amd/common: correctly detect if we need ring buffers

When allocate_user_sgprs() was called, ctx->stage was actually
unset and 0 is for the vertex shader. This doesn't change
anything for now because of the spill support thing.

Though, the number of user SGPRs has to be fixed for merged
shaders on GFX9. It was broken before anyway.

Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>

1 files changed, 9 insertions, 7 deletions

diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 90faf0537f7..ac0a2e67d40 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -541,19 +541,20 @@ struct user_sgpr_info {
 };
 
 static void allocate_user_sgprs(struct nir_to_llvm_context *ctx,
+				gl_shader_stage stage,
 				struct user_sgpr_info *user_sgpr_info)
 {
 	memset(user_sgpr_info, 0, sizeof(struct user_sgpr_info));
 
 	/* until we sort out scratch/global buffers always assign ring offsets for gs/vs/es */
-	if (ctx->stage == MESA_SHADER_GEOMETRY ||
-	    ctx->stage == MESA_SHADER_VERTEX ||
-	    ctx->stage == MESA_SHADER_TESS_CTRL ||
-	    ctx->stage == MESA_SHADER_TESS_EVAL ||
+	if (stage == MESA_SHADER_GEOMETRY ||
+	    stage == MESA_SHADER_VERTEX ||
+	    stage == MESA_SHADER_TESS_CTRL ||
+	    stage == MESA_SHADER_TESS_EVAL ||
 	    ctx->is_gs_copy_shader)
 		user_sgpr_info->need_ring_offsets = true;
 
-	if (ctx->stage == MESA_SHADER_FRAGMENT &&
+	if (stage == MESA_SHADER_FRAGMENT &&
 	    ctx->shader_info->info.ps.needs_sample_positions)
 		user_sgpr_info->need_ring_offsets = true;
 
@@ -562,7 +563,8 @@ static void allocate_user_sgprs(struct nir_to_llvm_context *ctx,
 		user_sgpr_info->sgpr_count += 2;
 	}
 
-	switch (ctx->stage) {
+	/* FIXME: fix the number of user sgprs for merged shaders on GFX9 */
+	switch (stage) {
 	case MESA_SHADER_COMPUTE:
 		if (ctx->shader_info->info.cs.uses_grid_size)
 			user_sgpr_info->sgpr_count += 3;
@@ -760,7 +762,7 @@ static void create_function(struct nir_to_llvm_context *ctx,
 	struct arg_info args = {};
 	LLVMValueRef desc_sets;
 
-	allocate_user_sgprs(ctx, &user_sgpr_info);
+	allocate_user_sgprs(ctx, stage, &user_sgpr_info);
 
 	if (user_sgpr_info.need_ring_offsets && !ctx->options->supports_spill) {
 		add_arg(&args, ARG_SGPR, const_array(ctx->ac.v4i32, 16),