103 lines · plain
1; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-module-lds --amdgpu-lower-module-lds-strategy=module < %s | FileCheck %s2; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds --amdgpu-lower-module-lds-strategy=module < %s | FileCheck %s3 4@lds.size.1.align.1 = internal unnamed_addr addrspace(3) global [1 x i8] poison, align 15@lds.size.2.align.2 = internal unnamed_addr addrspace(3) global [2 x i8] poison, align 26@lds.size.4.align.4 = internal unnamed_addr addrspace(3) global [4 x i8] poison, align 47@lds.size.8.align.8 = internal unnamed_addr addrspace(3) global [8 x i8] poison, align 88@lds.size.16.align.16 = internal unnamed_addr addrspace(3) global [16 x i8] poison, align 169 10; CHECK: %llvm.amdgcn.module.lds.t = type { [8 x i8], [1 x i8] }11; CHECK: %llvm.amdgcn.kernel.k0.lds.t = type { [16 x i8], [4 x i8], [2 x i8], [1 x i8] }12; CHECK: %llvm.amdgcn.kernel.k1.lds.t = type { [16 x i8], [4 x i8], [2 x i8] }13; CHECK: %llvm.amdgcn.kernel.k2.lds.t = type { [2 x i8] }14; CHECK: %llvm.amdgcn.kernel.k3.lds.t = type { [4 x i8] }15 16;.17; CHECK: @llvm.amdgcn.module.lds = internal addrspace(3) global %llvm.amdgcn.module.lds.t poison, align 8, !absolute_symbol !018; CHECK: @llvm.compiler.used = appending addrspace(1) global [1 x ptr] [ptr addrspacecast (ptr addrspace(3) @llvm.amdgcn.module.lds to ptr)], section "llvm.metadata"19; CHECK: @llvm.amdgcn.kernel.k0.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k0.lds.t poison, align 16, !absolute_symbol !020; CHECK: @llvm.amdgcn.kernel.k1.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k1.lds.t poison, align 16, !absolute_symbol !021; CHECK: @llvm.amdgcn.kernel.k2.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k2.lds.t poison, align 2, !absolute_symbol !022; CHECK: @llvm.amdgcn.kernel.k3.lds = internal addrspace(3) global %llvm.amdgcn.kernel.k3.lds.t poison, align 4, !absolute_symbol !023;.24define amdgpu_kernel void @k0() #0 {25; CHECK-LABEL: @k0() #026; CHECK-NEXT: store i8 1, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, ptr addrspace(3) @llvm.amdgcn.kernel.k0.lds, i32 0, i32 3), align 2, !alias.scope !1, !noalias !427; CHECK-NEXT: store i8 2, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, ptr addrspace(3) @llvm.amdgcn.kernel.k0.lds, i32 0, i32 2), align 4, !alias.scope !8, !noalias !928; CHECK-NEXT: store i8 4, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.kernel.k0.lds.t, ptr addrspace(3) @llvm.amdgcn.kernel.k0.lds, i32 0, i32 1), align 16, !alias.scope !10, !noalias !1129; CHECK-NEXT: store i8 16, ptr addrspace(3) @llvm.amdgcn.kernel.k0.lds, align 16, !alias.scope !12, !noalias !1330; CHECK-NEXT: ret void31 store i8 1, ptr addrspace(3) @lds.size.1.align.1, align 132 33 store i8 2, ptr addrspace(3) @lds.size.2.align.2, align 234 35 store i8 4, ptr addrspace(3) @lds.size.4.align.4, align 436 37 store i8 16, ptr addrspace(3) @lds.size.16.align.16, align 1638 39 ret void40}41 42define amdgpu_kernel void @k1() #0 {43; CHECK-LABEL: @k1() #144; CHECK-NEXT: store i8 2, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, ptr addrspace(3) @llvm.amdgcn.kernel.k1.lds, i32 0, i32 2), align 4, !alias.scope !14, !noalias !1745; CHECK-NEXT: store i8 4, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.kernel.k1.lds.t, ptr addrspace(3) @llvm.amdgcn.kernel.k1.lds, i32 0, i32 1), align 16, !alias.scope !20, !noalias !2146; CHECK-NEXT: store i8 16, ptr addrspace(3) @llvm.amdgcn.kernel.k1.lds, align 16, !alias.scope !22, !noalias !2347; CHECK-NEXT: ret void48;49 store i8 2, ptr addrspace(3) @lds.size.2.align.2, align 250 51 store i8 4, ptr addrspace(3) @lds.size.4.align.4, align 452 53 store i8 16, ptr addrspace(3) @lds.size.16.align.16, align 1654 55 ret void56}57 58define amdgpu_kernel void @k2() #0 {59; CHECK-LABEL: @k2() #260; CHECK-NEXT: store i8 2, ptr addrspace(3) @llvm.amdgcn.kernel.k2.lds, align 261; CHECK-NEXT: ret void62;63 store i8 2, ptr addrspace(3) @lds.size.2.align.2, align 264 65 ret void66}67 68define amdgpu_kernel void @k3() #0 {69; CHECK-LABEL: @k3() #370; CHECK-NEXT: store i8 4, ptr addrspace(3) @llvm.amdgcn.kernel.k3.lds, align 471; CHECK-NEXT: ret void72;73 store i8 4, ptr addrspace(3) @lds.size.4.align.4, align 474 75 ret void76}77 78; CHECK-LABEL: @calls_f0() #479define amdgpu_kernel void @calls_f0() {80 call void @f0()81 ret void82}83 84define void @f0() {85; CHECK-LABEL: define void @f0()86; CHECK-NEXT: store i8 1, ptr addrspace(3) getelementptr inbounds (%llvm.amdgcn.module.lds.t, ptr addrspace(3) @llvm.amdgcn.module.lds, i32 0, i32 1), align 8, !noalias !2487; CHECK-NEXT: store i8 8, ptr addrspace(3) @llvm.amdgcn.module.lds, align 8, !noalias !2988; CHECK-NEXT: ret void89 store i8 1, ptr addrspace(3) @lds.size.1.align.1, align 190 91 store i8 8, ptr addrspace(3) @lds.size.8.align.8, align 492 93 ret void94}95 96; CHECK: attributes #0 = { "amdgpu-lds-size"="23" }97; CHECK: attributes #1 = { "amdgpu-lds-size"="22" }98; CHECK: attributes #2 = { "amdgpu-lds-size"="2" }99; CHECK: attributes #3 = { "amdgpu-lds-size"="4" }100; CHECK: attributes #4 = { "amdgpu-lds-size"="9" }101 102; CHECK: !0 = !{i32 0, i32 1}103