[Mesa-dev] [PATCH 1/2] ac/nir: init full exec mask for merged shaders.

[Bas Nieuwenhuizen]

From: Dave Airlie <airlied at redhat.com>

Signed-off-by: Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
---
 src/amd/common/ac_llvm_build.c  | 8 ++++++++
 src/amd/common/ac_llvm_build.h  | 1 +
 src/amd/common/ac_nir_to_llvm.c | 3 +++
 3 files changed, 12 insertions(+)

diff --git a/src/amd/common/ac_llvm_build.c b/src/amd/common/ac_llvm_build.c
index 949f181aace..e5cd23e0251 100644
--- a/src/amd/common/ac_llvm_build.c
+++ b/src/amd/common/ac_llvm_build.c
@@ -1734,3 +1734,11 @@ void ac_optimize_vs_outputs(struct ac_llvm_context *ctx,
 		*num_param_exports = exports.num;
 	}
 }
+
+void ac_init_exec_full_mask(struct ac_llvm_context *ctx)
+{
+	LLVMValueRef full_mask = LLVMConstInt(ctx->i64, ~0ull, 0);
+	ac_build_intrinsic(ctx,
+			   "llvm.amdgcn.init.exec", ctx->voidt,
+			   &full_mask, 1, AC_FUNC_ATTR_CONVERGENT);
+}
diff --git a/src/amd/common/ac_llvm_build.h b/src/amd/common/ac_llvm_build.h
index f0b5875b423..aa2a2899ab4 100644
--- a/src/amd/common/ac_llvm_build.h
+++ b/src/amd/common/ac_llvm_build.h
@@ -281,6 +281,7 @@ void ac_optimize_vs_outputs(struct ac_llvm_context *ac,
 			    uint8_t *vs_output_param_offset,
 			    uint32_t num_outputs,
 			    uint8_t *num_param_exports);
+void ac_init_exec_full_mask(struct ac_llvm_context *ctx);
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 242675654d2..b6a7d43789d 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -6488,6 +6488,9 @@ LLVMModuleRef ac_translate_nir_to_llvm(LLVMTargetMachineRef tm,
 	ctx.abi.load_ssbo = radv_load_ssbo;
 	ctx.abi.load_sampler_desc = radv_get_sampler_desc;
 
+	if (shader_count >= 2)
+		ac_init_exec_full_mask(&ctx.ac);
+
 	if (ctx.ac.chip_class == GFX9 &&
 	    shaders[shader_count - 1]->stage == MESA_SHADER_TESS_CTRL)
 		ac_nir_fixup_ls_hs_input_vgprs(&ctx);
-- 
2.14.2