var->type, false);
 	LLVMValueRef dw_addr;
 	LLVMValueRef stride = NULL;
 	LLVMValueRef buf_addr = NULL;
@@ -2904,7 +2906,12 @@ store_tcs_output(struct ac_shader_abi *abi,
 		dw_addr = get_tcs_out_current_patch_data_offset(ctx);
 	}
 
-	mark_tess_output(ctx, is_patch, param);
+
+	if (param_index) {
+		mark_tess_output(ctx, is_patch, param, count);
+	} else {
+		mark_tess_output(ctx, is_patch, param, 1);
+	}
 
 	dw_addr = get_dw_address(ctx, dw_addr, param, const_index, is_compact, vertex_index, stride,
 				 param_index);
@@ -6201,9 +6208,9 @@ handle_ls_outputs_post(struct radv_shader_context *ctx)
 		if (i == VARYING_SLOT_CLIP_DIST0)
 			length = ctx->num_output_clips + ctx->num_output_culls;
 		int param = shader_io_get_unique_index(i);
-		mark_tess_output(ctx, false, param);
+		mark_tess_output(ctx, false, param, 1);
 		if (length > 4)
-			mark_tess_output(ctx, false, param + 1);
+			mark_tess_output(ctx, false, param + 1, 1);
 		LLVMValueRef dw_addr = LLVMBuildAdd(ctx->ac.builder, base_dw_addr,
 						    LLVMConstInt(ctx->ac.i32, param * 4, false),
 						    "");
@@ -6347,13 +6354,13 @@ write_tess_factors(struct radv_shader_context *ctx)
 
 	if (inner_comps) {
 		tess_inner_index = shader_io_get_unique_index(VARYING_SLOT_TESS_LEVEL_INNER);
-		mark_tess_output(ctx, true, tess_inner_index);
+		mark_tess_output(ctx, true, tess_inner_index, 1);
 		lds_inner = LLVMBuildAdd(ctx->ac.builder, lds_base,
 					 LLVMConstInt(ctx->ac.i32, tess_inner_index * 4, false), "");
 	}
 
 	tess_outer_index = shader_io_get_unique_index(VARYING_SLOT_TESS_LEVEL_OUTER);
-	mark_tess_output(ctx, true, tess_outer_index);
+	mark_tess_output(ctx, true, tess_outer_index, 1);
 	lds_outer = LLVMBuildAdd(ctx->ac.builder, lds_base,
 				 LLVMConstInt(ctx->ac.i32, tess_outer_index * 4, false), "");
 
-- 
2.14.3