diff options
| author | Dimitry Andric <dim@FreeBSD.org> | 2015-12-30 11:46:15 +0000 |
|---|---|---|
| committer | Dimitry Andric <dim@FreeBSD.org> | 2015-12-30 11:46:15 +0000 |
| commit | dd58ef019b700900793a1eb48b52123db01b654e (patch) | |
| tree | fcfbb4df56a744f4ddc6122c50521dd3f1c5e196 /test/CodeGen/AMDGPU/array-ptr-calc-i32.ll | |
| parent | 2fe5752e3a7c345cdb59e869278d36af33c13fa4 (diff) | |
Notes
Diffstat (limited to 'test/CodeGen/AMDGPU/array-ptr-calc-i32.ll')
| -rw-r--r-- | test/CodeGen/AMDGPU/array-ptr-calc-i32.ll | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/test/CodeGen/AMDGPU/array-ptr-calc-i32.ll b/test/CodeGen/AMDGPU/array-ptr-calc-i32.ll index 8c2a0795860d8..f8a74222d5669 100644 --- a/test/CodeGen/AMDGPU/array-ptr-calc-i32.ll +++ b/test/CodeGen/AMDGPU/array-ptr-calc-i32.ll @@ -2,7 +2,7 @@ ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI -mattr=+promote-alloca < %s | FileCheck -check-prefix=SI-PROMOTE -check-prefix=SI %s declare i32 @llvm.SI.tid() nounwind readnone -declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate +declare void @llvm.AMDGPU.barrier.local() nounwind convergent ; The required pointer calculations for the alloca'd actually requires ; an add and won't be folded into the addressing, which fails with a @@ -14,7 +14,7 @@ declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate ; FIXME: We end up with zero argument for ADD, because ; SIRegisterInfo::eliminateFrameIndex() blindly replaces the frame index ; with the appropriate offset. We should fold this into the store. -; SI-ALLOCA: v_add_i32_e32 [[PTRREG:v[0-9]+]], 0, v{{[0-9]+}} +; SI-ALLOCA: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 0, v{{[0-9]+}} ; SI-ALLOCA: buffer_store_dword {{v[0-9]+}}, [[PTRREG]], s[{{[0-9]+:[0-9]+}}] ; ; FIXME: The AMDGPUPromoteAlloca pass should be able to convert this @@ -22,7 +22,7 @@ declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate ; to interpret: ; getelementptr [4 x i32], [4 x i32]* %alloca, i32 1, i32 %b -; SI-PROMOTE: v_add_i32_e32 [[PTRREG:v[0-9]+]], 16 +; SI-PROMOTE: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 16 ; SI-PROMOTE: ds_write_b32 [[PTRREG]] define void @test_private_array_ptr_calc(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %inA, i32 addrspace(1)* noalias %inB) { %alloca = alloca [4 x i32], i32 4, align 16 @@ -35,7 +35,7 @@ define void @test_private_array_ptr_calc(i32 addrspace(1)* noalias %out, i32 add %alloca_ptr = getelementptr [4 x i32], [4 x i32]* %alloca, i32 1, i32 %b store i32 %result, i32* %alloca_ptr, align 4 ; Dummy call - call void @llvm.AMDGPU.barrier.local() nounwind noduplicate + call void @llvm.AMDGPU.barrier.local() nounwind convergent %reload = load i32, i32* %alloca_ptr, align 4 %out_ptr = getelementptr i32, i32 addrspace(1)* %out, i32 %tid store i32 %reload, i32 addrspace(1)* %out_ptr, align 4 |