AMDGPU: Uniform branch conditions can originate with intrinsics

Summary:
Discovered by Dave Airlie, fixes an assertion in Khronos OpenGL CTS
GL43-CTS.shader_storage_buffer_object.advanced-matrix.

In this particular case, the buffer load intrinsic fed into a uniform
conditional branch, and led the brcond lowering down the wrong path.

Reviewers: tstellarAMD, arsenm

Subscribers: arsenm, llvm-commits

Differential Revision: http://reviews.llvm.org/D19931

llvm-svn: 268650

1 files changed, 27 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/AMDGPU/uniform-branch-intrinsic-cond.ll b/llvm/test/CodeGen/AMDGPU/uniform-branch-intrinsic-cond.ll
new file mode 100644
index 00000000000..93a2c6998be
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/uniform-branch-intrinsic-cond.ll
@@ -0,0 +1,27 @@
+; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck %s
+
+; This used to raise an assertion due to how the choice between uniform and
+; non-uniform branches was determined.
+;
+; CHECK-LABEL: {{^}}main:
+; CHECK: s_cbranch_vccnz
+define amdgpu_ps float @main(<4 x i32> inreg %rsrc) {
+main_body:
+  %v = call float @llvm.amdgcn.buffer.load.f32(<4 x i32> %rsrc, i32 0, i32 0, i1 true, i1 false)
+  %cc = fcmp une float %v, 1.000000e+00
+  br i1 %cc, label %if, label %else
+
+if:
+  %u = fadd float %v, %v
+  br label %else
+
+else:
+  %r = phi float [ %v, %main_body ], [ %u, %if ]
+  ret float %r
+}
+
+; Function Attrs: nounwind readonly
+declare float @llvm.amdgcn.buffer.load.f32(<4 x i32>, i32, i32, i1, i1) #0
+
+attributes #0 = { nounwind readonly }