diff options
Diffstat (limited to 'test/CodeGen/R600/array-ptr-calc-i32.ll')
| -rw-r--r-- | test/CodeGen/R600/array-ptr-calc-i32.ll | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/test/CodeGen/R600/array-ptr-calc-i32.ll b/test/CodeGen/R600/array-ptr-calc-i32.ll index 33a8aee..8c2a079 100644 --- a/test/CodeGen/R600/array-ptr-calc-i32.ll +++ b/test/CodeGen/R600/array-ptr-calc-i32.ll @@ -20,24 +20,24 @@ declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate ; FIXME: The AMDGPUPromoteAlloca pass should be able to convert this ; alloca to a vector. It currently fails because it does not know how ; to interpret: -; getelementptr [4 x i32]* %alloca, i32 1, i32 %b +; 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: 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 %tid = call i32 @llvm.SI.tid() readnone - %a_ptr = getelementptr i32 addrspace(1)* %inA, i32 %tid - %b_ptr = getelementptr i32 addrspace(1)* %inB, i32 %tid - %a = load i32 addrspace(1)* %a_ptr - %b = load i32 addrspace(1)* %b_ptr + %a_ptr = getelementptr i32, i32 addrspace(1)* %inA, i32 %tid + %b_ptr = getelementptr i32, i32 addrspace(1)* %inB, i32 %tid + %a = load i32, i32 addrspace(1)* %a_ptr + %b = load i32, i32 addrspace(1)* %b_ptr %result = add i32 %a, %b - %alloca_ptr = getelementptr [4 x i32]* %alloca, i32 1, i32 %b + %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 - %reload = load i32* %alloca_ptr, align 4 - %out_ptr = getelementptr i32 addrspace(1)* %out, i32 %tid + %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 ret void } |