|
120
|
1 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck %s
|
|
|
2 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck %s
|
|
|
3
|
|
|
4 ; This is used to crash in LiveIntervalAnalysis via SILoadStoreOptimizer
|
|
|
5 ; while fixing up the merge of two ds_write instructions.
|
|
|
6
|
|
|
7 @tess_lds = external addrspace(3) global [8192 x i32]
|
|
|
8
|
|
|
9 ; CHECK-LABEL: {{^}}main:
|
|
|
10 ; CHECK: ds_write2_b32
|
|
|
11 ; CHECK: v_mov_b32_e32 v1, v0
|
|
|
12 ; CHECK: tbuffer_store_format_xyzw v[0:3],
|
|
|
13 define amdgpu_vs void @main(i32 inreg %arg) {
|
|
|
14 main_body:
|
|
|
15 %tmp = load float, float addrspace(3)* undef, align 4
|
|
|
16 %tmp1 = load float, float addrspace(3)* undef, align 4
|
|
|
17 store float %tmp, float addrspace(3)* null, align 4
|
|
|
18 %tmp2 = bitcast float %tmp to i32
|
|
|
19 %tmp3 = add nuw nsw i32 0, 1
|
|
|
20 %tmp4 = zext i32 %tmp3 to i64
|
|
|
21 %tmp5 = getelementptr [8192 x i32], [8192 x i32] addrspace(3)* @tess_lds, i64 0, i64 %tmp4
|
|
|
22 %tmp6 = bitcast i32 addrspace(3)* %tmp5 to float addrspace(3)*
|
|
|
23 store float %tmp1, float addrspace(3)* %tmp6, align 4
|
|
|
24 %tmp7 = bitcast float %tmp1 to i32
|
|
|
25 %tmp8 = insertelement <4 x i32> undef, i32 %tmp2, i32 0
|
|
|
26 %tmp9 = insertelement <4 x i32> %tmp8, i32 %tmp7, i32 1
|
|
|
27 %tmp10 = insertelement <4 x i32> %tmp9, i32 undef, i32 2
|
|
|
28 %tmp11 = insertelement <4 x i32> %tmp10, i32 undef, i32 3
|
|
121
|
29 call void @llvm.amdgcn.tbuffer.store.v4i32(<4 x i32> %tmp11, <4 x i32> undef, i32 undef, i32 0, i32 %arg, i32 0, i32 14, i32 4, i1 1, i1 1)
|
|
120
|
30 ret void
|
|
|
31 }
|
|
|
32
|
|
|
33 ; Function Attrs: nounwind
|
|
121
|
34 declare void @llvm.amdgcn.tbuffer.store.v4i32(<4 x i32>, <4 x i32>, i32, i32, i32, i32, i32, i32, i1, i1) #0
|
|
120
|
35
|
|
|
36 attributes #0 = { nounwind }
|
