shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
llvm-project/clang/test/OpenMP/target_task_affinity_codegen.cpp
Nikita Popov cca5bb52f3
[OpenMP] Use CreatePtrDiff() (#184127) 
This will use ptrtoaddr instead of ptrtoint in the pointer difference
calculation, avoiding unnecessary provenance exposure.

Add an IsNUW flag to allow specifying nuw on the subtraction.
2026-03-02 14:34:40 +00:00

865 lines
67 KiB
C++
Raw Blame History

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --check-globals --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --global-value-regex "\.offload_.*"
// Test host codegen.
// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7
// Test target codegen
// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix=CHECK7
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-target-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
int main() {
int *A;
int *B;
#pragma omp target data map(tofrom: A[0:1024]) // (1)
#pragma omp target data use_device_ptr(A) // (2)
{
#pragma omp target defaultmap(none) is_device_ptr(A) map(tofrom: B[0:1024]) // (3)
{
#pragma omp task shared(B) affinity(A[0:1024])
{
for (int i = 0; i < 1024; i++) {
B[i] = 2 * A[i];
}
}
#pragma omp taskwait
}
}
return 0;
}
#endif
// Expected Maps:
// (1):
// &A[0], &A[/*lb=*/0], 1024 * sizeof(A[0]), TO | FROM
// &A, &A[/*lb=*/0], sizeof(A), ATTACH
// (2):
// &A[0], &A[0], 0, RETURN_PARAM
// (3):
// &B[0], &B[/*lb=*/0], 1024 * sizeof(B[0]), TO | FROM | PARAM
// &B, &B[/*lb=*/0], sizeof(B), ATTACH
// &A[0], &A[0], sizeof(A[0]), LITERAL | PARAM
//.
// CHECK1: @.offload_sizes = private unnamed_addr constant [2 x i64] [i64 4096, i64 8]
// CHECK1: @.offload_maptypes = private unnamed_addr constant [2 x i64] [i64 3, i64 16384]
// CHECK1: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] zeroinitializer
// CHECK1: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 64]
// CHECK1: @.offload_sizes.3 = private unnamed_addr constant [3 x i64] [i64 4096, i64 8, i64 8]
// CHECK1: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 35, i64 16384, i64 288]
//.
// CHECK3: @.offload_sizes = private unnamed_addr constant [2 x i64] [i64 4096, i64 4]
// CHECK3: @.offload_maptypes = private unnamed_addr constant [2 x i64] [i64 3, i64 16384]
// CHECK3: @.offload_sizes.1 = private unnamed_addr constant [1 x i64] zeroinitializer
// CHECK3: @.offload_maptypes.2 = private unnamed_addr constant [1 x i64] [i64 64]
// CHECK3: @.offload_sizes.3 = private unnamed_addr constant [3 x i64] [i64 4096, i64 4, i64 4]
// CHECK3: @.offload_maptypes.4 = private unnamed_addr constant [3 x i64] [i64 35, i64 16384, i64 288]
//.
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[B:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A]], align 8
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAYIDX]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: call void @__tgt_target_data_begin_mapper(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 2, ptr [[TMP9]], ptr [[TMP10]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr null, ptr null)
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[A]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: call void @__tgt_target_data_begin_mapper(ptr @[[GLOB1]], i64 -1, i32 1, ptr [[TMP15]], ptr [[TMP16]], ptr @.offload_sizes.1, ptr @.offload_maptypes.2, ptr null, ptr null)
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[B]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[B]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = load ptr, ptr [[B]], align 8
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP21]], i64 0
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARRAYIDX4]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[B]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX4]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 3, ptr [[TMP33]], align 4
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP34]], align 4
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP31]], ptr [[TMP35]], align 8
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP32]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP40]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP41]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT: store i64 0, ptr [[TMP42]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP43]], align 4
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP44]], align 4
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT: store i32 0, ptr [[TMP45]], align 4
// CHECK1-NEXT: [[TMP46:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP47:%.*]] = icmp ne i32 [[TMP46]], 0
// CHECK1-NEXT: br i1 [[TMP47]], 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_l44(ptr [[TMP18]], ptr [[TMP19]]) #[[ATTR1:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: call void @__tgt_target_data_end_mapper(ptr @[[GLOB1]], i64 -1, i32 1, ptr [[TMP48]], ptr [[TMP49]], ptr @.offload_sizes.1, ptr @.offload_maptypes.2, ptr null, ptr null)
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: call void @__tgt_target_data_end_mapper(ptr @[[GLOB1]], i64 -1, i32 2, ptr [[TMP50]], ptr [[TMP51]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr null, ptr null)
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44
// CHECK1-SAME: (ptr noundef [[B:%.*]], ptr noundef [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK1-NEXT: [[DOTAFFS_ARR_ADDR:%.*]] = alloca [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T:%.*]]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[B_ADDR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 48, i64 8, ptr @.omp_task_entry.)
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T]]], ptr [[DOTAFFS_ARR_ADDR]], i64 0, i64 0
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP4]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP5]], i64 1023
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[ARRAYIDX1]], i32 1
// CHECK1-NEXT: [[TMP7:%.*]] = ptrtoaddr ptr [[TMP6]] to i64
// CHECK1-NEXT: [[TMP8:%.*]] = ptrtoaddr ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = sub nuw i64 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP3]], i64 0
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 0
// CHECK1-NEXT: [[TMP12:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT: store i64 [[TMP12]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 1
// CHECK1-NEXT: store i64 [[TMP9]], ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_omp_reg_task_with_affinity(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]], i32 1, ptr [[TMP3]])
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP17]], ptr align 8 [[AGG_CAPTURED]], i64 8, i1 false)
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP2]], i32 0, i32 1
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP18]], i32 0, i32 0
// CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]])
// CHECK1-NEXT: [[TMP22:%.*]] = call i32 @__kmpc_omp_taskwait(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK1-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[I_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META12:![0-9]+]])
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META14:![0-9]+]]
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR1]]
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META14]]
// CHECK1-NEXT: store i32 0, ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: br label [[FOR_COND_I:%.*]]
// CHECK1: for.cond.i:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: [[CMP_I:%.*]] = icmp slt i32 [[TMP13]], 1024
// CHECK1-NEXT: br i1 [[CMP_I]], label [[FOR_BODY_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
// CHECK1: for.body.i:
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: [[IDXPROM_I:%.*]] = sext i32 [[TMP15]] to i64
// CHECK1-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i64 [[IDXPROM_I]]
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX_I]], align 4
// CHECK1-NEXT: [[MUL_I:%.*]] = mul nsw i32 2, [[TMP16]]
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP9]], align 8, !nonnull [[META15:![0-9]+]], !align [[META16:![0-9]+]]
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: [[IDXPROM1_I:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i32, ptr [[TMP18]], i64 [[IDXPROM1_I]]
// CHECK1-NEXT: store i32 [[MUL_I]], ptr [[ARRAYIDX2_I]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: [[INC_I:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT: store i32 [[INC_I]], ptr [[I_I]], align 4, !noalias [[META14]]
// CHECK1-NEXT: br label [[FOR_COND_I]], !llvm.loop [[LOOP17:![0-9]+]]
// CHECK1: .omp_outlined..exit:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[A:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[B:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A]], align 4
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP2]], i32 0
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[ARRAYIDX]], ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[A]], ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[ARRAYIDX]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: call void @__tgt_target_data_begin_mapper(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 2, ptr [[TMP9]], ptr [[TMP10]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr null, ptr null)
// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[A]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK3-NEXT: call void @__tgt_target_data_begin_mapper(ptr @[[GLOB1]], i64 -1, i32 1, ptr [[TMP15]], ptr [[TMP16]], ptr @.offload_sizes.1, ptr @.offload_maptypes.2, ptr null, ptr null)
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP12]], align 4
// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[B]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = load ptr, ptr [[B]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = load ptr, ptr [[B]], align 4
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP21]], i32 0
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP20]], ptr [[TMP22]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[ARRAYIDX4]], ptr [[TMP23]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP24]], align 4
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[B]], ptr [[TMP25]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[ARRAYIDX4]], ptr [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP28]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 2
// CHECK3-NEXT: store ptr null, ptr [[TMP30]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 3, ptr [[TMP33]], align 4
// CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 3, ptr [[TMP34]], align 4
// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP31]], ptr [[TMP35]], align 4
// CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP32]], ptr [[TMP36]], align 4
// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP37]], align 4
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP38]], align 4
// CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP39]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP40]], align 4
// CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP41]], align 8
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT: store i64 0, ptr [[TMP42]], align 8
// CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT: store [3 x i32] [i32 -1, i32 0, i32 0], ptr [[TMP43]], align 4
// CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP44]], align 4
// CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT: store i32 0, ptr [[TMP45]], align 4
// CHECK3-NEXT: [[TMP46:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP47:%.*]] = icmp ne i32 [[TMP46]], 0
// CHECK3-NEXT: br i1 [[TMP47]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44(ptr [[TMP18]], ptr [[TMP19]]) #[[ATTR1:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK3-NEXT: call void @__tgt_target_data_end_mapper(ptr @[[GLOB1]], i64 -1, i32 1, ptr [[TMP48]], ptr [[TMP49]], ptr @.offload_sizes.1, ptr @.offload_maptypes.2, ptr null, ptr null)
// CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: call void @__tgt_target_data_end_mapper(ptr @[[GLOB1]], i64 -1, i32 2, ptr [[TMP50]], ptr [[TMP51]], ptr @.offload_sizes, ptr @.offload_maptypes, ptr null, ptr null)
// CHECK3-NEXT: ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44
// CHECK3-SAME: (ptr noundef [[B:%.*]], ptr noundef [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 4
// CHECK3-NEXT: [[DOTAFFS_ARR_ADDR:%.*]] = alloca [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T:%.*]]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[B_ADDR]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 24, i32 4, ptr @.omp_task_entry.)
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T]]], ptr [[DOTAFFS_ARR_ADDR]], i32 0, i32 0
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP4]], i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK3-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP5]], i32 1023
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[ARRAYIDX1]], i32 1
// CHECK3-NEXT: [[TMP7:%.*]] = ptrtoaddr ptr [[TMP6]] to i32
// CHECK3-NEXT: [[TMP8:%.*]] = ptrtoaddr ptr [[ARRAYIDX]] to i32
// CHECK3-NEXT: [[TMP9:%.*]] = sub nuw i32 [[TMP7]], [[TMP8]]
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP3]], i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 0
// CHECK3-NEXT: [[TMP12:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i32
// CHECK3-NEXT: store i32 [[TMP12]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 1
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_omp_reg_task_with_affinity(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]], i32 1, ptr [[TMP3]])
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP17]], ptr align 4 [[AGG_CAPTURED]], i32 4, i1 false)
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP2]], i32 0, i32 1
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP18]], i32 0, i32 0
// CHECK3-NEXT: [[TMP20:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]])
// CHECK3-NEXT: [[TMP22:%.*]] = call i32 @__kmpc_omp_taskwait(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK3-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 4
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK3-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[I_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK3-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META13:![0-9]+]])
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META15:![0-9]+]]
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR1]]
// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: store i32 0, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: br label [[FOR_COND_I:%.*]]
// CHECK3: for.cond.i:
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[CMP_I:%.*]] = icmp slt i32 [[TMP13]], 1024
// CHECK3-NEXT: br i1 [[CMP_I]], label [[FOR_BODY_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
// CHECK3: for.body.i:
// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i32 [[TMP15]]
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX_I]], align 4
// CHECK3-NEXT: [[MUL_I:%.*]] = mul nsw i32 2, [[TMP16]]
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP9]], align 4, !nonnull [[META16:![0-9]+]], !align [[META17:![0-9]+]]
// CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[ARRAYIDX1_I:%.*]] = getelementptr inbounds i32, ptr [[TMP18]], i32 [[TMP19]]
// CHECK3-NEXT: store i32 [[MUL_I]], ptr [[ARRAYIDX1_I]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: [[INC_I:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK3-NEXT: store i32 [[INC_I]], ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK3-NEXT: br label [[FOR_COND_I]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK3: .omp_outlined..exit:
// CHECK3-NEXT: ret i32 0
//
//
// CHECK5-LABEL: define {{[^@]+}}@main
// CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[B:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK5-NEXT: store i32 0, ptr [[I]], align 4
// CHECK5-NEXT: br label [[FOR_COND:%.*]]
// CHECK5: for.cond:
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[I]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP0]], 1024
// CHECK5-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK5: for.body:
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[I]], align 4
// CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP2]] to i64
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]]
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 2, [[TMP3]]
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[B]], align 8
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4
// CHECK5-NEXT: [[IDXPROM1:%.*]] = sext i32 [[TMP5]] to i64
// CHECK5-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 [[IDXPROM1]]
// CHECK5-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
// CHECK5-NEXT: br label [[FOR_INC:%.*]]
// CHECK5: for.inc:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4
// CHECK5-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK5-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK5-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK5: for.end:
// CHECK5-NEXT: ret i32 0
//
//
// CHECK7-LABEL: define {{[^@]+}}@main
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK7-NEXT: entry:
// CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[A:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[B:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK7-NEXT: store i32 0, ptr [[I]], align 4
// CHECK7-NEXT: br label [[FOR_COND:%.*]]
// CHECK7: for.cond:
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[I]], align 4
// CHECK7-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP0]], 1024
// CHECK7-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK7: for.body:
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[I]], align 4
// CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 [[TMP2]]
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 2, [[TMP3]]
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[B]], align 4
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4
// CHECK7-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 [[TMP5]]
// CHECK7-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX1]], align 4
// CHECK7-NEXT: br label [[FOR_INC:%.*]]
// CHECK7: for.inc:
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4
// CHECK7-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[INC]], ptr [[I]], align 4
// CHECK7-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK7: for.end:
// CHECK7-NEXT: ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44
// CHECK9-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], ptr noundef [[B:%.*]], ptr noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK9-NEXT: [[DOTAFFS_ARR_ADDR:%.*]] = alloca [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T:%.*]]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
// CHECK9-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 8
// CHECK9-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[B_ADDR]], ptr [[TMP1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i64 48, i64 8, ptr @.omp_task_entry.)
// CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T]]], ptr [[DOTAFFS_ARR_ADDR]], i64 0, i64 0
// CHECK9-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP4]], i64 0
// CHECK9-NEXT: [[TMP5:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP5]], i64 1023
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[ARRAYIDX1]], i32 1
// CHECK9-NEXT: [[TMP7:%.*]] = ptrtoaddr ptr [[TMP6]] to i64
// CHECK9-NEXT: [[TMP8:%.*]] = ptrtoaddr ptr [[ARRAYIDX]] to i64
// CHECK9-NEXT: [[TMP9:%.*]] = sub nuw i64 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP3]], i64 0
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK9-NEXT: store i64 [[TMP12]], ptr [[TMP11]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_omp_reg_task_with_affinity(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]], i32 1, ptr [[TMP3]])
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
// CHECK9-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
// CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP17]], ptr align 8 [[AGG_CAPTURED]], i64 8, i1 false)
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP2]], i32 0, i32 1
// CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP18]], i32 0, i32 0
// CHECK9-NEXT: [[TMP20:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]])
// CHECK9-NEXT: [[TMP22:%.*]] = call i32 @__kmpc_omp_taskwait(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK9-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK9-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK9-SAME: (i32 noundef signext [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[I_I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK9-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
// CHECK9-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK9-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK9-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META13:![0-9]+]])
// CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META15:![0-9]+]]
// CHECK9-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias [[META15]]
// CHECK9-NEXT: store i32 0, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: br label [[FOR_COND_I:%.*]]
// CHECK9: for.cond.i:
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: [[CMP_I:%.*]] = icmp slt i32 [[TMP13]], 1024
// CHECK9-NEXT: br i1 [[CMP_I]], label [[FOR_BODY_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
// CHECK9: for.body.i:
// CHECK9-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: [[IDXPROM_I:%.*]] = sext i32 [[TMP15]] to i64
// CHECK9-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i64 [[IDXPROM_I]]
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX_I]], align 4
// CHECK9-NEXT: [[MUL_I:%.*]] = mul nsw i32 2, [[TMP16]]
// CHECK9-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP9]], align 8, !nonnull [[META16:![0-9]+]], !align [[META17:![0-9]+]]
// CHECK9-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: [[IDXPROM1_I:%.*]] = sext i32 [[TMP19]] to i64
// CHECK9-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i32, ptr [[TMP18]], i64 [[IDXPROM1_I]]
// CHECK9-NEXT: store i32 [[MUL_I]], ptr [[ARRAYIDX2_I]], align 4
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: [[INC_I:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK9-NEXT: store i32 [[INC_I]], ptr [[I_I]], align 4, !noalias [[META15]]
// CHECK9-NEXT: br label [[FOR_COND_I]], !llvm.loop [[LOOP18:![0-9]+]]
// CHECK9: .omp_outlined..exit:
// CHECK9-NEXT: ret i32 0
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l44
// CHECK11-SAME: (ptr noalias noundef [[DYN_PTR:%.*]], ptr noundef [[B:%.*]], ptr noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DYN_PTR_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 4
// CHECK11-NEXT: [[DOTAFFS_ARR_ADDR:%.*]] = alloca [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T:%.*]]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1:[0-9]+]])
// CHECK11-NEXT: store ptr [[DYN_PTR]], ptr [[DYN_PTR_ADDR]], align 4
// CHECK11-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[B_ADDR]], ptr [[TMP1]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 24, i32 4, ptr @.omp_task_entry.)
// CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x [[STRUCT_KMP_TASK_AFFINITY_INFO_T]]], ptr [[DOTAFFS_ARR_ADDR]], i32 0, i32 0
// CHECK11-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP4]], i32 0
// CHECK11-NEXT: [[TMP5:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP5]], i32 1023
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[ARRAYIDX1]], i32 1
// CHECK11-NEXT: [[TMP7:%.*]] = ptrtoaddr ptr [[TMP6]] to i32
// CHECK11-NEXT: [[TMP8:%.*]] = ptrtoaddr ptr [[ARRAYIDX]] to i32
// CHECK11-NEXT: [[TMP9:%.*]] = sub nuw i32 [[TMP7]], [[TMP8]]
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP3]], i32 0
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i32
// CHECK11-NEXT: store i32 [[TMP12]], ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_AFFINITY_INFO_T]], ptr [[TMP10]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP9]], ptr [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_omp_reg_task_with_affinity(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]], i32 1, ptr [[TMP3]])
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP15]], i32 0, i32 0
// CHECK11-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 4
// CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP17]], ptr align 4 [[AGG_CAPTURED]], i32 4, i1 false)
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP2]], i32 0, i32 1
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP18]], i32 0, i32 0
// CHECK11-NEXT: [[TMP20:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP0]], ptr [[TMP2]])
// CHECK11-NEXT: [[TMP22:%.*]] = call i32 @__kmpc_omp_taskwait(ptr @[[GLOB1]], i32 [[TMP0]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK11-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK11-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP2]], i32 0, i32 0
// CHECK11-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK11-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: [[I_I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK11-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK11-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK11-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
// CHECK11-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META12:![0-9]+]])
// CHECK11-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META14:![0-9]+]])
// CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META16:![0-9]+]]
// CHECK11-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: store i32 0, ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: br label [[FOR_COND_I:%.*]]
// CHECK11: for.cond.i:
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[CMP_I:%.*]] = icmp slt i32 [[TMP13]], 1024
// CHECK11-NEXT: br i1 [[CMP_I]], label [[FOR_BODY_I:%.*]], label [[DOTOMP_OUTLINED__EXIT:%.*]]
// CHECK11: for.body.i:
// CHECK11-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i32 [[TMP15]]
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX_I]], align 4
// CHECK11-NEXT: [[MUL_I:%.*]] = mul nsw i32 2, [[TMP16]]
// CHECK11-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP9]], align 4, !nonnull [[META17:![0-9]+]], !align [[META18:![0-9]+]]
// CHECK11-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[ARRAYIDX1_I:%.*]] = getelementptr inbounds i32, ptr [[TMP18]], i32 [[TMP19]]
// CHECK11-NEXT: store i32 [[MUL_I]], ptr [[ARRAYIDX1_I]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: [[INC_I:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK11-NEXT: store i32 [[INC_I]], ptr [[I_I]], align 4, !noalias [[META16]]
// CHECK11-NEXT: br label [[FOR_COND_I]], !llvm.loop [[LOOP19:![0-9]+]]
// CHECK11: .omp_outlined..exit:
// CHECK11-NEXT: ret i32 0
//
Reference in New Issue View Git Blame Copy Permalink