llvm/clang/test/OpenMP/teams_distribute_parallel_for_private_codegen.cpp

// 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 _
// RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK1
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK1
// RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK3
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK3

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

// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++  -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap  %s --check-prefix=CHECK9

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

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

struct St {
  int a, b;
  St() : a(0), b(0) {}
  St(const St &st) : a(st.a + st.b), b(0) {}
  ~St() {}
};

volatile int g = 1212;
volatile int &g1 = g;

template <class T>
struct S {
  T f;
  S(T a) : f(a + g) {}
  S() : f(g) {}
  S(const S &s, St t = St()) : f(s.f + t.a) {}
  operator T() { return T(); }
  ~S() {}
};


template <typename T>
T tmain() {
  S<T> test;
  T t_var = T();
  T vec[] = {1, 2};
  S<T> s_arr[] = {1, 2};
  S<T> &var = test;
#pragma omp target
#pragma omp teams distribute parallel for private(t_var, vec, s_arr, var)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
  return T();
}

S<float> test;
int t_var = 333;
int vec[] = {1, 2};
S<float> s_arr[] = {1, 2};
S<float> var(3);

int main() {
  static int sivar;
#ifdef LAMBDA
  [&]() {
#pragma omp target
#pragma omp teams distribute parallel for private(g, g1, sivar)
  for (int i = 0; i < 2; ++i) {

    // Skip global, bound tid and loop vars

    g = 1;
    g1 = 1;
    sivar = 2;

    // Skip global, bound tid and loop vars
    [&]() {
      g = 2;
      g1 = 2;
      sivar = 4;

    }();
  }
  }();
  return 0;
#else
#pragma omp target
#pragma omp teams distribute parallel for private(t_var, vec, s_arr, var, sivar)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
    sivar += i;
  }
  return tmain<int>();
#endif
}



// Skip global, bound tid and loop vars

// private(s_arr)

// private(var)


// Skip global, bound tid and loop vars

// private(s_arr)

// private(var)




// Skip global, bound tid and loop vars

// private(s_arr)


// private(var)


// Skip global, bound tid and loop vars
// prev lb and ub
// iter variables

// private(s_arr)


// private(var)



#endif
// CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) @test)
// CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @test, ptr @__dso_handle) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 4
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK1-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @s_arr, float noundef 1.000000e+00)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i64 1), float noundef 2.000000e+00)
// CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR2]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ef(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], @s_arr
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done1:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 4
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
// CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
// CHECK1-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @var, float noundef 3.000000e+00)
// CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @var, ptr @__dso_handle) #[[ATTR2]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT:    store i32 0, ptr [[TMP1]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT:    store ptr null, ptr [[TMP3]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT:    store ptr null, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT:    store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP7]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT:    store i64 2, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP10]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP11]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP12]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = 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_l96.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT:    [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK1-NEXT:    br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1:       omp_offload.failed:
// CHECK1-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96() #[[ATTR2]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK1:       omp_offload.cont:
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT:    ret i32 [[CALL]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96
// CHECK1-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined)
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[SIVAR:%.*]] = 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 i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 1, 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br 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 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
// 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:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP14]])
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAY_BEGIN2:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN2]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN2]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE3:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR4]] {
// 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:    [[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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[SIVAR:%.*]] = 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 i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV1:%.*]] = trunc i64 [[TMP1]] 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// 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:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP10]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM3:%.*]] = sext i32 [[TMP12]] to i64
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i64 0, i64 [[IDXPROM3]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX4]], ptr align 4 [[VAR]], i64 4, i1 false)
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK1-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP13]]
// CHECK1-NEXT:    store i32 [[ADD5]], ptr [[SIVAR]], 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:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT:    store i32 [[ADD6]], 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:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP17]])
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN7]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP18]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[S_ARR]], i32 noundef signext 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[S_ARR]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 noundef signext 2)
// CHECK1-NEXT:    store ptr [[TEST]], ptr [[VAR]], align 8
// CHECK1-NEXT:    store ptr undef, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT:    store i32 0, ptr [[TMP1]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT:    store ptr null, ptr [[TMP3]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT:    store ptr null, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT:    store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP7]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT:    store i64 2, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP10]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP11]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP12]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT:    [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK1-NEXT:    br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1:       omp_offload.failed:
// CHECK1-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56() #[[ATTR2]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK1:       omp_offload.cont:
// CHECK1-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP16]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], i32 noundef signext [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
// CHECK1-SAME: () #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined)
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// 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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// 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 undef, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 1, 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[_TMP2]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br 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 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
// 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:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP14]])
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAY_BEGIN4:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN4]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN4]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE5:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done5:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR4]] {
// 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:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// 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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[_TMP3:%.*]] = alloca ptr, align 8
// 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 undef, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[CONV2]], 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[_TMP3]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// 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:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP10]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[_TMP3]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM5:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i64 0, i64 [[IDXPROM5]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX6]], ptr align 4 [[TMP12]], i64 4, i1 false)
// 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:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], 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:    [[TMP15:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP16]])
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN8]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN8]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE9:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done9:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 4
// CHECK1-NEXT:    store i32 [[TMP0]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_private_codegen.cpp
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    call void @__cxx_global_var_init()
// CHECK1-NEXT:    call void @__cxx_global_var_init.1()
// CHECK1-NEXT:    call void @__cxx_global_var_init.2()
// CHECK1-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__cxx_global_var_init
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) @test)
// CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @test, ptr @__dso_handle) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIfEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIfED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 4
// CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK3-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @s_arr, float noundef 1.000000e+00)
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i32 1), float noundef 2.000000e+00)
// CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR2]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIfEC2Ef(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i32 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], @s_arr
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done1:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 4
// CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
// CHECK3-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
// CHECK3-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @var, float noundef 3.000000e+00)
// CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @var, ptr @__dso_handle) #[[ATTR2]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT:    store i32 0, ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT:    store ptr null, ptr [[TMP3]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT:    store ptr null, ptr [[TMP4]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT:    store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT:    store i64 2, ptr [[TMP8]], align 8
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP9]], align 8
// CHECK3-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP10]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP12]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = 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_l96.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT:    [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK3-NEXT:    br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3:       omp_offload.failed:
// CHECK3-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96() #[[ATTR2]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK3:       omp_offload.cont:
// CHECK3-NEXT:    [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT:    ret i32 [[CALL]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96
// CHECK3-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined)
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK3-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK3-NEXT:    [[SIVAR:%.*]] = 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 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 1, 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR]], i32 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK3:       omp.inner.for.cond.cleanup:
// CHECK3-NEXT:    br 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 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
// 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:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP12]])
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAY_BEGIN2:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN2]], i32 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN2]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE3:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done3:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l96.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR4]] {
// 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:    [[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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK3-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK3-NEXT:    [[SIVAR:%.*]] = 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 i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR]], i32 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK3-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// 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:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK3-NEXT:    br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK3:       omp.inner.for.cond.cleanup:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT:    store i32 [[TMP10]], ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 [[TMP12]]
// CHECK3-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[ARRAYIDX2]], ptr align 4 [[VAR]], i32 4, i1 false)
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[TMP14:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK3-NEXT:    [[ADD3:%.*]] = add nsw i32 [[TMP14]], [[TMP13]]
// CHECK3-NEXT:    store i32 [[ADD3]], ptr [[SIVAR]], 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:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK3-NEXT:    store i32 [[ADD4]], 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:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP17]])
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAY_BEGIN5:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN5]], i32 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP18]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN5]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE6:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done6:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR1]] comdat {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK3-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK3-NEXT:    [[VAR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK3-NEXT:    store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[S_ARR]], i32 noundef 1)
// CHECK3-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[S_ARR]], i32 1
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 noundef 2)
// CHECK3-NEXT:    store ptr [[TEST]], ptr [[VAR]], align 4
// CHECK3-NEXT:    store ptr undef, ptr [[_TMP1]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT:    store i32 0, ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT:    store ptr null, ptr [[TMP3]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT:    store ptr null, ptr [[TMP4]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT:    store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT:    store i64 2, ptr [[TMP8]], align 8
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP9]], align 8
// CHECK3-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP10]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP12]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT:    [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK3-NEXT:    br i1 [[TMP14]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3:       omp_offload.failed:
// CHECK3-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56() #[[ATTR2]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK3:       omp_offload.cont:
// CHECK3-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done2:
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK3-NEXT:    ret i32 [[TMP16]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIiEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIiEC2Ei(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], i32 noundef [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
// CHECK3-SAME: () #[[ATTR4]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined)
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca ptr, 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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK3-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK3-NEXT:    [[_TMP2:%.*]] = alloca ptr, 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 undef, ptr [[_TMP1]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 1, 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYCTOR_CUR]], i32 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK3-NEXT:    store ptr [[VAR]], ptr [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK3-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK3:       omp.inner.for.cond.cleanup:
// CHECK3-NEXT:    br 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 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]])
// 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:    br label [[OMP_INNER_FOR_COND]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3:       omp.loop.exit:
// CHECK3-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP12]])
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAY_BEGIN4:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN4]], i32 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN4]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE5:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done5:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR4]] {
// 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:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca ptr, 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:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK3-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK3-NEXT:    [[_TMP2:%.*]] = alloca ptr, 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 undef, ptr [[_TMP1]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP0]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], 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:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYCTOR_CUR]], i32 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK3-NEXT:    store ptr [[VAR]], ptr [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// 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:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK3-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK3:       omp.inner.for.cond.cleanup:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT:    store i32 [[TMP10]], ptr [[ARRAYIDX]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 [[TMP13]]
// CHECK3-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[ARRAYIDX4]], ptr align 4 [[TMP12]], i32 4, i1 false)
// 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:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK3-NEXT:    store i32 [[ADD5]], 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:    [[TMP15:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP16]])
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAY_BEGIN6:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN6]], i32 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN6]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE7:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done7:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIiED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 4
// CHECK3-NEXT:    store i32 [[TMP0]], ptr [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 4
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_private_codegen.cpp
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    call void @__cxx_global_var_init()
// CHECK3-NEXT:    call void @__cxx_global_var_init.1()
// CHECK3-NEXT:    call void @__cxx_global_var_init.2()
// CHECK3-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@__cxx_global_var_init
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) @test)
// CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @test, ptr @__dso_handle) #[[ATTR2:[0-9]+]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 4
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK9-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
// CHECK9-SAME: () #[[ATTR0]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @s_arr, float noundef 1.000000e+00)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i64 1), float noundef 2.000000e+00)
// CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @__cxx_global_array_dtor, ptr null, ptr @__dso_handle) #[[ATTR2]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfEC2Ef(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
// CHECK9-SAME: (ptr noundef [[TMP0:%.*]]) #[[ATTR0]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ getelementptr inbounds ([[STRUCT_S:%.*]], ptr @s_arr, i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], @s_arr
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done1:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 4
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
// CHECK9-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
// CHECK9-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
// CHECK9-SAME: () #[[ATTR0]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) @var, float noundef 3.000000e+00)
// CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(ptr @_ZN1SIfED1Ev, ptr @var, ptr @__dso_handle) #[[ATTR2]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK9-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK9-NEXT:    ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75
// CHECK9-SAME: (i64 noundef [[G1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store i64 [[G1]], ptr [[G1_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[G1_ADDR]], ptr [[TMP]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3:[0-9]+]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75.omp_outlined)
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// 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:    [[G:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[G1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[SIVAR:%.*]] = 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 undef, ptr [[_TMP1]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 1, 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:    store ptr [[G1]], ptr [[_TMP2]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// 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 [[TMP1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR4]] {
// 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:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// 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:    [[G:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[G1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP3:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[SIVAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], 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 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 undef, ptr [[_TMP1]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV2]], 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:    store ptr [[G1]], ptr [[_TMP3]], align 8
// 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 @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// 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_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_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:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK9-NEXT:    store i32 1, ptr [[G]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[_TMP3]], align 8
// CHECK9-NEXT:    store volatile i32 1, ptr [[TMP10]], align 4
// CHECK9-NEXT:    store i32 2, ptr [[SIVAR]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[G]], ptr [[TMP11]], align 8
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 1
// CHECK9-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[_TMP3]], align 8
// CHECK9-NEXT:    store ptr [[TMP13]], ptr [[TMP12]], align 8
// CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[SIVAR]], ptr [[TMP14]], align 8
// CHECK9-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(24) [[REF_TMP]])
// 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:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK9-NEXT:    store i32 [[ADD5]], 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 [[TMP3]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_private_codegen.cpp
// CHECK9-SAME: () #[[ATTR0]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    call void @__cxx_global_var_init()
// CHECK9-NEXT:    call void @__cxx_global_var_init.1()
// CHECK9-NEXT:    call void @__cxx_global_var_init.2()
// CHECK9-NEXT:    ret void
//