101 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals --include-generated-funcs --version 32; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-ctor-dtor < %s | FileCheck %s3; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-ctor-dtor < %s | FileCheck %s4 5; Make sure we get the same result if we run multiple times6; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-ctor-dtor,amdgpu-lower-ctor-dtor < %s | FileCheck %s7; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx700 -filetype=obj -o - < %s | llvm-readelf -s - 2>&1 | FileCheck %s -check-prefix=VISIBILITY8; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx700 -filetype=obj -o - < %s | llvm-readelf -S - 2>&1 | FileCheck %s -check-prefix=SECTION9; RUN: llc -mtriple=amdgcn-amd-amdhsa -amdgpu-lower-global-ctor-dtor=0 -mcpu=gfx700 -filetype=obj -o - < %s | llvm-readelf -s - 2>&1 | FileCheck %s -check-prefix=DISABLED10; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx700 -filetype=obj -o - < %s | llvm-readelf --notes - 2>&1 | FileCheck %s -check-prefix=METADATA11 12@llvm.global_ctors = appending addrspace(1) global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @foo, ptr null }]13@llvm.global_dtors = appending addrspace(1) global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @bar, ptr null }]14 15; VISIBILITY: FUNC WEAK PROTECTED {{.*}} amdgcn.device.init16; VISIBILITY: OBJECT WEAK DEFAULT {{.*}} amdgcn.device.init.kd17; VISIBILITY: FUNC WEAK PROTECTED {{.*}} amdgcn.device.fini18; VISIBILITY: OBJECT WEAK DEFAULT {{.*}} amdgcn.device.fini.kd19 20; SECTION: .init_array.1 INIT_ARRAY {{.*}} {{.*}} 000008 00 WA 0 0 821; SECTION: .fini_array.1 FINI_ARRAY {{.*}} {{.*}} 000008 00 WA 0 0 822 23; DISABLED-NOT: FUNC GLOBAL PROTECTED {{.*}} amdgcn.device.init24; DISABLED-NOT: OBJECT GLOBAL DEFAULT {{.*}} amdgcn.device.init.kd25; DISABLED-NOT: FUNC GLOBAL PROTECTED {{.*}} amdgcn.device.fini26; DISABLED-NOT: OBJECT GLOBAL DEFAULT {{.*}} amdgcn.device.fini.kd27 28; METADATA: amdhsa.kernels:29; METADATA: .kind: init30; METADATA: .max_flat_workgroup_size: 131; METADATA: .name: amdgcn.device.init32; METADATA: .symbol: amdgcn.device.init.kd33; METADATA: .kind: fini34; METADATA: .max_flat_workgroup_size: 135; METADATA: .name: amdgcn.device.fini36; METADATA: .symbol: amdgcn.device.fini.kd37 38define internal void @foo() {39 ret void40}41 42define internal void @bar() {43 ret void44}45 46;.47; CHECK: @llvm.global_ctors = appending addrspace(1) global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @foo, ptr null }]48; CHECK: @llvm.global_dtors = appending addrspace(1) global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @bar, ptr null }]49; CHECK: @__init_array_start = external addrspace(1) constant [0 x ptr addrspace(1)]50; CHECK: @__init_array_end = external addrspace(1) constant [0 x ptr addrspace(1)]51; CHECK: @__fini_array_start = external addrspace(1) constant [0 x ptr addrspace(1)]52; CHECK: @__fini_array_end = external addrspace(1) constant [0 x ptr addrspace(1)]53; CHECK: @llvm.used = appending addrspace(1) global [2 x ptr] [ptr @amdgcn.device.init, ptr @amdgcn.device.fini], section "llvm.metadata"54;.55; CHECK-LABEL: define internal void @foo() {56; CHECK-NEXT: ret void57;58;59; CHECK-LABEL: define internal void @bar() {60; CHECK-NEXT: ret void61;62;63; CHECK-LABEL: define weak_odr amdgpu_kernel void @amdgcn.device.init(64; CHECK-SAME: ) #[[ATTR0:[0-9]+]] {65; CHECK-NEXT: entry:66; CHECK-NEXT: [[TMP0:%.*]] = icmp ne ptr addrspace(1) @__init_array_start, @__init_array_end67; CHECK-NEXT: br i1 [[TMP0]], label [[WHILE_ENTRY:%.*]], label [[WHILE_END:%.*]]68; CHECK: while.entry:69; CHECK-NEXT: [[PTR:%.*]] = phi ptr addrspace(1) [ @__init_array_start, [[ENTRY:%.*]] ], [ [[NEXT:%.*]], [[WHILE_ENTRY]] ]70; CHECK-NEXT: [[CALLBACK:%.*]] = load ptr, ptr addrspace(1) [[PTR]], align 871; CHECK-NEXT: call void [[CALLBACK]]()72; CHECK-NEXT: [[NEXT]] = getelementptr ptr addrspace(1), ptr addrspace(1) [[PTR]], i64 173; CHECK-NEXT: [[END:%.*]] = icmp eq ptr addrspace(1) [[NEXT]], @__init_array_end74; CHECK-NEXT: br i1 [[END]], label [[WHILE_END]], label [[WHILE_ENTRY]]75; CHECK: while.end:76; CHECK-NEXT: ret void77;78;79; CHECK-LABEL: define weak_odr amdgpu_kernel void @amdgcn.device.fini(80; CHECK-SAME: ) #[[ATTR1:[0-9]+]] {81; CHECK-NEXT: entry:82; CHECK-NEXT: [[TMP0:%.*]] = ashr exact i64 sub nuw nsw (i64 ptrtoint (ptr addrspace(1) @__fini_array_end to i64), i64 ptrtoint (ptr addrspace(1) @__fini_array_start to i64)), 383; CHECK-NEXT: [[TMP1:%.*]] = sub nuw nsw i64 [[TMP0]], 184; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [0 x ptr addrspace(1)], ptr addrspace(1) @__fini_array_start, i64 0, i64 [[TMP1]]85; CHECK-NEXT: [[TMP3:%.*]] = icmp uge ptr addrspace(1) [[TMP2]], @__fini_array_start86; CHECK-NEXT: br i1 [[TMP3]], label [[WHILE_ENTRY:%.*]], label [[WHILE_END:%.*]]87; CHECK: while.entry:88; CHECK-NEXT: [[PTR:%.*]] = phi ptr addrspace(1) [ [[TMP2]], [[ENTRY:%.*]] ], [ [[NEXT:%.*]], [[WHILE_ENTRY]] ]89; CHECK-NEXT: [[CALLBACK:%.*]] = load ptr, ptr addrspace(1) [[PTR]], align 890; CHECK-NEXT: call void [[CALLBACK]]()91; CHECK-NEXT: [[NEXT]] = getelementptr ptr addrspace(1), ptr addrspace(1) [[PTR]], i64 -192; CHECK-NEXT: [[END:%.*]] = icmp ult ptr addrspace(1) [[NEXT]], @__fini_array_start93; CHECK-NEXT: br i1 [[END]], label [[WHILE_END]], label [[WHILE_ENTRY]]94; CHECK: while.end:95; CHECK-NEXT: ret void96;97;.98; CHECK: attributes #[[ATTR0]] = { "amdgpu-flat-work-group-size"="1,1" "device-init" }99; CHECK: attributes #[[ATTR1]] = { "amdgpu-flat-work-group-size"="1,1" "device-fini" }100;.101