caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
1fff116645b3ee108e8d8edbd5b90bed8304a4a8
clang-p2996/clang/test/OpenMP/nvptx_lambda_capturing.cpp
Joseph Huber 1fff116645 [OpenMP] Change OpenMP code generation for target region entries 
This patch changes the code we generate to enter a target region on the
device. This is in-line with the new definition in the runtime that was
added previously. Additionally we implement this in the OpenMPIRBuilder
so that this code can be shared with Flang in the future.

Reviewed By: ABataev

Differential Revision: https://reviews.llvm.org/D128550
2022-07-08 14:44:11 -04:00

1573 lines
121 KiB
C++
Raw Blame History

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// REQUIRES: powerpc-registered-target
// REQUIRES: nvptx-registered-target
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple nvptx64-nvidia-cuda -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK2
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple nvptx64-nvidia-cuda -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -emit-pch -o %t
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple nvptx64-nvidia-cuda -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple nvptx64-nvidia-cuda -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -o - | FileCheck %s --check-prefix=CHECK3
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
template <typename T>
int foo(const T &t) {
#pragma omp target parallel
t();
return 0;
}
struct S {
int a = 15;
int foo() {
auto &&L = [&]() { return a; };
#pragma omp target
L();
#pragma omp target parallel
L();
return a + ::foo(L);
}
} s;
int main(int argc, char **argv) {
int &b = argc;
int &&c = 1;
int *d = &argc;
int a;
auto &&L = [&]() { return argc + b + c + reinterpret_cast<long int>(d) + a; };
#pragma omp target firstprivate(argc) map(to : a)
L();
#pragma omp target parallel
L();
return argc + s.foo();
}
#endif // HEADER
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK1-NEXT: [[B:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[REF_TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[D:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[L:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[REF_TMP1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP3:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: [[_TMP4:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP5:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP6:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS7:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS8:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS9:%.*]] = alloca [11 x i8*], align 8
// CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[B]], align 8
// CHECK1-NEXT: store i32 1, i32* [[REF_TMP]], align 4
// CHECK1-NEXT: store i32* [[REF_TMP]], i32** [[C]], align 8
// CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[D]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP1]], i32 0, i32 0
// CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP0]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP1]], i32 0, i32 1
// CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[B]], align 8
// CHECK1-NEXT: store i32* [[TMP2]], i32** [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP1]], i32 0, i32 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i32*, i32** [[C]], align 8
// CHECK1-NEXT: store i32* [[TMP4]], i32** [[TMP3]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP1]], i32 0, i32 3
// CHECK1-NEXT: store i32** [[D]], i32*** [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP1]], i32 0, i32 4
// CHECK1-NEXT: store i32* [[A]], i32** [[TMP6]], align 8
// CHECK1-NEXT: store %class.anon* [[REF_TMP1]], %class.anon** [[L]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i32*, i32** [[B]], align 8
// CHECK1-NEXT: store i32* [[TMP7]], i32** [[TMP]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[C]], align 8
// CHECK1-NEXT: store i32* [[TMP8]], i32** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load %class.anon*, %class.anon** [[L]], align 8
// CHECK1-NEXT: store %class.anon* [[TMP9]], %class.anon** [[_TMP3]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP10]], i32* [[CONV]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = load i64, i64* [[ARGC_CASTED]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32*, i32** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = load i32*, i32** [[D]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = load %class.anon*, %class.anon** [[_TMP3]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 0
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 0
// CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 1
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 1
// CHECK1-NEXT: [[TMP21:%.*]] = load i32*, i32** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 2
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 2
// CHECK1-NEXT: [[TMP24:%.*]] = load i32*, i32** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 3
// CHECK1-NEXT: [[TMP27:%.*]] = load i32**, i32*** [[TMP26]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 4
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP15]], i32 0, i32 4
// CHECK1-NEXT: [[TMP30:%.*]] = load i32*, i32** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP32:%.*]] = bitcast i8** [[TMP31]] to i64*
// CHECK1-NEXT: store i64 [[TMP11]], i64* [[TMP32]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP34:%.*]] = bitcast i8** [[TMP33]] to i64*
// CHECK1-NEXT: store i64 [[TMP11]], i64* [[TMP34]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP35]], align 8
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to i32**
// CHECK1-NEXT: store i32* [[TMP12]], i32** [[TMP37]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32**
// CHECK1-NEXT: store i32* [[TMP12]], i32** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP40]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32**
// CHECK1-NEXT: store i32* [[TMP13]], i32** [[TMP42]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to i32**
// CHECK1-NEXT: store i32* [[TMP13]], i32** [[TMP44]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP45]], align 8
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to i32**
// CHECK1-NEXT: store i32* [[TMP14]], i32** [[TMP47]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to i32**
// CHECK1-NEXT: store i32* [[TMP14]], i32** [[TMP49]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP50]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP52:%.*]] = bitcast i8** [[TMP51]] to i32**
// CHECK1-NEXT: store i32* [[A]], i32** [[TMP52]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: [[TMP54:%.*]] = bitcast i8** [[TMP53]] to i32**
// CHECK1-NEXT: store i32* [[A]], i32** [[TMP54]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP55]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK1-NEXT: [[TMP57:%.*]] = bitcast i8** [[TMP56]] to %class.anon**
// CHECK1-NEXT: store %class.anon* [[TMP15]], %class.anon** [[TMP57]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK1-NEXT: [[TMP59:%.*]] = bitcast i8** [[TMP58]] to %class.anon**
// CHECK1-NEXT: store %class.anon* [[TMP15]], %class.anon** [[TMP59]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP60]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK1-NEXT: [[TMP62:%.*]] = bitcast i8** [[TMP61]] to i32***
// CHECK1-NEXT: store i32** [[TMP16]], i32*** [[TMP62]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK1-NEXT: [[TMP64:%.*]] = bitcast i8** [[TMP63]] to i32**
// CHECK1-NEXT: store i32* [[TMP18]], i32** [[TMP64]], align 8
// CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK1-NEXT: store i8* null, i8** [[TMP65]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK1-NEXT: [[TMP67:%.*]] = bitcast i8** [[TMP66]] to i32***
// CHECK1-NEXT: store i32** [[TMP19]], i32*** [[TMP67]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK1-NEXT: [[TMP69:%.*]] = bitcast i8** [[TMP68]] to i32**
// CHECK1-NEXT: store i32* [[TMP21]], i32** [[TMP69]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK1-NEXT: store i8* null, i8** [[TMP70]], align 8
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK1-NEXT: [[TMP72:%.*]] = bitcast i8** [[TMP71]] to i32***
// CHECK1-NEXT: store i32** [[TMP22]], i32*** [[TMP72]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK1-NEXT: [[TMP74:%.*]] = bitcast i8** [[TMP73]] to i32**
// CHECK1-NEXT: store i32* [[TMP24]], i32** [[TMP74]], align 8
// CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP75]], align 8
// CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK1-NEXT: [[TMP77:%.*]] = bitcast i8** [[TMP76]] to i32****
// CHECK1-NEXT: store i32*** [[TMP25]], i32**** [[TMP77]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK1-NEXT: [[TMP79:%.*]] = bitcast i8** [[TMP78]] to i32***
// CHECK1-NEXT: store i32** [[TMP27]], i32*** [[TMP79]], align 8
// CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK1-NEXT: store i8* null, i8** [[TMP80]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 10
// CHECK1-NEXT: [[TMP82:%.*]] = bitcast i8** [[TMP81]] to i32***
// CHECK1-NEXT: store i32** [[TMP28]], i32*** [[TMP82]], align 8
// CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 10
// CHECK1-NEXT: [[TMP84:%.*]] = bitcast i8** [[TMP83]] to i32**
// CHECK1-NEXT: store i32* [[TMP30]], i32** [[TMP84]], align 8
// CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 10
// CHECK1-NEXT: store i8* null, i8** [[TMP85]], align 8
// CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP88]], align 4
// CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 11, i32* [[TMP89]], align 4
// CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP86]], i8*** [[TMP90]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP87]], i8*** [[TMP91]], align 8
// CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP92]], align 8
// CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP93]], align 8
// CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP94]], align 8
// CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP95]], align 8
// CHECK1-NEXT: [[TMP96:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP97:%.*]] = icmp ne i32 [[TMP96]], 0
// CHECK1-NEXT: br i1 [[TMP97]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41(i64 [[TMP11]], i32* [[TMP12]], i32* [[TMP13]], i32* [[TMP14]], i32* [[A]], %class.anon* [[TMP15]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP98:%.*]] = load i32*, i32** [[B]], align 8
// CHECK1-NEXT: store i32* [[TMP98]], i32** [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP99:%.*]] = load i32*, i32** [[C]], align 8
// CHECK1-NEXT: store i32* [[TMP99]], i32** [[_TMP5]], align 8
// CHECK1-NEXT: [[TMP100:%.*]] = load %class.anon*, %class.anon** [[L]], align 8
// CHECK1-NEXT: store %class.anon* [[TMP100]], %class.anon** [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP101:%.*]] = load i32*, i32** [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = load i32*, i32** [[_TMP5]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = load i32*, i32** [[D]], align 8
// CHECK1-NEXT: [[TMP104:%.*]] = load %class.anon*, %class.anon** [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 0
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 0
// CHECK1-NEXT: [[TMP107:%.*]] = load i32*, i32** [[TMP106]], align 8
// CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 1
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 1
// CHECK1-NEXT: [[TMP110:%.*]] = load i32*, i32** [[TMP109]], align 8
// CHECK1-NEXT: [[TMP111:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 2
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 2
// CHECK1-NEXT: [[TMP113:%.*]] = load i32*, i32** [[TMP112]], align 8
// CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 3
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 3
// CHECK1-NEXT: [[TMP116:%.*]] = load i32**, i32*** [[TMP115]], align 8
// CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 4
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP104]], i32 0, i32 4
// CHECK1-NEXT: [[TMP119:%.*]] = load i32*, i32** [[TMP118]], align 8
// CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP121:%.*]] = bitcast i8** [[TMP120]] to i32**
// CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP121]], align 8
// CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK1-NEXT: [[TMP123:%.*]] = bitcast i8** [[TMP122]] to i32**
// CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP123]], align 8
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP124]], align 8
// CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 1
// CHECK1-NEXT: [[TMP126:%.*]] = bitcast i8** [[TMP125]] to i32**
// CHECK1-NEXT: store i32* [[TMP101]], i32** [[TMP126]], align 8
// CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK1-NEXT: [[TMP128:%.*]] = bitcast i8** [[TMP127]] to i32**
// CHECK1-NEXT: store i32* [[TMP101]], i32** [[TMP128]], align 8
// CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP129]], align 8
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 2
// CHECK1-NEXT: [[TMP131:%.*]] = bitcast i8** [[TMP130]] to i32**
// CHECK1-NEXT: store i32* [[TMP102]], i32** [[TMP131]], align 8
// CHECK1-NEXT: [[TMP132:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 2
// CHECK1-NEXT: [[TMP133:%.*]] = bitcast i8** [[TMP132]] to i32**
// CHECK1-NEXT: store i32* [[TMP102]], i32** [[TMP133]], align 8
// CHECK1-NEXT: [[TMP134:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP134]], align 8
// CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 3
// CHECK1-NEXT: [[TMP136:%.*]] = bitcast i8** [[TMP135]] to i32**
// CHECK1-NEXT: store i32* [[TMP103]], i32** [[TMP136]], align 8
// CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 3
// CHECK1-NEXT: [[TMP138:%.*]] = bitcast i8** [[TMP137]] to i32**
// CHECK1-NEXT: store i32* [[TMP103]], i32** [[TMP138]], align 8
// CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 3
// CHECK1-NEXT: store i8* null, i8** [[TMP139]], align 8
// CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 4
// CHECK1-NEXT: [[TMP141:%.*]] = bitcast i8** [[TMP140]] to i32**
// CHECK1-NEXT: store i32* [[A]], i32** [[TMP141]], align 8
// CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 4
// CHECK1-NEXT: [[TMP143:%.*]] = bitcast i8** [[TMP142]] to i32**
// CHECK1-NEXT: store i32* [[A]], i32** [[TMP143]], align 8
// CHECK1-NEXT: [[TMP144:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 4
// CHECK1-NEXT: store i8* null, i8** [[TMP144]], align 8
// CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 5
// CHECK1-NEXT: [[TMP146:%.*]] = bitcast i8** [[TMP145]] to %class.anon**
// CHECK1-NEXT: store %class.anon* [[TMP104]], %class.anon** [[TMP146]], align 8
// CHECK1-NEXT: [[TMP147:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 5
// CHECK1-NEXT: [[TMP148:%.*]] = bitcast i8** [[TMP147]] to %class.anon**
// CHECK1-NEXT: store %class.anon* [[TMP104]], %class.anon** [[TMP148]], align 8
// CHECK1-NEXT: [[TMP149:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 5
// CHECK1-NEXT: store i8* null, i8** [[TMP149]], align 8
// CHECK1-NEXT: [[TMP150:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 6
// CHECK1-NEXT: [[TMP151:%.*]] = bitcast i8** [[TMP150]] to i32***
// CHECK1-NEXT: store i32** [[TMP105]], i32*** [[TMP151]], align 8
// CHECK1-NEXT: [[TMP152:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 6
// CHECK1-NEXT: [[TMP153:%.*]] = bitcast i8** [[TMP152]] to i32**
// CHECK1-NEXT: store i32* [[TMP107]], i32** [[TMP153]], align 8
// CHECK1-NEXT: [[TMP154:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 6
// CHECK1-NEXT: store i8* null, i8** [[TMP154]], align 8
// CHECK1-NEXT: [[TMP155:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 7
// CHECK1-NEXT: [[TMP156:%.*]] = bitcast i8** [[TMP155]] to i32***
// CHECK1-NEXT: store i32** [[TMP108]], i32*** [[TMP156]], align 8
// CHECK1-NEXT: [[TMP157:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 7
// CHECK1-NEXT: [[TMP158:%.*]] = bitcast i8** [[TMP157]] to i32**
// CHECK1-NEXT: store i32* [[TMP110]], i32** [[TMP158]], align 8
// CHECK1-NEXT: [[TMP159:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 7
// CHECK1-NEXT: store i8* null, i8** [[TMP159]], align 8
// CHECK1-NEXT: [[TMP160:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 8
// CHECK1-NEXT: [[TMP161:%.*]] = bitcast i8** [[TMP160]] to i32***
// CHECK1-NEXT: store i32** [[TMP111]], i32*** [[TMP161]], align 8
// CHECK1-NEXT: [[TMP162:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 8
// CHECK1-NEXT: [[TMP163:%.*]] = bitcast i8** [[TMP162]] to i32**
// CHECK1-NEXT: store i32* [[TMP113]], i32** [[TMP163]], align 8
// CHECK1-NEXT: [[TMP164:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 8
// CHECK1-NEXT: store i8* null, i8** [[TMP164]], align 8
// CHECK1-NEXT: [[TMP165:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 9
// CHECK1-NEXT: [[TMP166:%.*]] = bitcast i8** [[TMP165]] to i32****
// CHECK1-NEXT: store i32*** [[TMP114]], i32**** [[TMP166]], align 8
// CHECK1-NEXT: [[TMP167:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 9
// CHECK1-NEXT: [[TMP168:%.*]] = bitcast i8** [[TMP167]] to i32***
// CHECK1-NEXT: store i32** [[TMP116]], i32*** [[TMP168]], align 8
// CHECK1-NEXT: [[TMP169:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 9
// CHECK1-NEXT: store i8* null, i8** [[TMP169]], align 8
// CHECK1-NEXT: [[TMP170:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 10
// CHECK1-NEXT: [[TMP171:%.*]] = bitcast i8** [[TMP170]] to i32***
// CHECK1-NEXT: store i32** [[TMP117]], i32*** [[TMP171]], align 8
// CHECK1-NEXT: [[TMP172:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 10
// CHECK1-NEXT: [[TMP173:%.*]] = bitcast i8** [[TMP172]] to i32**
// CHECK1-NEXT: store i32* [[TMP119]], i32** [[TMP173]], align 8
// CHECK1-NEXT: [[TMP174:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 10
// CHECK1-NEXT: store i8* null, i8** [[TMP174]], align 8
// CHECK1-NEXT: [[TMP175:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP176:%.*]] = getelementptr inbounds [11 x i8*], [11 x i8*]* [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP177:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP177]], align 4
// CHECK1-NEXT: [[TMP178:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 1
// CHECK1-NEXT: store i32 11, i32* [[TMP178]], align 4
// CHECK1-NEXT: [[TMP179:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP175]], i8*** [[TMP179]], align 8
// CHECK1-NEXT: [[TMP180:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP176]], i8*** [[TMP180]], align 8
// CHECK1-NEXT: [[TMP181:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_sizes.1, i32 0, i32 0), i64** [[TMP181]], align 8
// CHECK1-NEXT: [[TMP182:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([11 x i64], [11 x i64]* @.offload_maptypes.2, i32 0, i32 0), i64** [[TMP182]], align 8
// CHECK1-NEXT: [[TMP183:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP183]], align 8
// CHECK1-NEXT: [[TMP184:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP184]], align 8
// CHECK1-NEXT: [[TMP185:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB1]], i64 -1, i32 1, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS10]])
// CHECK1-NEXT: [[TMP186:%.*]] = icmp ne i32 [[TMP185]], 0
// CHECK1-NEXT: br i1 [[TMP186]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK1: omp_offload.failed11:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43(i32* [[ARGC_ADDR]], i32* [[TMP101]], i32* [[TMP102]], i32* [[TMP103]], i32* [[A]], %class.anon* [[TMP104]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK1: omp_offload.cont12:
// CHECK1-NEXT: [[TMP187:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN1S3fooEv(%struct.S* noundef nonnull align 4 dereferenceable(4) @s)
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP187]], [[CALL]]
// CHECK1-NEXT: ret i32 [[ADD]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK1-SAME: (i64 noundef [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK1-NEXT: [[_TMP4:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[B5:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C7:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP8:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: store i64 [[ARGC]], i64* [[ARGC_ADDR]], align 8
// CHECK1-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK1-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK1-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK1-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_ADDR]] to i32*
// CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: store i32* [[TMP0]], i32** [[TMP]], align 8
// CHECK1-NEXT: store i32* [[TMP1]], i32** [[_TMP1]], align 8
// CHECK1-NEXT: store %class.anon* [[TMP3]], %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast %class.anon* [[L3]] to i8*
// CHECK1-NEXT: [[TMP6:%.*]] = bitcast %class.anon* [[TMP4]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP5]], i8* align 8 [[TMP6]], i64 40, i1 false)
// CHECK1-NEXT: store %class.anon* [[L3]], %class.anon** [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], i32* [[B5]], align 4
// CHECK1-NEXT: store i32* [[B5]], i32** [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK1-NEXT: store i32 [[TMP10]], i32* [[C7]], align 4
// CHECK1-NEXT: store i32* [[C7]], i32** [[_TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[TMP11]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK1-SAME: (i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK1-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK1-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK1-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK1-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK1-NEXT: store %class.anon* [[TMP4]], %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*, i32*, i32*, %class.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[TMP0]], i32* [[TMP5]], i32* [[TMP6]], i32* [[TMP7]], i32* [[TMP3]], %class.anon* [[TMP8]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK1-NEXT: [[_TMP4:%.*]] = alloca %class.anon*, align 8
// CHECK1-NEXT: [[ARGC5:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[B6:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP7:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[C8:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP9:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A10:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK1-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK1-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK1-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK1-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK1-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK1-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK1-NEXT: store %class.anon* [[TMP4]], %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = bitcast %class.anon* [[L3]] to i8*
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast %class.anon* [[TMP5]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 [[TMP7]], i64 40, i1 false)
// CHECK1-NEXT: store %class.anon* [[L3]], %class.anon** [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], i32* [[ARGC5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK1-NEXT: store i32 [[TMP10]], i32* [[B6]], align 4
// CHECK1-NEXT: store i32* [[B6]], i32** [[_TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK1-NEXT: store i32 [[TMP12]], i32* [[C8]], align 4
// CHECK1-NEXT: store i32* [[C8]], i32** [[_TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK1-NEXT: store i32 [[TMP13]], i32* [[A10]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[TMP14]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1S3fooEv
// CHECK1-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR3:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[L:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK1-NEXT: store %struct.S* [[THIS1]], %struct.S** [[TMP0]], align 8
// CHECK1-NEXT: store %class.anon.0* [[REF_TMP]], %class.anon.0** [[L]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to %struct.S**
// CHECK1-NEXT: store %struct.S* [[THIS1]], %struct.S** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to %struct.S**
// CHECK1-NEXT: store %struct.S* [[THIS1]], %struct.S** [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8** [[TMP10]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP2]], %class.anon.0** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP2]], %class.anon.0** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP16:%.*]] = bitcast i8** [[TMP15]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP3]], %struct.S*** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP4]], %struct.S*** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP22]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, i32* [[TMP23]], align 4
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP20]], i8*** [[TMP24]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP21]], i8*** [[TMP25]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.3, i32 0, i32 0), i64** [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.4, i32 0, i32 0), i64** [[TMP27]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB1]], i64 -1, i32 -1, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
// CHECK1-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27(%struct.S* [[THIS1]], %class.anon.0* [[TMP2]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP32:%.*]] = load %class.anon.0*, %class.anon.0** [[L]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP32]], %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP33]], i32 0, i32 0
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP33]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = bitcast i8** [[TMP36]] to %struct.S**
// CHECK1-NEXT: store %struct.S* [[THIS1]], %struct.S** [[TMP37]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to %struct.S**
// CHECK1-NEXT: store %struct.S* [[THIS1]], %struct.S** [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP40]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 1
// CHECK1-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP33]], %class.anon.0** [[TMP42]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 1
// CHECK1-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP33]], %class.anon.0** [[TMP44]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP45]], align 8
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 2
// CHECK1-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP34]], %struct.S*** [[TMP47]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP49:%.*]] = bitcast i8** [[TMP48]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP35]], %struct.S*** [[TMP49]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 2
// CHECK1-NEXT: store i8* null, i8** [[TMP50]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS6:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP53]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, i32* [[TMP54]], align 4
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP51]], i8*** [[TMP55]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP52]], i8*** [[TMP56]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.6, i32 0, i32 0), i64** [[TMP57]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.7, i32 0, i32 0), i64** [[TMP58]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP59]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP60]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB1]], i64 -1, i32 1, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS6]])
// CHECK1-NEXT: [[TMP62:%.*]] = icmp ne i32 [[TMP61]], 0
// CHECK1-NEXT: br i1 [[TMP62]], label [[OMP_OFFLOAD_FAILED7:%.*]], label [[OMP_OFFLOAD_CONT8:%.*]]
// CHECK1: omp_offload.failed7:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29(%struct.S* [[THIS1]], %class.anon.0* [[TMP33]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT8]]
// CHECK1: omp_offload.cont8:
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP63:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: [[TMP64:%.*]] = load %class.anon.0*, %class.anon.0** [[L]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z3fooIZN1S3fooEvEUlvE_EiRKT_(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP64]])
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP63]], [[CALL]]
// CHECK1-NEXT: ret i32 [[ADD]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK1-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast %class.anon.0* [[L1]] to i8*
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast %class.anon.0* [[TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP3]], i8* align 8 [[TMP4]], i64 8, i1 false)
// CHECK1-NEXT: store %class.anon.0* [[L1]], %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP5]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK1-SAME: (%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR3]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP1:%.*]] = load %struct.S*, %struct.S** [[TMP0]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[A]], align 4
// CHECK1-NEXT: ret i32 [[TMP2]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29
// CHECK1-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S*, %class.anon.0*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), %struct.S* [[TMP0]], %class.anon.0* [[TMP2]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast %class.anon.0* [[L1]] to i8*
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast %class.anon.0* [[TMP2]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP3]], i8* align 8 [[TMP4]], i64 8, i1 false)
// CHECK1-NEXT: store %class.anon.0* [[L1]], %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP5]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z3fooIZN1S3fooEvEUlvE_EiRKT_
// CHECK1-SAME: (%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT: store %class.anon.0* [[T]], %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %class.anon.0*, %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP0]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[TMP1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to %class.anon.0**
// CHECK1-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP2]], %struct.S*** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to %struct.S***
// CHECK1-NEXT: store %struct.S** [[TMP3]], %struct.S*** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store i8* null, i8** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP16]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 2, i32* [[TMP17]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP14]], i8*** [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP15]], i8*** [[TMP19]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.9, i32 0, i32 0), i64** [[TMP20]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.10, i32 0, i32 0), i64** [[TMP21]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP22]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB1]], i64 -1, i32 1, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK1-NEXT: br i1 [[TMP25]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18(%class.anon.0* [[TMP1]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
// CHECK1-SAME: (%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store %class.anon.0* [[T]], %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %class.anon.0*, %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP0]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %class.anon.0*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), %class.anon.0* [[TMP1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[T_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[T]], %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load %class.anon.0*, %class.anon.0** [[T_ADDR]], align 8
// CHECK1-NEXT: store %class.anon.0* [[TMP0]], %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = bitcast %class.anon.0* [[T1]] to i8*
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast %class.anon.0* [[TMP1]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP2]], i8* align 8 [[TMP3]], i64 8, i1 false)
// CHECK1-NEXT: store %class.anon.0* [[T1]], %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP4]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK2-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[TMP1]], %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP3:%.*]] = load %class.anon*, %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = bitcast %class.anon* [[L1]] to i8*
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast %class.anon* [[TMP3]] to i8*
// CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP4]], i8* align 8 [[TMP5]], i64 8, i1 false)
// CHECK2-NEXT: store %class.anon* [[L1]], %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP6]], i32 0, i32 0
// CHECK2-NEXT: store %struct.S* [[TMP0]], %struct.S** [[TMP7]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon* noundef nonnull align 8 dereferenceable(8) [[TMP8]]) #[[ATTR7:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK2-SAME: (%class.anon* noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: store %class.anon* [[THIS]], %class.anon** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[THIS1:%.*]] = load %class.anon*, %class.anon** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON:%.*]], %class.anon* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP1:%.*]] = load %struct.S*, %struct.S** [[TMP0]], align 8
// CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[A]], align 4
// CHECK2-NEXT: ret i32 [[TMP2]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29
// CHECK2-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[TMP1]], %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK2-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: [[TMP6:%.*]] = bitcast %struct.S* [[TMP0]] to i8*
// CHECK2-NEXT: store i8* [[TMP6]], i8** [[TMP5]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast %class.anon* [[TMP4]] to i8*
// CHECK2-NEXT: store i8* [[TMP8]], i8** [[TMP7]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK2-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, %struct.S*, %class.anon*)* @__omp_outlined__ to i8*), i8* null, i8** [[TMP9]], i64 2)
// CHECK2-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.S* noundef [[THIS:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[TMP1]], %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load %class.anon*, %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast %class.anon* [[L1]] to i8*
// CHECK2-NEXT: [[TMP4:%.*]] = bitcast %class.anon* [[TMP2]] to i8*
// CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP3]], i8* align 8 [[TMP4]], i64 8, i1 false)
// CHECK2-NEXT: store %class.anon* [[L1]], %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT: store %struct.S* [[TMP0]], %struct.S** [[TMP6]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon* noundef nonnull align 8 dereferenceable(8) [[TMP7]]) #[[ATTR7]]
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK2-SAME: (i64 noundef [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK2-NEXT: [[_TMP4:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[B5:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[C7:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP8:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: store i64 [[ARGC]], i64* [[ARGC_ADDR]], align 8
// CHECK2-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK2-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK2-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK2-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK2-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_ADDR]] to i32*
// CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: store i32* [[TMP0]], i32** [[TMP]], align 8
// CHECK2-NEXT: store i32* [[TMP1]], i32** [[_TMP1]], align 8
// CHECK2-NEXT: store %class.anon.0* [[TMP3]], %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = bitcast %class.anon.0* [[L3]] to i8*
// CHECK2-NEXT: [[TMP7:%.*]] = bitcast %class.anon.0* [[TMP5]] to i8*
// CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 [[TMP7]], i64 40, i1 false)
// CHECK2-NEXT: store %class.anon.0* [[L3]], %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], i32* [[B5]], align 4
// CHECK2-NEXT: store i32* [[B5]], i32** [[_TMP6]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
// CHECK2-NEXT: store i32 [[TMP11]], i32* [[C7]], align 4
// CHECK2-NEXT: store i32* [[C7]], i32** [[_TMP8]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP12]], i32 0, i32 0
// CHECK2-NEXT: store i32* [[CONV]], i32** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP12]], i32 0, i32 1
// CHECK2-NEXT: [[TMP15:%.*]] = load i32*, i32** [[_TMP6]], align 8
// CHECK2-NEXT: store i32* [[TMP15]], i32** [[TMP14]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP12]], i32 0, i32 2
// CHECK2-NEXT: [[TMP17:%.*]] = load i32*, i32** [[_TMP8]], align 8
// CHECK2-NEXT: store i32* [[TMP17]], i32** [[TMP16]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP12]], i32 0, i32 3
// CHECK2-NEXT: store i32** [[D_ADDR]], i32*** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP12]], i32 0, i32 4
// CHECK2-NEXT: store i32* [[TMP2]], i32** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK2-SAME: (i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [6 x i8*], align 8
// CHECK2-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK2-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK2-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK2-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK2-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK2-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK2-NEXT: store %class.anon.0* [[TMP4]], %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
// CHECK2-NEXT: [[TMP7:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: [[TMP12:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK2-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: [[TMP14:%.*]] = bitcast i32* [[TMP7]] to i8*
// CHECK2-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP16:%.*]] = bitcast i32* [[TMP8]] to i8*
// CHECK2-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: [[TMP18:%.*]] = bitcast i32* [[TMP9]] to i8*
// CHECK2-NEXT: store i8* [[TMP18]], i8** [[TMP17]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP20:%.*]] = bitcast i32* [[TMP3]] to i8*
// CHECK2-NEXT: store i8* [[TMP20]], i8** [[TMP19]], align 8
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
// CHECK2-NEXT: [[TMP22:%.*]] = bitcast %class.anon.0* [[TMP10]] to i8*
// CHECK2-NEXT: store i8* [[TMP22]], i8** [[TMP21]], align 8
// CHECK2-NEXT: [[TMP23:%.*]] = bitcast [6 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK2-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, i32*, i32*, i32*, i32*, %class.anon.0*)* @__omp_outlined__1 to i8*), i8* null, i8** [[TMP23]], i64 6)
// CHECK2-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK2-NEXT: [[_TMP4:%.*]] = alloca %class.anon.0*, align 8
// CHECK2-NEXT: [[ARGC5:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B6:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP7:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[C8:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP9:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A10:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK2-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK2-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK2-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK2-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK2-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK2-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK2-NEXT: store %class.anon.0* [[TMP4]], %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = bitcast %class.anon.0* [[L3]] to i8*
// CHECK2-NEXT: [[TMP7:%.*]] = bitcast %class.anon.0* [[TMP5]] to i8*
// CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 [[TMP7]], i64 40, i1 false)
// CHECK2-NEXT: store %class.anon.0* [[L3]], %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: store i32 [[TMP8]], i32* [[ARGC5]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK2-NEXT: store i32 [[TMP10]], i32* [[B6]], align 4
// CHECK2-NEXT: store i32* [[B6]], i32** [[_TMP7]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT: store i32 [[TMP12]], i32* [[C8]], align 4
// CHECK2-NEXT: store i32* [[C8]], i32** [[_TMP9]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK2-NEXT: store i32 [[TMP13]], i32* [[A10]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP14]], i32 0, i32 0
// CHECK2-NEXT: store i32* [[ARGC5]], i32** [[TMP15]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP14]], i32 0, i32 1
// CHECK2-NEXT: [[TMP17:%.*]] = load i32*, i32** [[_TMP7]], align 8
// CHECK2-NEXT: store i32* [[TMP17]], i32** [[TMP16]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP14]], i32 0, i32 2
// CHECK2-NEXT: [[TMP19:%.*]] = load i32*, i32** [[_TMP9]], align 8
// CHECK2-NEXT: store i32* [[TMP19]], i32** [[TMP18]], align 8
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP14]], i32 0, i32 3
// CHECK2-NEXT: store i32** [[D_ADDR]], i32*** [[TMP20]], align 8
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP14]], i32 0, i32 4
// CHECK2-NEXT: store i32* [[A10]], i32** [[TMP21]], align 8
// CHECK2-NEXT: [[TMP22:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP4]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(40) [[TMP22]]) #[[ATTR7]]
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
// CHECK2-SAME: (%class.anon* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[T_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: store %class.anon* [[T]], %class.anon** [[T_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %class.anon*, %class.anon** [[T_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[TMP0]], %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
// CHECK2-NEXT: [[TMP3:%.*]] = load %class.anon*, %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: [[TMP5:%.*]] = bitcast %class.anon* [[TMP3]] to i8*
// CHECK2-NEXT: store i8* [[TMP5]], i8** [[TMP4]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = bitcast [1 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK2-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, %class.anon*)* @__omp_outlined__2 to i8*), i8* null, i8** [[TMP6]], i64 1)
// CHECK2-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK2-NEXT: ret void
// CHECK2: worker.exit:
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK2-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[T_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[T]], %class.anon** [[T_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load %class.anon*, %class.anon** [[T_ADDR]], align 8
// CHECK2-NEXT: store %class.anon* [[TMP0]], %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load %class.anon*, %class.anon** [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = bitcast %class.anon* [[T1]] to i8*
// CHECK2-NEXT: [[TMP3:%.*]] = bitcast %class.anon* [[TMP1]] to i8*
// CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP2]], i8* align 8 [[TMP3]], i64 8, i1 false)
// CHECK2-NEXT: store %class.anon* [[T1]], %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon* noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]]
// CHECK2-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK3-SAME: (i64 noundef [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK3-NEXT: [[_TMP4:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[B5:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[C7:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: store i64 [[ARGC]], i64* [[ARGC_ADDR]], align 8
// CHECK3-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK3-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK3-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK3-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK3-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: [[CONV:%.*]] = bitcast i64* [[ARGC_ADDR]] to i32*
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: store i32* [[TMP0]], i32** [[TMP]], align 8
// CHECK3-NEXT: store i32* [[TMP1]], i32** [[_TMP1]], align 8
// CHECK3-NEXT: store %class.anon* [[TMP3]], %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP5:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = bitcast %class.anon* [[L3]] to i8*
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast %class.anon* [[TMP5]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 [[TMP7]], i64 40, i1 false)
// CHECK3-NEXT: store %class.anon* [[L3]], %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], i32* [[B5]], align 4
// CHECK3-NEXT: store i32* [[B5]], i32** [[_TMP6]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
// CHECK3-NEXT: store i32 [[TMP11]], i32* [[C7]], align 4
// CHECK3-NEXT: store i32* [[C7]], i32** [[_TMP8]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP12]], i32 0, i32 0
// CHECK3-NEXT: store i32* [[CONV]], i32** [[TMP13]], align 8
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP12]], i32 0, i32 1
// CHECK3-NEXT: [[TMP15:%.*]] = load i32*, i32** [[_TMP6]], align 8
// CHECK3-NEXT: store i32* [[TMP15]], i32** [[TMP14]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP12]], i32 0, i32 2
// CHECK3-NEXT: [[TMP17:%.*]] = load i32*, i32** [[_TMP8]], align 8
// CHECK3-NEXT: store i32* [[TMP17]], i32** [[TMP16]], align 8
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP12]], i32 0, i32 3
// CHECK3-NEXT: store i32** [[D_ADDR]], i32*** [[TMP18]], align 8
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP12]], i32 0, i32 4
// CHECK3-NEXT: store i32* [[TMP2]], i32** [[TMP19]], align 8
// CHECK3-NEXT: [[TMP20:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7:[0-9]+]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: ret void
// CHECK3: worker.exit:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK3-SAME: (i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [6 x i8*], align 8
// CHECK3-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK3-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK3-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK3-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK3-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK3-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK3-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK3-NEXT: store %class.anon* [[TMP4]], %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
// CHECK3-NEXT: [[TMP7:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: [[TMP12:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK3-NEXT: store i8* [[TMP12]], i8** [[TMP11]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK3-NEXT: [[TMP14:%.*]] = bitcast i32* [[TMP7]] to i8*
// CHECK3-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK3-NEXT: [[TMP16:%.*]] = bitcast i32* [[TMP8]] to i8*
// CHECK3-NEXT: store i8* [[TMP16]], i8** [[TMP15]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK3-NEXT: [[TMP18:%.*]] = bitcast i32* [[TMP9]] to i8*
// CHECK3-NEXT: store i8* [[TMP18]], i8** [[TMP17]], align 8
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK3-NEXT: [[TMP20:%.*]] = bitcast i32* [[TMP3]] to i8*
// CHECK3-NEXT: store i8* [[TMP20]], i8** [[TMP19]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x i8*], [6 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
// CHECK3-NEXT: [[TMP22:%.*]] = bitcast %class.anon* [[TMP10]] to i8*
// CHECK3-NEXT: store i8* [[TMP22]], i8** [[TMP21]], align 8
// CHECK3-NEXT: [[TMP23:%.*]] = bitcast [6 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK3-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, i32*, i32*, i32*, i32*, i32*, %class.anon*)* @__omp_outlined__ to i8*), i8* null, i8** [[TMP23]], i64 6)
// CHECK3-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK3-NEXT: ret void
// CHECK3: worker.exit:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[B:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[C:%.*]], i32* noundef [[D:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]], %class.anon* noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP1:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[L3:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK3-NEXT: [[_TMP4:%.*]] = alloca %class.anon*, align 8
// CHECK3-NEXT: [[ARGC5:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[B6:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP7:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[C8:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP9:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[A10:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK3-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK3-NEXT: store i32* [[B]], i32** [[B_ADDR]], align 8
// CHECK3-NEXT: store i32* [[C]], i32** [[C_ADDR]], align 8
// CHECK3-NEXT: store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK3-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8
// CHECK3-NEXT: store %class.anon* [[L]], %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load i32*, i32** [[B_ADDR]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load i32*, i32** [[C_ADDR]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load i32*, i32** [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load %class.anon*, %class.anon** [[L_ADDR]], align 8
// CHECK3-NEXT: store i32* [[TMP1]], i32** [[TMP]], align 8
// CHECK3-NEXT: store i32* [[TMP2]], i32** [[_TMP1]], align 8
// CHECK3-NEXT: store %class.anon* [[TMP4]], %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load %class.anon*, %class.anon** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = bitcast %class.anon* [[L3]] to i8*
// CHECK3-NEXT: [[TMP7:%.*]] = bitcast %class.anon* [[TMP5]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 [[TMP7]], i64 40, i1 false)
// CHECK3-NEXT: store %class.anon* [[L3]], %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK3-NEXT: store i32 [[TMP8]], i32* [[ARGC5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK3-NEXT: store i32 [[TMP10]], i32* [[B6]], align 4
// CHECK3-NEXT: store i32* [[B6]], i32** [[_TMP7]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = load i32*, i32** [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK3-NEXT: store i32 [[TMP12]], i32* [[C8]], align 4
// CHECK3-NEXT: store i32* [[C8]], i32** [[_TMP9]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK3-NEXT: store i32 [[TMP13]], i32* [[A10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP14]], i32 0, i32 0
// CHECK3-NEXT: store i32* [[ARGC5]], i32** [[TMP15]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP14]], i32 0, i32 1
// CHECK3-NEXT: [[TMP17:%.*]] = load i32*, i32** [[_TMP7]], align 8
// CHECK3-NEXT: store i32* [[TMP17]], i32** [[TMP16]], align 8
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP14]], i32 0, i32 2
// CHECK3-NEXT: [[TMP19:%.*]] = load i32*, i32** [[_TMP9]], align 8
// CHECK3-NEXT: store i32* [[TMP19]], i32** [[TMP18]], align 8
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP14]], i32 0, i32 3
// CHECK3-NEXT: store i32** [[D_ADDR]], i32*** [[TMP20]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[TMP14]], i32 0, i32 4
// CHECK3-NEXT: store i32* [[A10]], i32** [[TMP21]], align 8
// CHECK3-NEXT: [[TMP22:%.*]] = load %class.anon*, %class.anon** [[_TMP4]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 8 dereferenceable(40) [[TMP22]]) #[[ATTR7]]
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK3-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP3:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = bitcast %class.anon.0* [[L1]] to i8*
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast %class.anon.0* [[TMP3]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP4]], i8* align 8 [[TMP5]], i64 8, i1 false)
// CHECK3-NEXT: store %class.anon.0* [[L1]], %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP6]], i32 0, i32 0
// CHECK3-NEXT: store %struct.S* [[TMP0]], %struct.S** [[TMP7]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP8]]) #[[ATTR7]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: ret void
// CHECK3: worker.exit:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK3-SAME: (%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP1:%.*]] = load %struct.S*, %struct.S** [[TMP0]], align 8
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[A]], align 4
// CHECK3-NEXT: ret i32 [[TMP2]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29
// CHECK3-SAME: (%struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x i8*], align 8
// CHECK3-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
// CHECK3-NEXT: [[TMP4:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: [[TMP6:%.*]] = bitcast %struct.S* [[TMP0]] to i8*
// CHECK3-NEXT: store i8* [[TMP6]], i8** [[TMP5]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK3-NEXT: [[TMP8:%.*]] = bitcast %class.anon.0* [[TMP4]] to i8*
// CHECK3-NEXT: store i8* [[TMP8]], i8** [[TMP7]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = bitcast [2 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK3-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, %struct.S*, %class.anon.0*)* @__omp_outlined__1 to i8*), i8* null, i8** [[TMP9]], i64 2)
// CHECK3-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK3-NEXT: ret void
// CHECK3: worker.exit:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.S* noundef [[THIS:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK3-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[L]], %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[L_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[TMP1]], %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast %class.anon.0* [[L1]] to i8*
// CHECK3-NEXT: [[TMP4:%.*]] = bitcast %class.anon.0* [[TMP2]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP3]], i8* align 8 [[TMP4]], i64 8, i1 false)
// CHECK3-NEXT: store %class.anon.0* [[L1]], %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[TMP5]], i32 0, i32 0
// CHECK3-NEXT: store %struct.S* [[TMP0]], %struct.S** [[TMP6]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP7]]) #[[ATTR7]]
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
// CHECK3-SAME: (%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x i8*], align 8
// CHECK3-NEXT: store %class.anon.0* [[T]], %class.anon.0** [[T_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %class.anon.0*, %class.anon.0** [[T_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[TMP0]], %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
// CHECK3-NEXT: [[TMP3:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast %class.anon.0* [[TMP3]] to i8*
// CHECK3-NEXT: store i8* [[TMP5]], i8** [[TMP4]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = bitcast [1 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
// CHECK3-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*, %class.anon.0*)* @__omp_outlined__2 to i8*), i8* null, i8** [[TMP6]], i64 1)
// CHECK3-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
// CHECK3-NEXT: ret void
// CHECK3: worker.exit:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %class.anon.0* noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT: [[T_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca %class.anon.0*, align 8
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[T]], %class.anon.0** [[T_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load %class.anon.0*, %class.anon.0** [[T_ADDR]], align 8
// CHECK3-NEXT: store %class.anon.0* [[TMP0]], %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load %class.anon.0*, %class.anon.0** [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = bitcast %class.anon.0* [[T1]] to i8*
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast %class.anon.0* [[TMP1]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP2]], i8* align 8 [[TMP3]], i64 8, i1 false)
// CHECK3-NEXT: store %class.anon.0* [[T1]], %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load %class.anon.0*, %class.anon.0** [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]]
// CHECK3-NEXT: ret void
//
Reference in New Issue View Git Blame Copy Permalink