// 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 _
// expected-no-diagnostics
#ifndef HEADER
#define HEADER

// Test host codegen.
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -DCK1 -fopenmp -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 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3

// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// 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 -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// 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 -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
#ifdef CK1

template <typename T, int X, long long Y>
struct SS{
  T a[X];
  float b;
  int foo(void) {

    #pragma omp target
    #pragma omp teams distribute parallel for
    for(int i = 0; i < X; i++) {
      a[i] = (T)0;
    }
    #pragma omp target
    #pragma omp teams distribute parallel for dist_schedule(static)
    for(int i = 0; i < X; i++) {
      a[i] = (T)0;
    }
    #pragma omp target
    #pragma omp teams distribute parallel for dist_schedule(static, X/2)
    for(int i = 0; i < X; i++) {
      a[i] = (T)0;
    }


    return a[0];
  }
};

int teams_template_struct(void) {
  SS<int, 123, 456> V;
  return V.foo();

}
#endif // CK1

// Test host codegen.
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11

// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK2 -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 -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK2 -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 -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
#ifdef CK2

template <typename T, int n>
int tmain(T argc) {
  T a[n];
  int m = 10;
#pragma omp target
#pragma omp teams distribute parallel for
  for(int i = 0; i < n; i++) {
    a[i] = (T)0;
  }
#pragma omp target
#pragma omp teams distribute parallel for dist_schedule(static)
  for(int i = 0; i < n; i++) {
    a[i] = (T)0;
  }
#pragma omp target
#pragma omp teams distribute parallel for dist_schedule(static, m)
  for(int i = 0; i < n; i++) {
    a[i] = (T)0;
  }
  return 0;
}

int main (int argc, char **argv) {
  int n = 100;
  int a[n];
  int m = 10;
#pragma omp target
#pragma omp teams distribute parallel for
  for(int i = 0; i < n; i++) {
    a[i] = 0;
  }
#pragma omp target
#pragma omp teams distribute parallel for dist_schedule(static)
  for(int i = 0; i < n; i++) {
    a[i] = 0;
  }
#pragma omp target
#pragma omp teams distribute parallel for dist_schedule(static, m)
  for(int i = 0; i < n; i++) {
    a[i] = 0;
  }
  return tmain<int, 10>(argc);
}


#endif // CK2
#endif // #ifndef HEADER
// CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
// CHECK1-NEXT:    ret i32 [[CALL]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[_TMP14:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT:    store i64 123, ptr [[TMP13]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK1-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1:       omp_offload.failed:
// CHECK1-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK1:       omp_offload.cont:
// CHECK1-NEXT:    [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[THIS1]], ptr [[TMP20]], align 8
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A2]], ptr [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
// CHECK1-NEXT:    store ptr null, ptr [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP25]], align 4
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
// CHECK1-NEXT:    store i32 1, ptr [[TMP26]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
// CHECK1-NEXT:    store ptr [[TMP23]], ptr [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
// CHECK1-NEXT:    store ptr [[TMP24]], ptr [[TMP28]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
// CHECK1-NEXT:    store ptr @.offload_sizes.1, ptr [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
// CHECK1-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP31]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP32]], align 8
// CHECK1-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
// CHECK1-NEXT:    store i64 123, ptr [[TMP33]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP34]], align 8
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
// CHECK1-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
// CHECK1-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP37]], align 4
// CHECK1-NEXT:    [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.region_id, ptr [[KERNEL_ARGS7]])
// CHECK1-NEXT:    [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
// CHECK1-NEXT:    br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
// CHECK1:       omp_offload.failed8:
// CHECK1-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33(ptr [[THIS1]]) #[[ATTR2]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT9]]
// CHECK1:       omp_offload.cont9:
// CHECK1-NEXT:    [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[THIS1]], ptr [[TMP40]], align 8
// CHECK1-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A10]], ptr [[TMP41]], align 8
// CHECK1-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i64 0, i64 0
// CHECK1-NEXT:    store ptr null, ptr [[TMP42]], align 8
// CHECK1-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP45]], align 4
// CHECK1-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
// CHECK1-NEXT:    store i32 1, ptr [[TMP46]], align 4
// CHECK1-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
// CHECK1-NEXT:    store ptr [[TMP43]], ptr [[TMP47]], align 8
// CHECK1-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
// CHECK1-NEXT:    store ptr [[TMP44]], ptr [[TMP48]], align 8
// CHECK1-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
// CHECK1-NEXT:    store ptr @.offload_sizes.3, ptr [[TMP49]], align 8
// CHECK1-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
// CHECK1-NEXT:    store ptr @.offload_maptypes.4, ptr [[TMP50]], align 8
// CHECK1-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP51]], align 8
// CHECK1-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP52]], align 8
// CHECK1-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
// CHECK1-NEXT:    store i64 123, ptr [[TMP53]], align 8
// CHECK1-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP54]], align 8
// CHECK1-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
// CHECK1-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
// CHECK1-NEXT:    [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP57]], align 4
// CHECK1-NEXT:    [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.region_id, ptr [[KERNEL_ARGS15]])
// CHECK1-NEXT:    [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
// CHECK1-NEXT:    br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK1:       omp_offload.failed16:
// CHECK1-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38(ptr [[THIS1]]) #[[ATTR2]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT17]]
// CHECK1:       omp_offload.cont17:
// CHECK1-NEXT:    [[A18:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    ret i32 [[TMP60]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined, ptr [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined, ptr [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 123
// CHECK1-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]], ptr [[TMP0]])
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK1-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP17]], 122
// CHECK1-NEXT:    br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK1:       cond.true5:
// CHECK1-NEXT:    br label [[COND_END7:%.*]]
// CHECK1:       cond.false6:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END7]]
// CHECK1:       cond.end7:
// CHECK1-NEXT:    [[COND8:%.*]] = phi i32 [ 122, [[COND_TRUE5]] ], [ [[TMP18]], [[COND_FALSE6]] ]
// CHECK1-NEXT:    store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP19]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK3-NEXT:    [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]])
// CHECK3-NEXT:    ret i32 [[CALL]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT:    [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[_TMP6:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK3-NEXT:    [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT:    [[_TMP14:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT:    store i64 123, ptr [[TMP13]], align 8
// CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK3-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK3-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3:       omp_offload.failed:
// CHECK3-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK3:       omp_offload.cont:
// CHECK3-NEXT:    [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[THIS1]], ptr [[TMP20]], align 4
// CHECK3-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[A2]], ptr [[TMP21]], align 4
// CHECK3-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0
// CHECK3-NEXT:    store ptr null, ptr [[TMP22]], align 4
// CHECK3-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP25]], align 4
// CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1
// CHECK3-NEXT:    store i32 1, ptr [[TMP26]], align 4
// CHECK3-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2
// CHECK3-NEXT:    store ptr [[TMP23]], ptr [[TMP27]], align 4
// CHECK3-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3
// CHECK3-NEXT:    store ptr [[TMP24]], ptr [[TMP28]], align 4
// CHECK3-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4
// CHECK3-NEXT:    store ptr @.offload_sizes.1, ptr [[TMP29]], align 4
// CHECK3-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5
// CHECK3-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP30]], align 4
// CHECK3-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP31]], align 4
// CHECK3-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP32]], align 4
// CHECK3-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8
// CHECK3-NEXT:    store i64 123, ptr [[TMP33]], align 8
// CHECK3-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP34]], align 8
// CHECK3-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
// CHECK3-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
// CHECK3-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP37]], align 4
// CHECK3-NEXT:    [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.region_id, ptr [[KERNEL_ARGS7]])
// CHECK3-NEXT:    [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
// CHECK3-NEXT:    br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]]
// CHECK3:       omp_offload.failed8:
// CHECK3-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33(ptr [[THIS1]]) #[[ATTR2]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT9]]
// CHECK3:       omp_offload.cont9:
// CHECK3-NEXT:    [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[THIS1]], ptr [[TMP40]], align 4
// CHECK3-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[A10]], ptr [[TMP41]], align 4
// CHECK3-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i32 0, i32 0
// CHECK3-NEXT:    store ptr null, ptr [[TMP42]], align 4
// CHECK3-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP45]], align 4
// CHECK3-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
// CHECK3-NEXT:    store i32 1, ptr [[TMP46]], align 4
// CHECK3-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
// CHECK3-NEXT:    store ptr [[TMP43]], ptr [[TMP47]], align 4
// CHECK3-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
// CHECK3-NEXT:    store ptr [[TMP44]], ptr [[TMP48]], align 4
// CHECK3-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
// CHECK3-NEXT:    store ptr @.offload_sizes.3, ptr [[TMP49]], align 4
// CHECK3-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
// CHECK3-NEXT:    store ptr @.offload_maptypes.4, ptr [[TMP50]], align 4
// CHECK3-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP51]], align 4
// CHECK3-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP52]], align 4
// CHECK3-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
// CHECK3-NEXT:    store i64 123, ptr [[TMP53]], align 8
// CHECK3-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP54]], align 8
// CHECK3-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4
// CHECK3-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4
// CHECK3-NEXT:    [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP57]], align 4
// CHECK3-NEXT:    [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.region_id, ptr [[KERNEL_ARGS15]])
// CHECK3-NEXT:    [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0
// CHECK3-NEXT:    br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK3:       omp_offload.failed16:
// CHECK3-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38(ptr [[THIS1]]) #[[ATTR2]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT17]]
// CHECK3:       omp_offload.cont17:
// CHECK3-NEXT:    [[A18:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    ret i32 [[TMP60]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3:       omp.body.continue:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK3-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined, ptr [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3:       omp.body.continue:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK3-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined, ptr [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61)
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 123
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]], ptr [[TMP0]])
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK3-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP13]], [[TMP14]]
// CHECK3-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP15]], 122
// CHECK3-NEXT:    br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK3:       cond.true5:
// CHECK3-NEXT:    br label [[COND_END7:%.*]]
// CHECK3:       cond.false6:
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END7]]
// CHECK3:       cond.end7:
// CHECK3-NEXT:    [[COND8:%.*]] = phi i32 [ 122, [[COND_TRUE5]] ], [ [[TMP16]], [[COND_FALSE6]] ]
// CHECK3-NEXT:    store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 122, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122
// CHECK3-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3:       cond.true:
// CHECK3-NEXT:    br label [[COND_END:%.*]]
// CHECK3:       cond.false:
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3:       omp.body.continue:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK3-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[ARGV_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT:    [[N_CASTED3:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 8
// CHECK9-NEXT:    [[_TMP8:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK9-NEXT:    [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_CASTED18:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 8
// CHECK9-NEXT:    [[_TMP23:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT:    store i32 100, ptr [[N]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT:    [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK9-NEXT:    store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT:    store i32 10, ptr [[M]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false)
// CHECK9-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP4]], ptr [[TMP6]], align 8
// CHECK9-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP4]], ptr [[TMP7]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP8]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP9]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP10]], align 8
// CHECK9-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT:    store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP12]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP13]], align 8
// CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP5]], ptr [[TMP14]], align 8
// CHECK9-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT:    store ptr null, ptr [[TMP15]], align 8
// CHECK9-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK9-NEXT:    [[TMP22:%.*]] = zext i32 [[ADD]] to i64
// CHECK9-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP23]], align 4
// CHECK9-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT:    store i32 3, ptr [[TMP24]], align 4
// CHECK9-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP16]], ptr [[TMP25]], align 8
// CHECK9-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP17]], ptr [[TMP26]], align 8
// CHECK9-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[TMP18]], ptr [[TMP27]], align 8
// CHECK9-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes, ptr [[TMP28]], align 8
// CHECK9-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP29]], align 8
// CHECK9-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP30]], align 8
// CHECK9-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT:    store i64 [[TMP22]], ptr [[TMP31]], align 8
// CHECK9-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP32]], align 8
// CHECK9-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
// CHECK9-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
// CHECK9-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP35]], align 4
// CHECK9-NEXT:    [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT:    [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
// CHECK9-NEXT:    br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9:       omp_offload.failed:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK9:       omp_offload.cont:
// CHECK9-NEXT:    [[TMP38:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
// CHECK9-NEXT:    [[TMP39:%.*]] = load i64, ptr [[N_CASTED3]], align 8
// CHECK9-NEXT:    [[TMP40:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES7]], ptr align 8 @.offload_sizes.1, i64 24, i1 false)
// CHECK9-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP39]], ptr [[TMP41]], align 8
// CHECK9-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP39]], ptr [[TMP42]], align 8
// CHECK9-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP44]], align 8
// CHECK9-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP45]], align 8
// CHECK9-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 1
// CHECK9-NEXT:    store ptr null, ptr [[TMP46]], align 8
// CHECK9-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP47]], align 8
// CHECK9-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP48]], align 8
// CHECK9-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP40]], ptr [[TMP49]], align 8
// CHECK9-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 2
// CHECK9-NEXT:    store ptr null, ptr [[TMP50]], align 8
// CHECK9-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP54:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT:    [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
// CHECK9-NEXT:    [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0
// CHECK9-NEXT:    [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK9-NEXT:    [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK9-NEXT:    store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT:    [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
// CHECK9-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1
// CHECK9-NEXT:    [[TMP57:%.*]] = zext i32 [[ADD14]] to i64
// CHECK9-NEXT:    [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP58]], align 4
// CHECK9-NEXT:    [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
// CHECK9-NEXT:    store i32 3, ptr [[TMP59]], align 4
// CHECK9-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP51]], ptr [[TMP60]], align 8
// CHECK9-NEXT:    [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP52]], ptr [[TMP61]], align 8
// CHECK9-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[TMP53]], ptr [[TMP62]], align 8
// CHECK9-NEXT:    [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8
// CHECK9-NEXT:    [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP64]], align 8
// CHECK9-NEXT:    [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP65]], align 8
// CHECK9-NEXT:    [[TMP66:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
// CHECK9-NEXT:    store i64 [[TMP57]], ptr [[TMP66]], align 8
// CHECK9-NEXT:    [[TMP67:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP67]], align 8
// CHECK9-NEXT:    [[TMP68:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4
// CHECK9-NEXT:    [[TMP69:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
// CHECK9-NEXT:    [[TMP70:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP70]], align 4
// CHECK9-NEXT:    [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, ptr [[KERNEL_ARGS15]])
// CHECK9-NEXT:    [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0
// CHECK9-NEXT:    br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK9:       omp_offload.failed16:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i64 [[TMP39]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT17]]
// CHECK9:       omp_offload.cont17:
// CHECK9-NEXT:    [[TMP73:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT:    store i32 [[TMP73]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT:    [[TMP74:%.*]] = load i64, ptr [[M_CASTED]], align 8
// CHECK9-NEXT:    [[TMP75:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP75]], ptr [[N_CASTED18]], align 4
// CHECK9-NEXT:    [[TMP76:%.*]] = load i64, ptr [[N_CASTED18]], align 8
// CHECK9-NEXT:    [[TMP77:%.*]] = mul nuw i64 [[TMP1]], 4
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES22]], ptr align 8 @.offload_sizes.3, i64 32, i1 false)
// CHECK9-NEXT:    [[TMP78:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP74]], ptr [[TMP78]], align 8
// CHECK9-NEXT:    [[TMP79:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP74]], ptr [[TMP79]], align 8
// CHECK9-NEXT:    [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP80]], align 8
// CHECK9-NEXT:    [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP76]], ptr [[TMP81]], align 8
// CHECK9-NEXT:    [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP76]], ptr [[TMP82]], align 8
// CHECK9-NEXT:    [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1
// CHECK9-NEXT:    store ptr null, ptr [[TMP83]], align 8
// CHECK9-NEXT:    [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP84]], align 8
// CHECK9-NEXT:    [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP85]], align 8
// CHECK9-NEXT:    [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2
// CHECK9-NEXT:    store ptr null, ptr [[TMP86]], align 8
// CHECK9-NEXT:    [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP87]], align 8
// CHECK9-NEXT:    [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP88]], align 8
// CHECK9-NEXT:    [[TMP89:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
// CHECK9-NEXT:    store i64 [[TMP77]], ptr [[TMP89]], align 8
// CHECK9-NEXT:    [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 3
// CHECK9-NEXT:    store ptr null, ptr [[TMP90]], align 8
// CHECK9-NEXT:    [[TMP91:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP93:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP94:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP94]], ptr [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT:    [[TMP95:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
// CHECK9-NEXT:    [[SUB26:%.*]] = sub nsw i32 [[TMP95]], 0
// CHECK9-NEXT:    [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK9-NEXT:    [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK9-NEXT:    store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT:    [[TMP96:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
// CHECK9-NEXT:    [[ADD29:%.*]] = add nsw i32 [[TMP96]], 1
// CHECK9-NEXT:    [[TMP97:%.*]] = zext i32 [[ADD29]] to i64
// CHECK9-NEXT:    [[TMP98:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP98]], align 4
// CHECK9-NEXT:    [[TMP99:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
// CHECK9-NEXT:    store i32 4, ptr [[TMP99]], align 4
// CHECK9-NEXT:    [[TMP100:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP91]], ptr [[TMP100]], align 8
// CHECK9-NEXT:    [[TMP101:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP92]], ptr [[TMP101]], align 8
// CHECK9-NEXT:    [[TMP102:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[TMP93]], ptr [[TMP102]], align 8
// CHECK9-NEXT:    [[TMP103:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.4, ptr [[TMP103]], align 8
// CHECK9-NEXT:    [[TMP104:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP104]], align 8
// CHECK9-NEXT:    [[TMP105:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP105]], align 8
// CHECK9-NEXT:    [[TMP106:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
// CHECK9-NEXT:    store i64 [[TMP97]], ptr [[TMP106]], align 8
// CHECK9-NEXT:    [[TMP107:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP107]], align 8
// CHECK9-NEXT:    [[TMP108:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP108]], align 4
// CHECK9-NEXT:    [[TMP109:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP109]], align 4
// CHECK9-NEXT:    [[TMP110:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP110]], align 4
// CHECK9-NEXT:    [[TMP111:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, ptr [[KERNEL_ARGS30]])
// CHECK9-NEXT:    [[TMP112:%.*]] = icmp ne i32 [[TMP111]], 0
// CHECK9-NEXT:    br i1 [[TMP112]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK9:       omp_offload.failed31:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i64 [[TMP74]], i64 [[TMP76]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT32]]
// CHECK9:       omp_offload.cont32:
// CHECK9-NEXT:    [[TMP113:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP113]])
// CHECK9-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT:    [[TMP114:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP114]])
// CHECK9-NEXT:    [[TMP115:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP115]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
// CHECK9-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
// CHECK9-NEXT:    [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT:    br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK9-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I4]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK9-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111
// CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
// CHECK9-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = zext i32 [[TMP18]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
// CHECK9-NEXT:    [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT:    br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK9-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I4]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK9-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116
// CHECK9-SAME: (i64 noundef [[M:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK9-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
// CHECK9-NEXT:    br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK9-NEXT:    [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK9-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP21]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK9-NEXT:    [[TMP22:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined.omp_outlined, i64 [[TMP18]], i64 [[TMP20]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]], i64 [[TMP22]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK9-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
// CHECK9-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
// CHECK9-NEXT:    store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[CMP10:%.*]] = icmp sgt i32 [[TMP29]], [[TMP30]]
// CHECK9-NEXT:    br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
// CHECK9:       cond.true11:
// CHECK9-NEXT:    [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    br label [[COND_END13:%.*]]
// CHECK9:       cond.false12:
// CHECK9-NEXT:    [[TMP32:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END13]]
// CHECK9:       cond.end13:
// CHECK9-NEXT:    [[COND14:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE11]] ], [ [[TMP32]], [[COND_FALSE12]] ]
// CHECK9-NEXT:    store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP33]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP35]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I5:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK9-NEXT:    store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP7]] to i32
// CHECK9-NEXT:    [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV4:%.*]] = trunc i64 [[TMP8]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[CMP6:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT:    br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK9-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I5]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I5]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK9-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[A:%.*]] = alloca [10 x i32], align 4
// CHECK9-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK9-NEXT:    [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 8
// CHECK9-NEXT:    [[_TMP11:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK9-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    store i32 10, ptr [[M]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr @.offload_sizes.5, ptr [[TMP9]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.6, ptr [[TMP10]], align 8
// CHECK9-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT:    store i64 10, ptr [[TMP13]], align 8
// CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK9-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK9-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK9-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK9-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9:       omp_offload.failed:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84(ptr [[A]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK9:       omp_offload.cont:
// CHECK9-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP20]], align 8
// CHECK9-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP21]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP22]], align 8
// CHECK9-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP25]], align 4
// CHECK9-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
// CHECK9-NEXT:    store i32 1, ptr [[TMP26]], align 4
// CHECK9-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP23]], ptr [[TMP27]], align 8
// CHECK9-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP24]], ptr [[TMP28]], align 8
// CHECK9-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
// CHECK9-NEXT:    store ptr @.offload_sizes.7, ptr [[TMP29]], align 8
// CHECK9-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.8, ptr [[TMP30]], align 8
// CHECK9-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP31]], align 8
// CHECK9-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP32]], align 8
// CHECK9-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
// CHECK9-NEXT:    store i64 10, ptr [[TMP33]], align 8
// CHECK9-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP34]], align 8
// CHECK9-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
// CHECK9-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
// CHECK9-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP37]], align 4
// CHECK9-NEXT:    [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS5]])
// CHECK9-NEXT:    [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
// CHECK9-NEXT:    br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK9:       omp_offload.failed6:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT7]]
// CHECK9:       omp_offload.cont7:
// CHECK9-NEXT:    [[TMP40:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT:    store i32 [[TMP40]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT:    [[TMP41:%.*]] = load i64, ptr [[M_CASTED]], align 8
// CHECK9-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP41]], ptr [[TMP42]], align 8
// CHECK9-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP41]], ptr [[TMP43]], align 8
// CHECK9-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP44]], align 8
// CHECK9-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP45]], align 8
// CHECK9-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP46]], align 8
// CHECK9-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1
// CHECK9-NEXT:    store ptr null, ptr [[TMP47]], align 8
// CHECK9-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP50]], align 4
// CHECK9-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
// CHECK9-NEXT:    store i32 2, ptr [[TMP51]], align 4
// CHECK9-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP48]], ptr [[TMP52]], align 8
// CHECK9-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP49]], ptr [[TMP53]], align 8
// CHECK9-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
// CHECK9-NEXT:    store ptr @.offload_sizes.9, ptr [[TMP54]], align 8
// CHECK9-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.10, ptr [[TMP55]], align 8
// CHECK9-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP56]], align 8
// CHECK9-NEXT:    [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP57]], align 8
// CHECK9-NEXT:    [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
// CHECK9-NEXT:    store i64 10, ptr [[TMP58]], align 8
// CHECK9-NEXT:    [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP59]], align 8
// CHECK9-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
// CHECK9-NEXT:    [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
// CHECK9-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP62]], align 4
// CHECK9-NEXT:    [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, ptr [[KERNEL_ARGS12]])
// CHECK9-NEXT:    [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
// CHECK9-NEXT:    br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
// CHECK9:       omp_offload.failed13:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i64 [[TMP41]], ptr [[A]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT14]]
// CHECK9:       omp_offload.cont14:
// CHECK9-NEXT:    ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined, ptr [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94
// CHECK9-SAME: (i64 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined, ptr [[TMP0]], i64 [[TMP3]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP3]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP1]])
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK9-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK9-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK9-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK9-NEXT:    br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK9:       cond.true5:
// CHECK9-NEXT:    br label [[COND_END7:%.*]]
// CHECK9:       cond.false6:
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END7]]
// CHECK9:       cond.end7:
// CHECK9-NEXT:    [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ]
// CHECK9-NEXT:    store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK9-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK9:       omp.inner.for.end:
// CHECK9-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9:       omp.loop.exit:
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK9-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@main
// CHECK11-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[ARGV_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT:    [[N_CASTED3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4
// CHECK11-NEXT:    [[_TMP8:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK11-NEXT:    [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_CASTED18:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [4 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS20:%.*]] = alloca [4 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [4 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_SIZES22:%.*]] = alloca [4 x i64], align 4
// CHECK11-NEXT:    [[_TMP23:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK11-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
// CHECK11-NEXT:    store i32 100, ptr [[N]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK11-NEXT:    store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT:    [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
// CHECK11-NEXT:    store i32 10, ptr [[M]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT:    [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK11-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false)
// CHECK11-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[TMP6]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP8]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP9]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP10]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT:    store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP12]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP13]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2
// CHECK11-NEXT:    store i64 [[TMP5]], ptr [[TMP14]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr null, ptr [[TMP15]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK11-NEXT:    [[TMP22:%.*]] = zext i32 [[ADD]] to i64
// CHECK11-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP23]], align 4
// CHECK11-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 3, ptr [[TMP24]], align 4
// CHECK11-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP16]], ptr [[TMP25]], align 4
// CHECK11-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP17]], ptr [[TMP26]], align 4
// CHECK11-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[TMP18]], ptr [[TMP27]], align 4
// CHECK11-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes, ptr [[TMP28]], align 4
// CHECK11-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP29]], align 4
// CHECK11-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP30]], align 4
// CHECK11-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT:    store i64 [[TMP22]], ptr [[TMP31]], align 8
// CHECK11-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP32]], align 8
// CHECK11-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4
// CHECK11-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4
// CHECK11-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP35]], align 4
// CHECK11-NEXT:    [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT:    [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0
// CHECK11-NEXT:    br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11:       omp_offload.failed:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK11:       omp_offload.cont:
// CHECK11-NEXT:    [[TMP38:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP38]], ptr [[N_CASTED3]], align 4
// CHECK11-NEXT:    [[TMP39:%.*]] = load i32, ptr [[N_CASTED3]], align 4
// CHECK11-NEXT:    [[TMP40:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT:    [[TMP41:%.*]] = sext i32 [[TMP40]] to i64
// CHECK11-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES7]], ptr align 4 @.offload_sizes.1, i32 24, i1 false)
// CHECK11-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP39]], ptr [[TMP42]], align 4
// CHECK11-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP39]], ptr [[TMP43]], align 4
// CHECK11-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP44]], align 4
// CHECK11-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP45]], align 4
// CHECK11-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP46]], align 4
// CHECK11-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1
// CHECK11-NEXT:    store ptr null, ptr [[TMP47]], align 4
// CHECK11-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP48]], align 4
// CHECK11-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP49]], align 4
// CHECK11-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2
// CHECK11-NEXT:    store i64 [[TMP41]], ptr [[TMP50]], align 4
// CHECK11-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2
// CHECK11-NEXT:    store ptr null, ptr [[TMP51]], align 4
// CHECK11-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP55:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP55]], ptr [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT:    [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4
// CHECK11-NEXT:    [[SUB11:%.*]] = sub nsw i32 [[TMP56]], 0
// CHECK11-NEXT:    [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1
// CHECK11-NEXT:    [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1
// CHECK11-NEXT:    store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4
// CHECK11-NEXT:    [[TMP57:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4
// CHECK11-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP57]], 1
// CHECK11-NEXT:    [[TMP58:%.*]] = zext i32 [[ADD14]] to i64
// CHECK11-NEXT:    [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP59]], align 4
// CHECK11-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1
// CHECK11-NEXT:    store i32 3, ptr [[TMP60]], align 4
// CHECK11-NEXT:    [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP52]], ptr [[TMP61]], align 4
// CHECK11-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP53]], ptr [[TMP62]], align 4
// CHECK11-NEXT:    [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[TMP54]], ptr [[TMP63]], align 4
// CHECK11-NEXT:    [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP64]], align 4
// CHECK11-NEXT:    [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP65]], align 4
// CHECK11-NEXT:    [[TMP66:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP66]], align 4
// CHECK11-NEXT:    [[TMP67:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8
// CHECK11-NEXT:    store i64 [[TMP58]], ptr [[TMP67]], align 8
// CHECK11-NEXT:    [[TMP68:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP68]], align 8
// CHECK11-NEXT:    [[TMP69:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4
// CHECK11-NEXT:    [[TMP70:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP70]], align 4
// CHECK11-NEXT:    [[TMP71:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP71]], align 4
// CHECK11-NEXT:    [[TMP72:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.region_id, ptr [[KERNEL_ARGS15]])
// CHECK11-NEXT:    [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0
// CHECK11-NEXT:    br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]]
// CHECK11:       omp_offload.failed16:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111(i32 [[TMP39]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT17]]
// CHECK11:       omp_offload.cont17:
// CHECK11-NEXT:    [[TMP74:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT:    store i32 [[TMP74]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP75:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP76:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP76]], ptr [[N_CASTED18]], align 4
// CHECK11-NEXT:    [[TMP77:%.*]] = load i32, ptr [[N_CASTED18]], align 4
// CHECK11-NEXT:    [[TMP78:%.*]] = mul nuw i32 [[TMP0]], 4
// CHECK11-NEXT:    [[TMP79:%.*]] = sext i32 [[TMP78]] to i64
// CHECK11-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES22]], ptr align 4 @.offload_sizes.3, i32 32, i1 false)
// CHECK11-NEXT:    [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP75]], ptr [[TMP80]], align 4
// CHECK11-NEXT:    [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP75]], ptr [[TMP81]], align 4
// CHECK11-NEXT:    [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP82]], align 4
// CHECK11-NEXT:    [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP77]], ptr [[TMP83]], align 4
// CHECK11-NEXT:    [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP77]], ptr [[TMP84]], align 4
// CHECK11-NEXT:    [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1
// CHECK11-NEXT:    store ptr null, ptr [[TMP85]], align 4
// CHECK11-NEXT:    [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP86]], align 4
// CHECK11-NEXT:    [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP87]], align 4
// CHECK11-NEXT:    [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2
// CHECK11-NEXT:    store ptr null, ptr [[TMP88]], align 4
// CHECK11-NEXT:    [[TMP89:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP89]], align 4
// CHECK11-NEXT:    [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP90]], align 4
// CHECK11-NEXT:    [[TMP91:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 3
// CHECK11-NEXT:    store i64 [[TMP79]], ptr [[TMP91]], align 4
// CHECK11-NEXT:    [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 3
// CHECK11-NEXT:    store ptr null, ptr [[TMP92]], align 4
// CHECK11-NEXT:    [[TMP93:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP94:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP95:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP96:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP96]], ptr [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT:    [[TMP97:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4
// CHECK11-NEXT:    [[SUB26:%.*]] = sub nsw i32 [[TMP97]], 0
// CHECK11-NEXT:    [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1
// CHECK11-NEXT:    [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1
// CHECK11-NEXT:    store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT:    [[TMP98:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4
// CHECK11-NEXT:    [[ADD29:%.*]] = add nsw i32 [[TMP98]], 1
// CHECK11-NEXT:    [[TMP99:%.*]] = zext i32 [[ADD29]] to i64
// CHECK11-NEXT:    [[TMP100:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP100]], align 4
// CHECK11-NEXT:    [[TMP101:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1
// CHECK11-NEXT:    store i32 4, ptr [[TMP101]], align 4
// CHECK11-NEXT:    [[TMP102:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP93]], ptr [[TMP102]], align 4
// CHECK11-NEXT:    [[TMP103:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP94]], ptr [[TMP103]], align 4
// CHECK11-NEXT:    [[TMP104:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[TMP95]], ptr [[TMP104]], align 4
// CHECK11-NEXT:    [[TMP105:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.4, ptr [[TMP105]], align 4
// CHECK11-NEXT:    [[TMP106:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP106]], align 4
// CHECK11-NEXT:    [[TMP107:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP107]], align 4
// CHECK11-NEXT:    [[TMP108:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8
// CHECK11-NEXT:    store i64 [[TMP99]], ptr [[TMP108]], align 8
// CHECK11-NEXT:    [[TMP109:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP109]], align 8
// CHECK11-NEXT:    [[TMP110:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP110]], align 4
// CHECK11-NEXT:    [[TMP111:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP111]], align 4
// CHECK11-NEXT:    [[TMP112:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP112]], align 4
// CHECK11-NEXT:    [[TMP113:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.region_id, ptr [[KERNEL_ARGS30]])
// CHECK11-NEXT:    [[TMP114:%.*]] = icmp ne i32 [[TMP113]], 0
// CHECK11-NEXT:    br i1 [[TMP114]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]]
// CHECK11:       omp_offload.failed31:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116(i32 [[TMP75]], i32 [[TMP77]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT32]]
// CHECK11:       omp_offload.cont32:
// CHECK11-NEXT:    [[TMP115:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP115]])
// CHECK11-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT:    [[TMP116:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP116]])
// CHECK11-NEXT:    [[TMP117:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT:    ret i32 [[TMP117]]
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106
// CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
// CHECK11-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l106.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK11-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I3]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK11-NEXT:    store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111
// CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]]
// CHECK11-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l111.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK11-NEXT:    br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I3]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I3]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK11-NEXT:    store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116
// CHECK11-SAME: (i32 noundef [[M:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store i32 [[M]], ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]])
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK11-NEXT:    [[CMP6:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK11-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined.omp_outlined, i32 [[TMP17]], i32 [[TMP18]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]], i32 [[TMP20]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP21]], [[TMP22]]
// CHECK11-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP23:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP24:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP24]]
// CHECK11-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP25]], [[TMP26]]
// CHECK11-NEXT:    store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP27:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP28:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[CMP10:%.*]] = icmp sgt i32 [[TMP27]], [[TMP28]]
// CHECK11-NEXT:    br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]]
// CHECK11:       cond.true11:
// CHECK11-NEXT:    [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    br label [[COND_END13:%.*]]
// CHECK11:       cond.false12:
// CHECK11-NEXT:    [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END13]]
// CHECK11:       cond.end13:
// CHECK11-NEXT:    [[COND14:%.*]] = phi i32 [ [[TMP29]], [[COND_TRUE11]] ], [ [[TMP30]], [[COND_FALSE12]] ]
// CHECK11-NEXT:    store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP31]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP33]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l116.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1
// CHECK11-NEXT:    store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP5]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
// CHECK11-NEXT:    br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK11-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I4]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I4]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK11-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_
// CHECK11-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A:%.*]] = alloca [10 x i32], align 4
// CHECK11-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[_TMP4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK11-NEXT:    [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 4
// CHECK11-NEXT:    [[_TMP11:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK11-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    store i32 10, ptr [[M]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr @.offload_sizes.5, ptr [[TMP9]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.6, ptr [[TMP10]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT:    store i64 10, ptr [[TMP13]], align 8
// CHECK11-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK11-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11:       omp_offload.failed:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84(ptr [[A]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK11:       omp_offload.cont:
// CHECK11-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP20]], align 4
// CHECK11-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP21]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP25]], align 4
// CHECK11-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
// CHECK11-NEXT:    store i32 1, ptr [[TMP26]], align 4
// CHECK11-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP23]], ptr [[TMP27]], align 4
// CHECK11-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP24]], ptr [[TMP28]], align 4
// CHECK11-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
// CHECK11-NEXT:    store ptr @.offload_sizes.7, ptr [[TMP29]], align 4
// CHECK11-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.8, ptr [[TMP30]], align 4
// CHECK11-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP31]], align 4
// CHECK11-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP32]], align 4
// CHECK11-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
// CHECK11-NEXT:    store i64 10, ptr [[TMP33]], align 8
// CHECK11-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP34]], align 8
// CHECK11-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4
// CHECK11-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4
// CHECK11-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP37]], align 4
// CHECK11-NEXT:    [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS5]])
// CHECK11-NEXT:    [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0
// CHECK11-NEXT:    br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK11:       omp_offload.failed6:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT7]]
// CHECK11:       omp_offload.cont7:
// CHECK11-NEXT:    [[TMP40:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT:    store i32 [[TMP40]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP41:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP42:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP41]], ptr [[TMP42]], align 4
// CHECK11-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP41]], ptr [[TMP43]], align 4
// CHECK11-NEXT:    [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP44]], align 4
// CHECK11-NEXT:    [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP45]], align 4
// CHECK11-NEXT:    [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP46]], align 4
// CHECK11-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 1
// CHECK11-NEXT:    store ptr null, ptr [[TMP47]], align 4
// CHECK11-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP50]], align 4
// CHECK11-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1
// CHECK11-NEXT:    store i32 2, ptr [[TMP51]], align 4
// CHECK11-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP48]], ptr [[TMP52]], align 4
// CHECK11-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP49]], ptr [[TMP53]], align 4
// CHECK11-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4
// CHECK11-NEXT:    store ptr @.offload_sizes.9, ptr [[TMP54]], align 4
// CHECK11-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.10, ptr [[TMP55]], align 4
// CHECK11-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP56]], align 4
// CHECK11-NEXT:    [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP57]], align 4
// CHECK11-NEXT:    [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8
// CHECK11-NEXT:    store i64 10, ptr [[TMP58]], align 8
// CHECK11-NEXT:    [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP59]], align 8
// CHECK11-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP60]], align 4
// CHECK11-NEXT:    [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4
// CHECK11-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP62]], align 4
// CHECK11-NEXT:    [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.region_id, ptr [[KERNEL_ARGS12]])
// CHECK11-NEXT:    [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0
// CHECK11-NEXT:    br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]]
// CHECK11:       omp_offload.failed13:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94(i32 [[TMP41]], ptr [[A]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT14]]
// CHECK11:       omp_offload.cont14:
// CHECK11-NEXT:    ret i32 0
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84
// CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined, ptr [[TMP0]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK11-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89
// CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK11-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94
// CHECK11-SAME: (i32 noundef [[M:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store i32 [[M]], ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined, ptr [[TMP0]], i32 [[TMP3]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP3]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP1]])
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK11-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK11-NEXT:    store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9
// CHECK11-NEXT:    br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]]
// CHECK11:       cond.true5:
// CHECK11-NEXT:    br label [[COND_END7:%.*]]
// CHECK11:       cond.false6:
// CHECK11-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END7]]
// CHECK11:       cond.end7:
// CHECK11-NEXT:    [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ]
// CHECK11-NEXT:    store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l94.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK11-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK11-NEXT:    store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK11-NEXT:    ret void
//