1; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds < %s | FileCheck %s
2; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
3
4; CHECK: %llvm.amdgcn.module.lds.t = type { float, float }
5
6@func = addrspace(3) global float undef, align 4
7
8; @kern is only used from a kernel so it is left unchanged
9; CHECK: @kern = addrspace(3) global float undef, align 4
10@kern = addrspace(3) global float undef, align 4
11
12; @func is only used from a non-kernel function so is rewritten
13; CHECK-NOT: @func
14; @both is used from a non-kernel function so is rewritten
15; CHECK-NOT: @both
16; sorted both < func, so @both at null and @func at 4
17@both = addrspace(3) global float undef, align 4
18
19; CHECK: @llvm.amdgcn.module.lds = internal addrspace(3) global %llvm.amdgcn.module.lds.t undef, align 4
20
21; CHECK-LABEL: @get_func()
22; CHECK: %0 = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* getelementptr (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* null, i32 0, i32 1) to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* getelementptr (%llvm.amdgcn.module.lds.t, %llvm.amdgcn.module.lds.t addrspace(3)* null, i32 0, i32 1) to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
23define i32 @get_func() local_unnamed_addr #0 {
24entry:
25  %0 = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @func to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @func to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
26  ret i32 %0
27}
28
29; CHECK-LABEL: @set_func(i32 %x)
30; CHECK: store i32 %x, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* null to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* null to i32*) to i64)) to i32*), align 4
31define void @set_func(i32 %x) local_unnamed_addr #1 {
32entry:
33  store i32 %x, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
34  ret void
35}
36
37; CHECK-LABEL: @timestwo()
38; CHECK: call void @llvm.donothing() [ "ExplicitUse"(%llvm.amdgcn.module.lds.t addrspace(3)* @llvm.amdgcn.module.lds) ]
39; CHECK: %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* null to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
40; CHECK: %mul = mul i32 %ld, 2
41; CHECK: store i32 %mul, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* null to i32*) to i64)) to i32*), align 4
42define amdgpu_kernel void @timestwo() {
43  %ld = load i32, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
44  %mul = mul i32 %ld, 2
45  store i32 %mul, i32* inttoptr (i64 add (i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @kern to i32 addrspace(3)*) to i32*) to i64), i64 ptrtoint (i32* addrspacecast (i32 addrspace(3)* bitcast (float addrspace(3)* @both to i32 addrspace(3)*) to i32*) to i64)) to i32*), align 4
46  ret void
47}
48