clang/test/OpenMP/teams_distribute_parallel_for_firstprivate_codegen.cpp - llvm-project - Git at Google

 // 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 - -Wno-openmp-mapping | FileCheck %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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | FileCheck %s --check-prefix=CHECK2
 // RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | FileCheck %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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | FileCheck %s --check-prefix=CHECK4

 // RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - -Wno-openmp-mapping | 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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | 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 - -Wno-openmp-mapping | 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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | 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 - -Wno-openmp-mapping | FileCheck %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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | FileCheck %s --check-prefix=CHECK10

 // 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 - -Wno-openmp-mapping | 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 -Wno-openmp-mapping
 // 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 - -Wno-openmp-mapping | 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 firstprivate(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 firstprivate(g, g1, sivar)
   for (int i = 0; i < 2; ++i) {

     // Skip global and bound tid vars
     // skip loop vars
     g = 1;
     g1 = 1;
     sivar = 2;

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

     // use of private vars
     [&]() {
       g = 2;
       g1 = 2;
       sivar = 4;

     }();
   }
   }();
   return 0;
 #else
 #pragma omp target
 #pragma omp teams distribute parallel for firstprivate(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 and bound tid vars
 // Skip temp vars for loop

 // param copy

 // T_VAR and SIVAR

 // preparation vars

 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2

 // firstprivate(s_arr)

 // firstprivate(var)


 // Skip global and bound tid vars, and prev lb ub vars
 // Skip temp vars for loop

 // param copy

 // T_VAR and SIVAR

 // preparation vars

 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2

 // firstprivate(s_arr)

 // firstprivate(var)


 // Skip global and bound tid vars
 // Skip temp vars for loop

 // param copy

 // T_VAR and preparation variables

 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2

 // firstprivate(s_arr)

 // firstprivate(var)


 // Skip global and bound tid vars
 // Skip temp vars for loop

 // param copy

 // T_VAR and preparation variables

 // firstprivate vec(vec): copy from *_addr into priv1 and then from priv1 into priv2

 // firstprivate(s_arr)

 // firstprivate(var)


 #endif
 // CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK1-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
 // CHECK1-SAME: () #[[ATTR0]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 0), float 1.000000e+00)
 // CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 1), float 2.000000e+00)
 // CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK1-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK1-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
 // CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // 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: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK1-NEXT:    store float [[ADD]], float* [[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(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__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:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
 // CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* @t_var, align 4
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[CONV]], align 4
 // CHECK1-NEXT:    [[TMP1:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
 // CHECK1-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP2]], i32* [[CONV1]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP1]], i64* [[TMP5]], align 8
 // CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP1]], i64* [[TMP7]], align 8
 // CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
 // CHECK1-NEXT:    store i8* null, i8** [[TMP8]], align 8
 // CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK1-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [2 x i32]**
 // CHECK1-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP10]], align 8
 // CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK1-NEXT:    [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [2 x i32]**
 // CHECK1-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP12]], align 8
 // CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
 // CHECK1-NEXT:    store i8* null, i8** [[TMP13]], align 8
 // CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK1-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x %struct.S]**
 // CHECK1-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP15]], align 8
 // CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [2 x %struct.S]**
 // CHECK1-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP17]], align 8
 // CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
 // CHECK1-NEXT:    store i8* null, i8** [[TMP18]], align 8
 // CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.S**
 // CHECK1-NEXT:    store %struct.S* @var, %struct.S** [[TMP20]], align 8
 // CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK1-NEXT:    [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.S**
 // CHECK1-NEXT:    store %struct.S* @var, %struct.S** [[TMP22]], align 8
 // CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
 // CHECK1-NEXT:    store i8* null, i8** [[TMP23]], align 8
 // CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
 // CHECK1-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP3]], i64* [[TMP25]], align 8
 // CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
 // CHECK1-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP3]], i64* [[TMP27]], align 8
 // CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
 // CHECK1-NEXT:    store i8* null, i8** [[TMP28]], align 8
 // CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK1-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 2)
 // CHECK1-NEXT:    [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99.region_id, i32 5, i8** [[TMP29]], i8** [[TMP30]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK1-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
 // CHECK1-NEXT:    br i1 [[TMP32]], 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_l99(i64 [[TMP1]], [2 x i32]* @vec, [2 x %struct.S]* @s_arr, %struct.S* @var, i64 [[TMP3]]) #[[ATTR2]]
 // CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK1:       omp_offload.cont:
 // CHECK1-NEXT:    [[CALL:%.*]] = call signext i32 @_Z5tmainIiET_v()
 // CHECK1-NEXT:    ret i32 [[CALL]]
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99
 // CHECK1-SAME: (i64 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4:[0-9]+]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[CONV2:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP3]], i32* [[CONV2]], align 4
 // CHECK1-NEXT:    [[TMP4:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK1-NEXT:    [[CONV3:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP5]], i32* [[CONV3]], align 4
 // CHECK1-NEXT:    [[TMP6:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i64, [2 x %struct.S]*, %struct.S*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i64 [[TMP4]], [2 x %struct.S]* [[TMP1]], %struct.S* [[TMP2]], i64 [[TMP6]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, 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:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK1-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK1-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK1-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK1-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i64 8, i1 false)
 // CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK1:       omp.arraycpy.body:
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK1:       omp.arraycpy.done4:
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP6]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 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:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    br label [[COND_END]]
 // CHECK1:       cond.end:
 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK1:       omp.inner.for.cond:
 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
 // CHECK1-NEXT:    br i1 [[CMP7]], 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:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[CONV8:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP18]], i32* [[CONV8]], align 4
 // CHECK1-NEXT:    [[TMP19:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK1-NEXT:    [[CONV9:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP20]], i32* [[CONV9]], align 4
 // CHECK1-NEXT:    [[TMP21:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [2 x i32]*, i64, [2 x %struct.S]*, %struct.S*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP15]], i64 [[TMP17]], [2 x i32]* [[VEC2]], i64 [[TMP19]], [2 x %struct.S]* [[S_ARR3]], %struct.S* [[VAR5]], i64 [[TMP21]])
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK1:       omp.inner.for.inc:
 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[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:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]])
 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i64 2
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP26]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK1:       arraydestroy.done11:
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN2StC1Ev
 // CHECK1-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN2StC2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN2StD1Ev
 // CHECK1-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR2]]
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[SIVAR_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:    [[VEC4:%.*]] = alloca [2 x i32], align 4
 // CHECK1-NEXT:    [[S_ARR5:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK1-NEXT:    [[VAR7:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK1-NEXT:    [[AGG_TMP8:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK1-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
 // CHECK1-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK1-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP4]] to i32
 // CHECK1-NEXT:    store i32 [[CONV2]], i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
 // CHECK1-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i64 8, i1 false)
 // CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK1:       omp.arraycpy.body:
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK1:       omp.arraycpy.done6:
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]])
 // CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR7]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP8]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 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:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    br label [[COND_END]]
 // CHECK1:       cond.end:
 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK1:       omp.inner.for.cond:
 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
 // CHECK1-NEXT:    br i1 [[CMP9]], 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:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I]], align 4
 // CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64
 // CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 [[IDXPROM]]
 // CHECK1-NEXT:    store i32 [[TMP17]], i32* [[ARRAYIDX]], align 4
 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK1-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP19]] to i64
 // CHECK1-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i64 0, i64 [[IDXPROM10]]
 // CHECK1-NEXT:    [[TMP20:%.*]] = bitcast %struct.S* [[ARRAYIDX11]] to i8*
 // CHECK1-NEXT:    [[TMP21:%.*]] = bitcast %struct.S* [[VAR7]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP20]], i8* align 4 [[TMP21]], i64 4, i1 false)
 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
 // CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP23]], [[TMP22]]
 // CHECK1-NEXT:    store i32 [[ADD12]], i32* [[CONV1]], align 8
 // 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:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK1-NEXT:    store i32 [[ADD13]], i32* [[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:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR7]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAY_BEGIN14:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN14]], i64 2
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN14]]
 // CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE15:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK1:       arraydestroy.done15:
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
 // CHECK1-SAME: () #[[ATTR6:[0-9]+]] 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 %struct.S.0*, align 8
 // CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
 // CHECK1-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
 // CHECK1-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
 // CHECK1-NEXT:    store i32 0, i32* [[T_VAR]], align 4
 // CHECK1-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
 // CHECK1-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
 // CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 signext 1)
 // CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
 // CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 signext 2)
 // CHECK1-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
 // CHECK1-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR]], align 4
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP2]], i32* [[CONV]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP3]], i64* [[TMP8]], align 8
 // CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
 // CHECK1-NEXT:    store i64 [[TMP3]], i64* [[TMP10]], align 8
 // CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
 // CHECK1-NEXT:    store i8* null, i8** [[TMP11]], align 8
 // CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK1-NEXT:    [[TMP13:%.*]] = bitcast i8** [[TMP12]] to [2 x i32]**
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP13]], align 8
 // CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK1-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x i32]**
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP15]], align 8
 // CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
 // CHECK1-NEXT:    store i8* null, i8** [[TMP16]], align 8
 // CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK1-NEXT:    [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [2 x %struct.S.0]**
 // CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP18]], align 8
 // CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [2 x %struct.S.0]**
 // CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP20]], align 8
 // CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
 // CHECK1-NEXT:    store i8* null, i8** [[TMP21]], align 8
 // CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK1-NEXT:    [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.S.0**
 // CHECK1-NEXT:    store %struct.S.0* [[TMP5]], %struct.S.0** [[TMP23]], align 8
 // CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK1-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.S.0**
 // CHECK1-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[TMP25]], align 8
 // CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
 // CHECK1-NEXT:    store i8* null, i8** [[TMP26]], align 8
 // CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK1-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 2)
 // CHECK1-NEXT:    [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.6, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK1-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
 // CHECK1-NEXT:    br i1 [[TMP30]], 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(i64 [[TMP3]], [2 x i32]* [[VEC]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[TMP4]]) #[[ATTR2]]
 // CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK1:       omp_offload.cont:
 // CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP31]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
 // CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK1:       arraydestroy.done2:
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[TMP32:%.*]] = load i32, i32* [[RETVAL]], align 4
 // CHECK1-NEXT:    ret i32 [[TMP32]]
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN2StC2Ev
 // CHECK1-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
 // CHECK1-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 1
 // CHECK1-NEXT:    store i32 0, i32* [[B]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
 // CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP0]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP1:%.*]] = load float, float* [[F2]], align 4
 // CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
 // CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
 // CHECK1-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN2StD2Ev
 // CHECK1-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 signext [[TMP0]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
 // CHECK1-SAME: (i64 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[CONV1:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP3]], i32* [[CONV1]], align 4
 // CHECK1-NEXT:    [[TMP4:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i64, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i64 [[TMP4]], [2 x %struct.S.0]* [[TMP1]], %struct.S.0* [[TMP5]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    [[_TMP1:%.*]] = 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:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK1-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK1-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK1-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK1-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK1-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i64 8, i1 false)
 // CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK1:       omp.arraycpy.body:
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK1:       omp.arraycpy.done4:
 // CHECK1-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP7]], %struct.St* [[AGG_TMP6]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK1-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 8
 // CHECK1-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 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:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    br label [[COND_END]]
 // CHECK1:       cond.end:
 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK1:       omp.inner.for.cond:
 // CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
 // CHECK1-NEXT:    br i1 [[CMP8]], 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:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK1-NEXT:    [[TMP16:%.*]] = zext i32 [[TMP15]] to i64
 // CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK1-NEXT:    [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[CONV9:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK1-NEXT:    store i32 [[TMP19]], i32* [[CONV9]], align 4
 // CHECK1-NEXT:    [[TMP20:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK1-NEXT:    [[TMP21:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 8
 // CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [2 x i32]*, i64, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP16]], i64 [[TMP18]], [2 x i32]* [[VEC2]], i64 [[TMP20]], [2 x %struct.S.0]* [[S_ARR3]], %struct.S.0* [[TMP21]])
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK1:       omp.inner.for.inc:
 // CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[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:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]])
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN10]], i64 2
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP26]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK1:       arraydestroy.done11:
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    [[_TMP1:%.*]] = 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:    [[VEC4:%.*]] = alloca [2 x i32], align 4
 // CHECK1-NEXT:    [[S_ARR5:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK1-NEXT:    [[VAR7:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK1-NEXT:    [[AGG_TMP8:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK1-NEXT:    [[_TMP9:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK1-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK1-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
 // CHECK1-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK1-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP4]] to i32
 // CHECK1-NEXT:    store i32 [[CONV2]], i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
 // CHECK1-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i64 8, i1 false)
 // CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK1:       omp.arraycpy.body:
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK1:       omp.arraycpy.done6:
 // CHECK1-NEXT:    [[TMP9:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]])
 // CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR7]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP9]], %struct.St* [[AGG_TMP8]])
 // CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]]) #[[ATTR2]]
 // CHECK1-NEXT:    store %struct.S.0* [[VAR7]], %struct.S.0** [[_TMP9]], align 8
 // CHECK1-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 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:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    br label [[COND_END]]
 // CHECK1:       cond.end:
 // CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
 // CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK1-NEXT:    store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK1:       omp.inner.for.cond:
 // CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK1-NEXT:    [[CMP10:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
 // CHECK1-NEXT:    br i1 [[CMP10]], 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:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
 // CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 [[IDXPROM]]
 // CHECK1-NEXT:    store i32 [[TMP18]], i32* [[ARRAYIDX]], align 4
 // CHECK1-NEXT:    [[TMP20:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP9]], align 8
 // CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
 // CHECK1-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP21]] to i64
 // CHECK1-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i64 0, i64 [[IDXPROM11]]
 // CHECK1-NEXT:    [[TMP22:%.*]] = bitcast %struct.S.0* [[ARRAYIDX12]] to i8*
 // CHECK1-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP20]] to i8*
 // CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP22]], i8* align 4 [[TMP23]], 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:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK1-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK1-NEXT:    store i32 [[ADD13]], i32* [[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:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR7]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAY_BEGIN14:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN14]], i64 2
 // CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK1:       arraydestroy.body:
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN14]]
 // CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE15:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK1:       arraydestroy.done15:
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK1-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
 // CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP0]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[F2]], align 4
 // CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
 // CHECK1-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
 // CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_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
 //
 //
 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK1-SAME: () #[[ATTR0]] {
 // CHECK1-NEXT:  entry:
 // CHECK1-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK1-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK2-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK2-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
 // CHECK2-SAME: () #[[ATTR0]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 0), float 1.000000e+00)
 // CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 1), float 2.000000e+00)
 // CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK2-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK2-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
 // CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done1:
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK2-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK2-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK2-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
 // CHECK2-SAME: () #[[ATTR0]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__dso_handle) #[[ATTR2]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@main
 // CHECK2-SAME: () #[[ATTR3:[0-9]+]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 8
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* @t_var, align 4
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[CONV]], align 4
 // CHECK2-NEXT:    [[TMP1:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
 // CHECK2-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP2]], i32* [[CONV1]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP1]], i64* [[TMP5]], align 8
 // CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP1]], i64* [[TMP7]], align 8
 // CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
 // CHECK2-NEXT:    store i8* null, i8** [[TMP8]], align 8
 // CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK2-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [2 x i32]**
 // CHECK2-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP10]], align 8
 // CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK2-NEXT:    [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [2 x i32]**
 // CHECK2-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP12]], align 8
 // CHECK2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
 // CHECK2-NEXT:    store i8* null, i8** [[TMP13]], align 8
 // CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK2-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x %struct.S]**
 // CHECK2-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP15]], align 8
 // CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [2 x %struct.S]**
 // CHECK2-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP17]], align 8
 // CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
 // CHECK2-NEXT:    store i8* null, i8** [[TMP18]], align 8
 // CHECK2-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.S**
 // CHECK2-NEXT:    store %struct.S* @var, %struct.S** [[TMP20]], align 8
 // CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK2-NEXT:    [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.S**
 // CHECK2-NEXT:    store %struct.S* @var, %struct.S** [[TMP22]], align 8
 // CHECK2-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
 // CHECK2-NEXT:    store i8* null, i8** [[TMP23]], align 8
 // CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
 // CHECK2-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP3]], i64* [[TMP25]], align 8
 // CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
 // CHECK2-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP3]], i64* [[TMP27]], align 8
 // CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
 // CHECK2-NEXT:    store i8* null, i8** [[TMP28]], align 8
 // CHECK2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK2-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 2)
 // CHECK2-NEXT:    [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99.region_id, i32 5, i8** [[TMP29]], i8** [[TMP30]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK2-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
 // CHECK2-NEXT:    br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 // CHECK2:       omp_offload.failed:
 // CHECK2-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99(i64 [[TMP1]], [2 x i32]* @vec, [2 x %struct.S]* @s_arr, %struct.S* @var, i64 [[TMP3]]) #[[ATTR2]]
 // CHECK2-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK2:       omp_offload.cont:
 // CHECK2-NEXT:    [[CALL:%.*]] = call signext i32 @_Z5tmainIiET_v()
 // CHECK2-NEXT:    ret i32 [[CALL]]
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99
 // CHECK2-SAME: (i64 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4:[0-9]+]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[CONV2:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP3]], i32* [[CONV2]], align 4
 // CHECK2-NEXT:    [[TMP4:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK2-NEXT:    [[CONV3:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP5]], i32* [[CONV3]], align 4
 // CHECK2-NEXT:    [[TMP6:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i64, [2 x %struct.S]*, %struct.S*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i64 [[TMP4]], [2 x %struct.S]* [[TMP1]], %struct.S* [[TMP2]], i64 [[TMP6]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK2-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK2-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK2-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK2-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK2-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i64 8, i1 false)
 // CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK2:       omp.arraycpy.body:
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK2:       omp.arraycpy.done4:
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK2-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP6]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 1
 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK2:       cond.true:
 // CHECK2-NEXT:    br label [[COND_END:%.*]]
 // CHECK2:       cond.false:
 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    br label [[COND_END]]
 // CHECK2:       cond.end:
 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK2:       omp.inner.for.cond:
 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
 // CHECK2-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK2:       omp.inner.for.cond.cleanup:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK2:       omp.inner.for.body:
 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    [[TMP15:%.*]] = zext i32 [[TMP14]] to i64
 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[TMP17:%.*]] = zext i32 [[TMP16]] to i64
 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[CONV8:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP18]], i32* [[CONV8]], align 4
 // CHECK2-NEXT:    [[TMP19:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK2-NEXT:    [[CONV9:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP20]], i32* [[CONV9]], align 4
 // CHECK2-NEXT:    [[TMP21:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [2 x i32]*, i64, [2 x %struct.S]*, %struct.S*, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP15]], i64 [[TMP17]], [2 x i32]* [[VEC2]], i64 [[TMP19]], [2 x %struct.S]* [[S_ARR3]], %struct.S* [[VAR5]], i64 [[TMP21]])
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK2:       omp.inner.for.inc:
 // CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK2:       omp.inner.for.end:
 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK2:       omp.loop.exit:
 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]])
 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i64 2
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP26]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done11:
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN2StC1Ev
 // CHECK2-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK2-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN2StC2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN2StD1Ev
 // CHECK2-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK2-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR2]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[VEC4:%.*]] = alloca [2 x i32], align 4
 // CHECK2-NEXT:    [[S_ARR5:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK2-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK2-NEXT:    [[VAR7:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK2-NEXT:    [[AGG_TMP8:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK2-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
 // CHECK2-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK2-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP4]] to i32
 // CHECK2-NEXT:    store i32 [[CONV2]], i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK2-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
 // CHECK2-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i64 8, i1 false)
 // CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK2:       omp.arraycpy.body:
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK2:       omp.arraycpy.done6:
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]])
 // CHECK2-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR7]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP8]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK2:       cond.true:
 // CHECK2-NEXT:    br label [[COND_END:%.*]]
 // CHECK2:       cond.false:
 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    br label [[COND_END]]
 // CHECK2:       cond.end:
 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK2:       omp.inner.for.cond:
 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
 // CHECK2-NEXT:    br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK2:       omp.inner.for.cond.cleanup:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK2:       omp.inner.for.body:
 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I]], align 4
 // CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64
 // CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 [[IDXPROM]]
 // CHECK2-NEXT:    store i32 [[TMP17]], i32* [[ARRAYIDX]], align 4
 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK2-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP19]] to i64
 // CHECK2-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i64 0, i64 [[IDXPROM10]]
 // CHECK2-NEXT:    [[TMP20:%.*]] = bitcast %struct.S* [[ARRAYIDX11]] to i8*
 // CHECK2-NEXT:    [[TMP21:%.*]] = bitcast %struct.S* [[VAR7]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP20]], i8* align 4 [[TMP21]], i64 4, i1 false)
 // CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
 // CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK2-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP23]], [[TMP22]]
 // CHECK2-NEXT:    store i32 [[ADD12]], i32* [[CONV1]], align 8
 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 // CHECK2:       omp.body.continue:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK2:       omp.inner.for.inc:
 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK2-NEXT:    store i32 [[ADD13]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK2:       omp.inner.for.end:
 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK2:       omp.loop.exit:
 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR7]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAY_BEGIN14:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR5]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN14]], i64 2
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN14]]
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE15:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done15:
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
 // CHECK2-SAME: () #[[ATTR6:[0-9]+]] comdat {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK2-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
 // CHECK2-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK2-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8
 // CHECK2-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8
 // CHECK2-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
 // CHECK2-NEXT:    store i32 0, i32* [[T_VAR]], align 4
 // CHECK2-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
 // CHECK2-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
 // CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 signext 1)
 // CHECK2-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
 // CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 signext 2)
 // CHECK2-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
 // CHECK2-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR]], align 4
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP2]], i32* [[CONV]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP3]], i64* [[TMP8]], align 8
 // CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
 // CHECK2-NEXT:    store i64 [[TMP3]], i64* [[TMP10]], align 8
 // CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
 // CHECK2-NEXT:    store i8* null, i8** [[TMP11]], align 8
 // CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK2-NEXT:    [[TMP13:%.*]] = bitcast i8** [[TMP12]] to [2 x i32]**
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP13]], align 8
 // CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK2-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x i32]**
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP15]], align 8
 // CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
 // CHECK2-NEXT:    store i8* null, i8** [[TMP16]], align 8
 // CHECK2-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK2-NEXT:    [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [2 x %struct.S.0]**
 // CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP18]], align 8
 // CHECK2-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [2 x %struct.S.0]**
 // CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP20]], align 8
 // CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
 // CHECK2-NEXT:    store i8* null, i8** [[TMP21]], align 8
 // CHECK2-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK2-NEXT:    [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.S.0**
 // CHECK2-NEXT:    store %struct.S.0* [[TMP5]], %struct.S.0** [[TMP23]], align 8
 // CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK2-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.S.0**
 // CHECK2-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[TMP25]], align 8
 // CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
 // CHECK2-NEXT:    store i8* null, i8** [[TMP26]], align 8
 // CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK2-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 2)
 // CHECK2-NEXT:    [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.6, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK2-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
 // CHECK2-NEXT:    br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 // CHECK2:       omp_offload.failed:
 // CHECK2-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56(i64 [[TMP3]], [2 x i32]* [[VEC]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[TMP4]]) #[[ATTR2]]
 // CHECK2-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK2:       omp_offload.cont:
 // CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP31]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done2:
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[TMP32:%.*]] = load i32, i32* [[RETVAL]], align 4
 // CHECK2-NEXT:    ret i32 [[TMP32]]
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN2StC2Ev
 // CHECK2-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK2-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    store i32 0, i32* [[A]], align 4
 // CHECK2-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 1
 // CHECK2-NEXT:    store i32 0, i32* [[B]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
 // CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP0]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP1:%.*]] = load float, float* [[F2]], align 4
 // CHECK2-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK2-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
 // CHECK2-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
 // CHECK2-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN2StD2Ev
 // CHECK2-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
 // CHECK2-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK2-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 signext [[TMP0]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
 // CHECK2-SAME: (i64 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[CONV1:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP3]], i32* [[CONV1]], align 4
 // CHECK2-NEXT:    [[TMP4:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i64, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i64 [[TMP4]], [2 x %struct.S.0]* [[TMP1]], %struct.S.0* [[TMP5]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK2-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK2-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK2-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK2-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK2-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK2-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i64 8, i1 false)
 // CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK2:       omp.arraycpy.body:
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK2:       omp.arraycpy.done4:
 // CHECK2-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK2-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP7]], %struct.St* [[AGG_TMP6]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK2-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 8
 // CHECK2-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 1
 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK2:       cond.true:
 // CHECK2-NEXT:    br label [[COND_END:%.*]]
 // CHECK2:       cond.false:
 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    br label [[COND_END]]
 // CHECK2:       cond.end:
 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK2:       omp.inner.for.cond:
 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
 // CHECK2-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK2:       omp.inner.for.cond.cleanup:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK2:       omp.inner.for.body:
 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK2-NEXT:    [[TMP16:%.*]] = zext i32 [[TMP15]] to i64
 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK2-NEXT:    [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[CONV9:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
 // CHECK2-NEXT:    store i32 [[TMP19]], i32* [[CONV9]], align 4
 // CHECK2-NEXT:    [[TMP20:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
 // CHECK2-NEXT:    [[TMP21:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 8
 // CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [2 x i32]*, i64, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP16]], i64 [[TMP18]], [2 x i32]* [[VEC2]], i64 [[TMP20]], [2 x %struct.S.0]* [[S_ARR3]], %struct.S.0* [[TMP21]])
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK2:       omp.inner.for.inc:
 // CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP22]], [[TMP23]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK2:       omp.inner.for.end:
 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK2:       omp.loop.exit:
 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]])
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN10]], i64 2
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP26]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done11:
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK2-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
 // CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
 // CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    [[VEC4:%.*]] = alloca [2 x i32], align 4
 // CHECK2-NEXT:    [[S_ARR5:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK2-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK2-NEXT:    [[VAR7:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK2-NEXT:    [[AGG_TMP8:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK2-NEXT:    [[_TMP9:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
 // CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
 // CHECK2-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
 // CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
 // CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK2-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
 // CHECK2-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK2-NEXT:    [[CONV3:%.*]] = trunc i64 [[TMP4]] to i32
 // CHECK2-NEXT:    store i32 [[CONV2]], i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK2-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
 // CHECK2-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i64 8, i1 false)
 // CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK2:       omp.arraycpy.body:
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK2:       omp.arraycpy.done6:
 // CHECK2-NEXT:    [[TMP9:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
 // CHECK2-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]])
 // CHECK2-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR7]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP9]], %struct.St* [[AGG_TMP8]])
 // CHECK2-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP8]]) #[[ATTR2]]
 // CHECK2-NEXT:    store %struct.S.0* [[VAR7]], %struct.S.0** [[_TMP9]], align 8
 // CHECK2-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 1
 // CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK2:       cond.true:
 // CHECK2-NEXT:    br label [[COND_END:%.*]]
 // CHECK2:       cond.false:
 // CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    br label [[COND_END]]
 // CHECK2:       cond.end:
 // CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
 // CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK2-NEXT:    store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK2:       omp.inner.for.cond:
 // CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK2-NEXT:    [[CMP10:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
 // CHECK2-NEXT:    br i1 [[CMP10]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK2:       omp.inner.for.cond.cleanup:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK2:       omp.inner.for.body:
 // CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
 // CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 [[IDXPROM]]
 // CHECK2-NEXT:    store i32 [[TMP18]], i32* [[ARRAYIDX]], align 4
 // CHECK2-NEXT:    [[TMP20:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP9]], align 8
 // CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
 // CHECK2-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP21]] to i64
 // CHECK2-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i64 0, i64 [[IDXPROM11]]
 // CHECK2-NEXT:    [[TMP22:%.*]] = bitcast %struct.S.0* [[ARRAYIDX12]] to i8*
 // CHECK2-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP20]] to i8*
 // CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP22]], i8* align 4 [[TMP23]], i64 4, i1 false)
 // CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 // CHECK2:       omp.body.continue:
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK2:       omp.inner.for.inc:
 // CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK2-NEXT:    store i32 [[ADD13]], i32* [[DOTOMP_IV]], align 4
 // CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK2:       omp.inner.for.end:
 // CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK2:       omp.loop.exit:
 // CHECK2-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR7]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAY_BEGIN14:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR5]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN14]], i64 2
 // CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK2:       arraydestroy.body:
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN14]]
 // CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE15:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK2:       arraydestroy.done15:
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK2-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK2-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
 // CHECK2-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP0]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[F2]], align 4
 // CHECK2-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
 // CHECK2-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
 // CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
 // CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_codegen.cpp
 // CHECK2-SAME: () #[[ATTR0]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    call void @__cxx_global_var_init()
 // CHECK2-NEXT:    call void @__cxx_global_var_init.1()
 // CHECK2-NEXT:    call void @__cxx_global_var_init.2()
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK2-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK2-SAME: () #[[ATTR0]] {
 // CHECK2-NEXT:  entry:
 // CHECK2-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK2-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK3-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[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(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), float 1.000000e+00)
 // CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 1), float 2.000000e+00)
 // CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK3-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 4
 // CHECK3-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 4
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i32 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // 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: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK3-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK3-NEXT:    store float [[ADD]], float* [[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(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK3-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__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:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
 // CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* @t_var, align 4
 // CHECK3-NEXT:    store i32 [[TMP0]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
 // CHECK3-NEXT:    store i32 [[TMP2]], i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP1]], i32* [[TMP5]], align 4
 // CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP1]], i32* [[TMP7]], align 4
 // CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
 // CHECK3-NEXT:    store i8* null, i8** [[TMP8]], align 4
 // CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK3-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [2 x i32]**
 // CHECK3-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP10]], align 4
 // CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK3-NEXT:    [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [2 x i32]**
 // CHECK3-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP12]], align 4
 // CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
 // CHECK3-NEXT:    store i8* null, i8** [[TMP13]], align 4
 // CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK3-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x %struct.S]**
 // CHECK3-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP15]], align 4
 // CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK3-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [2 x %struct.S]**
 // CHECK3-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP17]], align 4
 // CHECK3-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
 // CHECK3-NEXT:    store i8* null, i8** [[TMP18]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK3-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.S**
 // CHECK3-NEXT:    store %struct.S* @var, %struct.S** [[TMP20]], align 4
 // CHECK3-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK3-NEXT:    [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.S**
 // CHECK3-NEXT:    store %struct.S* @var, %struct.S** [[TMP22]], align 4
 // CHECK3-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
 // CHECK3-NEXT:    store i8* null, i8** [[TMP23]], align 4
 // CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
 // CHECK3-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[TMP25]], align 4
 // CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
 // CHECK3-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[TMP27]], align 4
 // CHECK3-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
 // CHECK3-NEXT:    store i8* null, i8** [[TMP28]], align 4
 // CHECK3-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK3-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 2)
 // CHECK3-NEXT:    [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99.region_id, i32 5, i8** [[TMP29]], i8** [[TMP30]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK3-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
 // CHECK3-NEXT:    br i1 [[TMP32]], 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_l99(i32 [[TMP1]], [2 x i32]* @vec, [2 x %struct.S]* @s_arr, %struct.S* @var, i32 [[TMP3]]) #[[ATTR2]]
 // CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK3:       omp_offload.cont:
 // CHECK3-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
 // CHECK3-NEXT:    ret i32 [[CALL]]
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99
 // CHECK3-SAME: (i32 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4:[0-9]+]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP5]], i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32, [2 x %struct.S]*, %struct.S*, i32)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i32 [[TMP4]], [2 x %struct.S]* [[TMP1]], %struct.S* [[TMP2]], i32 [[TMP6]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, 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:    [[VEC1:%.*]] = alloca [2 x i32], align 4
 // CHECK3-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK3-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK3-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK3-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC1]] to i8*
 // CHECK3-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 8, i1 false)
 // CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK3:       omp.arraycpy.body:
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK3:       omp.arraycpy.done3:
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
 // CHECK3-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP5]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 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:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    br label [[COND_END]]
 // CHECK3:       cond.end:
 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK3:       omp.inner.for.cond:
 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
 // CHECK3-NEXT:    br i1 [[CMP6]], 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:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP16]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP18]], i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [2 x i32]*, i32, [2 x %struct.S]*, %struct.S*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP14]], i32 [[TMP15]], [2 x i32]* [[VEC1]], i32 [[TMP17]], [2 x %struct.S]* [[S_ARR2]], %struct.S* [[VAR4]], i32 [[TMP19]])
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK3:       omp.inner.for.inc:
 // CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[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:    [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]])
 // CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN7]], i32 2
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP24]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
 // CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK3:       arraydestroy.done8:
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN2StC1Ev
 // CHECK3-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK3-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN2StC2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN2StD1Ev
 // CHECK3-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK3-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR2]]
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[SIVAR_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:    [[VEC1:%.*]] = alloca [2 x i32], align 4
 // CHECK3-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK3-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK3-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK3-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK3-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC1]] to i8*
 // CHECK3-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i32 8, i1 false)
 // CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK3:       omp.arraycpy.body:
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK3:       omp.arraycpy.done3:
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
 // CHECK3-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP5]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 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:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    br label [[COND_END]]
 // CHECK3:       cond.end:
 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK3:       omp.inner.for.cond:
 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
 // CHECK3-NEXT:    br i1 [[CMP6]], 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:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I]], align 4
 // CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC1]], i32 0, i32 [[TMP18]]
 // CHECK3-NEXT:    store i32 [[TMP17]], i32* [[ARRAYIDX]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK3-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 [[TMP19]]
 // CHECK3-NEXT:    [[TMP20:%.*]] = bitcast %struct.S* [[ARRAYIDX7]] to i8*
 // CHECK3-NEXT:    [[TMP21:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP20]], i8* align 4 [[TMP21]], i32 4, i1 false)
 // CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
 // CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK3-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP22]]
 // CHECK3-NEXT:    store i32 [[ADD8]], i32* [[SIVAR_ADDR]], 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:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK3-NEXT:    store i32 [[ADD9]], i32* [[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:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i32 2
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK3:       arraydestroy.done11:
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
 // CHECK3-SAME: () #[[ATTR6:[0-9]+]] 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 %struct.S.0*, align 4
 // CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
 // CHECK3-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
 // CHECK3-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
 // CHECK3-NEXT:    store i32 0, i32* [[T_VAR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i32 8, i1 false)
 // CHECK3-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
 // CHECK3-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i32 1
 // CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
 // CHECK3-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP2]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[TMP8]], align 4
 // CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32*
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[TMP10]], align 4
 // CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
 // CHECK3-NEXT:    store i8* null, i8** [[TMP11]], align 4
 // CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK3-NEXT:    [[TMP13:%.*]] = bitcast i8** [[TMP12]] to [2 x i32]**
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP13]], align 4
 // CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK3-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x i32]**
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP15]], align 4
 // CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
 // CHECK3-NEXT:    store i8* null, i8** [[TMP16]], align 4
 // CHECK3-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK3-NEXT:    [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [2 x %struct.S.0]**
 // CHECK3-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP18]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK3-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [2 x %struct.S.0]**
 // CHECK3-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP20]], align 4
 // CHECK3-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
 // CHECK3-NEXT:    store i8* null, i8** [[TMP21]], align 4
 // CHECK3-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK3-NEXT:    [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.S.0**
 // CHECK3-NEXT:    store %struct.S.0* [[TMP5]], %struct.S.0** [[TMP23]], align 4
 // CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK3-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.S.0**
 // CHECK3-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[TMP25]], align 4
 // CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
 // CHECK3-NEXT:    store i8* null, i8** [[TMP26]], align 4
 // CHECK3-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK3-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 2)
 // CHECK3-NEXT:    [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.6, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK3-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
 // CHECK3-NEXT:    br i1 [[TMP30]], 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(i32 [[TMP3]], [2 x i32]* [[VEC]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[TMP4]]) #[[ATTR2]]
 // CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK3:       omp_offload.cont:
 // CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP31]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
 // CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK3:       arraydestroy.done2:
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[TMP32:%.*]] = load i32, i32* [[RETVAL]], align 4
 // CHECK3-NEXT:    ret i32 [[TMP32]]
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN2StC2Ev
 // CHECK3-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK3-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    store i32 0, i32* [[A]], align 4
 // CHECK3-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 1
 // CHECK3-NEXT:    store i32 0, i32* [[B]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
 // CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP0]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP1:%.*]] = load float, float* [[F2]], align 4
 // CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
 // CHECK3-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
 // CHECK3-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN2StD2Ev
 // CHECK3-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK3-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
 // CHECK3-SAME: (i32 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i32 [[TMP4]], [2 x %struct.S.0]* [[TMP1]], %struct.S.0* [[TMP5]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[_TMP1:%.*]] = 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:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK3-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK3-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK3-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK3-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK3-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK3-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 8, i1 false)
 // CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK3:       omp.arraycpy.body:
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK3:       omp.arraycpy.done4:
 // CHECK3-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK3-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP7]], %struct.St* [[AGG_TMP6]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK3-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 4
 // CHECK3-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 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:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    br label [[COND_END]]
 // CHECK3:       cond.end:
 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK3:       omp.inner.for.cond:
 // CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
 // CHECK3-NEXT:    br i1 [[CMP8]], 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:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP17]], i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 4
 // CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [2 x i32]*, i32, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP15]], i32 [[TMP16]], [2 x i32]* [[VEC2]], i32 [[TMP18]], [2 x %struct.S.0]* [[S_ARR3]], %struct.S.0* [[TMP19]])
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK3:       omp.inner.for.inc:
 // CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[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:    [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]])
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAY_BEGIN9:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN9]], i32 2
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP24]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN9]]
 // CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE10:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK3:       arraydestroy.done10:
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..5
 // CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK3-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK3-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    [[_TMP1:%.*]] = 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:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK3-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK3-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK3-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK3-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK3-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK3-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[TMP3]], i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK3-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK3-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i32 8, i1 false)
 // CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK3:       omp.arraycpy.body:
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK3-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK3-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK3:       omp.arraycpy.done4:
 // CHECK3-NEXT:    [[TMP9:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK3-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK3-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP9]], %struct.St* [[AGG_TMP6]])
 // CHECK3-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK3-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 4
 // CHECK3-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 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:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    br label [[COND_END]]
 // CHECK3:       cond.end:
 // CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
 // CHECK3-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK3-NEXT:    store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK3:       omp.inner.for.cond:
 // CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK3-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
 // CHECK3-NEXT:    br i1 [[CMP8]], 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:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i32 0, i32 [[TMP19]]
 // CHECK3-NEXT:    store i32 [[TMP18]], i32* [[ARRAYIDX]], align 4
 // CHECK3-NEXT:    [[TMP20:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 4
 // CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
 // CHECK3-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 [[TMP21]]
 // CHECK3-NEXT:    [[TMP22:%.*]] = bitcast %struct.S.0* [[ARRAYIDX9]] to i8*
 // CHECK3-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP20]] to i8*
 // CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP22]], i8* align 4 [[TMP23]], 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:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK3-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK3-NEXT:    store i32 [[ADD10]], i32* [[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:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK3-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK3-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAY_BEGIN11:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN11]], i32 2
 // CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK3:       arraydestroy.body:
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN11]]
 // CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE12:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK3:       arraydestroy.done12:
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK3-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK3-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 4
 // CHECK3-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP0]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP1:%.*]] = load i32, i32* [[F2]], align 4
 // CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
 // CHECK3-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
 // CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_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
 //
 //
 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK3-SAME: () #[[ATTR0]] {
 // CHECK3-NEXT:  entry:
 // CHECK3-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK3-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK4-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK4-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK4-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK4-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
 // CHECK4-SAME: () #[[ATTR0]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), float 1.000000e+00)
 // CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 1), float 2.000000e+00)
 // CHECK4-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK4-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 4
 // CHECK4-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 4
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i32 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done1:
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK4-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK4-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK4-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
 // CHECK4-SAME: () #[[ATTR0]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK4-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__dso_handle) #[[ATTR2]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@main
 // CHECK4-SAME: () #[[ATTR3:[0-9]+]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x i8*], align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x i8*], align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x i8*], align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* @t_var, align 4
 // CHECK4-NEXT:    store i32 [[TMP0]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
 // CHECK4-NEXT:    store i32 [[TMP2]], i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP1]], i32* [[TMP5]], align 4
 // CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP1]], i32* [[TMP7]], align 4
 // CHECK4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
 // CHECK4-NEXT:    store i8* null, i8** [[TMP8]], align 4
 // CHECK4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK4-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to [2 x i32]**
 // CHECK4-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP10]], align 4
 // CHECK4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK4-NEXT:    [[TMP12:%.*]] = bitcast i8** [[TMP11]] to [2 x i32]**
 // CHECK4-NEXT:    store [2 x i32]* @vec, [2 x i32]** [[TMP12]], align 4
 // CHECK4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
 // CHECK4-NEXT:    store i8* null, i8** [[TMP13]], align 4
 // CHECK4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK4-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x %struct.S]**
 // CHECK4-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP15]], align 4
 // CHECK4-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK4-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [2 x %struct.S]**
 // CHECK4-NEXT:    store [2 x %struct.S]* @s_arr, [2 x %struct.S]** [[TMP17]], align 4
 // CHECK4-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
 // CHECK4-NEXT:    store i8* null, i8** [[TMP18]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK4-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to %struct.S**
 // CHECK4-NEXT:    store %struct.S* @var, %struct.S** [[TMP20]], align 4
 // CHECK4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK4-NEXT:    [[TMP22:%.*]] = bitcast i8** [[TMP21]] to %struct.S**
 // CHECK4-NEXT:    store %struct.S* @var, %struct.S** [[TMP22]], align 4
 // CHECK4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
 // CHECK4-NEXT:    store i8* null, i8** [[TMP23]], align 4
 // CHECK4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
 // CHECK4-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[TMP25]], align 4
 // CHECK4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 4
 // CHECK4-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[TMP27]], align 4
 // CHECK4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
 // CHECK4-NEXT:    store i8* null, i8** [[TMP28]], align 4
 // CHECK4-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK4-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 2)
 // CHECK4-NEXT:    [[TMP31:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99.region_id, i32 5, i8** [[TMP29]], i8** [[TMP30]], i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([5 x i64], [5 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK4-NEXT:    [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
 // CHECK4-NEXT:    br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 // CHECK4:       omp_offload.failed:
 // CHECK4-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99(i32 [[TMP1]], [2 x i32]* @vec, [2 x %struct.S]* @s_arr, %struct.S* @var, i32 [[TMP3]]) #[[ATTR2]]
 // CHECK4-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK4:       omp_offload.cont:
 // CHECK4-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
 // CHECK4-NEXT:    ret i32 [[CALL]]
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l99
 // CHECK4-SAME: (i32 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4:[0-9]+]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP5]], i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32, [2 x %struct.S]*, %struct.S*, i32)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i32 [[TMP4]], [2 x %struct.S]* [[TMP1]], %struct.S* [[TMP2]], i32 [[TMP6]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 4
 // CHECK4-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK4-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK4-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK4-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC1]] to i8*
 // CHECK4-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 8, i1 false)
 // CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK4:       omp.arraycpy.body:
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK4:       omp.arraycpy.done3:
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
 // CHECK4-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP5]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 1
 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK4:       cond.true:
 // CHECK4-NEXT:    br label [[COND_END:%.*]]
 // CHECK4:       cond.false:
 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    br label [[COND_END]]
 // CHECK4:       cond.end:
 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP11]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK4:       omp.inner.for.cond:
 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
 // CHECK4-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK4:       omp.inner.for.cond.cleanup:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK4:       omp.inner.for.body:
 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP16]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP18]], i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 7, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [2 x i32]*, i32, [2 x %struct.S]*, %struct.S*, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP14]], i32 [[TMP15]], [2 x i32]* [[VEC1]], i32 [[TMP17]], [2 x %struct.S]* [[S_ARR2]], %struct.S* [[VAR4]], i32 [[TMP19]])
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK4:       omp.inner.for.inc:
 // CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK4:       omp.inner.for.end:
 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK4:       omp.loop.exit:
 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]])
 // CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN7]], i32 2
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP24]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done8:
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN2StC1Ev
 // CHECK4-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK4-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN2StC2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN2StD1Ev
 // CHECK4-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK4-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR2]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR4]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 4
 // CHECK4-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
 // CHECK4-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK4-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
 // CHECK4-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK4-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC1]] to i8*
 // CHECK4-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i32 8, i1 false)
 // CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
 // CHECK4-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK4:       omp.arraycpy.body:
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK4:       omp.arraycpy.done3:
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
 // CHECK4-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP2]], %struct.St* [[AGG_TMP5]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 1
 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK4:       cond.true:
 // CHECK4-NEXT:    br label [[COND_END:%.*]]
 // CHECK4:       cond.false:
 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    br label [[COND_END]]
 // CHECK4:       cond.end:
 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK4:       omp.inner.for.cond:
 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
 // CHECK4-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK4:       omp.inner.for.cond.cleanup:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK4:       omp.inner.for.body:
 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I]], align 4
 // CHECK4-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC1]], i32 0, i32 [[TMP18]]
 // CHECK4-NEXT:    store i32 [[TMP17]], i32* [[ARRAYIDX]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK4-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 [[TMP19]]
 // CHECK4-NEXT:    [[TMP20:%.*]] = bitcast %struct.S* [[ARRAYIDX7]] to i8*
 // CHECK4-NEXT:    [[TMP21:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP20]], i8* align 4 [[TMP21]], i32 4, i1 false)
 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
 // CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP23]], [[TMP22]]
 // CHECK4-NEXT:    store i32 [[ADD8]], i32* [[SIVAR_ADDR]], align 4
 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 // CHECK4:       omp.body.continue:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK4:       omp.inner.for.inc:
 // CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK4-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK4:       omp.inner.for.end:
 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK4:       omp.loop.exit:
 // CHECK4-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i32 2
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done11:
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
 // CHECK4-SAME: () #[[ATTR6:[0-9]+]] comdat {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK4-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
 // CHECK4-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK4-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4
 // CHECK4-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4
 // CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
 // CHECK4-NEXT:    store i32 0, i32* [[T_VAR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i32 8, i1 false)
 // CHECK4-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK4-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
 // CHECK4-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i32 1
 // CHECK4-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
 // CHECK4-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP2]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[TMP8]], align 4
 // CHECK4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32*
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[TMP10]], align 4
 // CHECK4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
 // CHECK4-NEXT:    store i8* null, i8** [[TMP11]], align 4
 // CHECK4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
 // CHECK4-NEXT:    [[TMP13:%.*]] = bitcast i8** [[TMP12]] to [2 x i32]**
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP13]], align 4
 // CHECK4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1
 // CHECK4-NEXT:    [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [2 x i32]**
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[TMP15]], align 4
 // CHECK4-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
 // CHECK4-NEXT:    store i8* null, i8** [[TMP16]], align 4
 // CHECK4-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
 // CHECK4-NEXT:    [[TMP18:%.*]] = bitcast i8** [[TMP17]] to [2 x %struct.S.0]**
 // CHECK4-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP18]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2
 // CHECK4-NEXT:    [[TMP20:%.*]] = bitcast i8** [[TMP19]] to [2 x %struct.S.0]**
 // CHECK4-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[TMP20]], align 4
 // CHECK4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
 // CHECK4-NEXT:    store i8* null, i8** [[TMP21]], align 4
 // CHECK4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
 // CHECK4-NEXT:    [[TMP23:%.*]] = bitcast i8** [[TMP22]] to %struct.S.0**
 // CHECK4-NEXT:    store %struct.S.0* [[TMP5]], %struct.S.0** [[TMP23]], align 4
 // CHECK4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3
 // CHECK4-NEXT:    [[TMP25:%.*]] = bitcast i8** [[TMP24]] to %struct.S.0**
 // CHECK4-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[TMP25]], align 4
 // CHECK4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
 // CHECK4-NEXT:    store i8* null, i8** [[TMP26]], align 4
 // CHECK4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
 // CHECK4-NEXT:    call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 2)
 // CHECK4-NEXT:    [[TMP29:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes.6, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes.7, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
 // CHECK4-NEXT:    [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
 // CHECK4-NEXT:    br i1 [[TMP30]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
 // CHECK4:       omp_offload.failed:
 // CHECK4-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56(i32 [[TMP3]], [2 x i32]* [[VEC]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[TMP4]]) #[[ATTR2]]
 // CHECK4-NEXT:    br label [[OMP_OFFLOAD_CONT]]
 // CHECK4:       omp_offload.cont:
 // CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP31]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done2:
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[TMP32:%.*]] = load i32, i32* [[RETVAL]], align 4
 // CHECK4-NEXT:    ret i32 [[TMP32]]
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN2StC2Ev
 // CHECK4-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK4-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    store i32 0, i32* [[A]], align 4
 // CHECK4-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 1
 // CHECK4-NEXT:    store i32 0, i32* [[B]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
 // CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
 // CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP0]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP1:%.*]] = load float, float* [[F2]], align 4
 // CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK4-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
 // CHECK4-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
 // CHECK4-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN2StD2Ev
 // CHECK4-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 4
 // CHECK4-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l56
 // CHECK4-SAME: (i32 [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP5:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), [2 x i32]* [[TMP0]], i32 [[TMP4]], [2 x %struct.S.0]* [[TMP1]], %struct.S.0* [[TMP5]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..4
 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK4-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK4-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK4-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK4-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK4-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK4-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 8, i1 false)
 // CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK4:       omp.arraycpy.body:
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK4:       omp.arraycpy.done4:
 // CHECK4-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK4-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP7]], %struct.St* [[AGG_TMP6]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK4-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 4
 // CHECK4-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP9]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 1
 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK4:       cond.true:
 // CHECK4-NEXT:    br label [[COND_END:%.*]]
 // CHECK4:       cond.false:
 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    br label [[COND_END]]
 // CHECK4:       cond.end:
 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK4:       omp.inner.for.cond:
 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
 // CHECK4-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK4:       omp.inner.for.cond.cleanup:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK4:       omp.inner.for.body:
 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP17]], i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 4
 // CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [2 x i32]*, i32, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32 [[TMP15]], i32 [[TMP16]], [2 x i32]* [[VEC2]], i32 [[TMP18]], [2 x %struct.S.0]* [[S_ARR3]], %struct.S.0* [[TMP19]])
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK4:       omp.inner.for.inc:
 // CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK4:       omp.inner.for.end:
 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK4:       omp.loop.exit:
 // CHECK4-NEXT:    [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]])
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAY_BEGIN9:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN9]], i32 2
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP24]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN9]]
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE10:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done10:
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..5
 // CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[DOTPREVIOUS_LB_:%.*]], i32 [[DOTPREVIOUS_UB_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
 // CHECK4-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
 // CHECK4-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
 // CHECK4-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
 // CHECK4-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
 // CHECK4-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
 // CHECK4-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
 // CHECK4-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
 // CHECK4-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK4-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[TMP3]], i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK4-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
 // CHECK4-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP5]], i8* align 4 [[TMP6]], i32 8, i1 false)
 // CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP1]] to %struct.S.0*
 // CHECK4-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
 // CHECK4:       omp.arraycpy.body:
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], %struct.S.0* nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], %struct.St* [[AGG_TMP]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
 // CHECK4-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
 // CHECK4-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
 // CHECK4:       omp.arraycpy.done4:
 // CHECK4-NEXT:    [[TMP9:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 4
 // CHECK4-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
 // CHECK4-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP9]], %struct.St* [[AGG_TMP6]])
 // CHECK4-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR2]]
 // CHECK4-NEXT:    store %struct.S.0* [[VAR5]], %struct.S.0** [[_TMP7]], align 4
 // CHECK4-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP11]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 1
 // CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK4:       cond.true:
 // CHECK4-NEXT:    br label [[COND_END:%.*]]
 // CHECK4:       cond.false:
 // CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    br label [[COND_END]]
 // CHECK4:       cond.end:
 // CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
 // CHECK4-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK4-NEXT:    store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK4:       omp.inner.for.cond:
 // CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK4-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
 // CHECK4-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
 // CHECK4:       omp.inner.for.cond.cleanup:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
 // CHECK4:       omp.inner.for.body:
 // CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I]], align 4
 // CHECK4-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i32 0, i32 [[TMP19]]
 // CHECK4-NEXT:    store i32 [[TMP18]], i32* [[ARRAYIDX]], align 4
 // CHECK4-NEXT:    [[TMP20:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP7]], align 4
 // CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
 // CHECK4-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 [[TMP21]]
 // CHECK4-NEXT:    [[TMP22:%.*]] = bitcast %struct.S.0* [[ARRAYIDX9]] to i8*
 // CHECK4-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[TMP20]] to i8*
 // CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP22]], i8* align 4 [[TMP23]], i32 4, i1 false)
 // CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 // CHECK4:       omp.body.continue:
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK4:       omp.inner.for.inc:
 // CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP24]], 1
 // CHECK4-NEXT:    store i32 [[ADD10]], i32* [[DOTOMP_IV]], align 4
 // CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK4:       omp.inner.for.end:
 // CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK4:       omp.loop.exit:
 // CHECK4-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
 // CHECK4-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
 // CHECK4-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP26]])
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAY_BEGIN11:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN11]], i32 2
 // CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK4:       arraydestroy.body:
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP27]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
 // CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN11]]
 // CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE12:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK4:       arraydestroy.done12:
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK4-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 // CHECK4-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[S:%.*]], %struct.St* [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 4
 // CHECK4-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP0]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[F2]], align 4
 // CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
 // CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
 // CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
 // CHECK4-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
 // CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
 // CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_codegen.cpp
 // CHECK4-SAME: () #[[ATTR0]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    call void @__cxx_global_var_init()
 // CHECK4-NEXT:    call void @__cxx_global_var_init.1()
 // CHECK4-NEXT:    call void @__cxx_global_var_init.2()
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK4-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK4-SAME: () #[[ATTR0]] {
 // CHECK4-NEXT:  entry:
 // CHECK4-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK4-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK9-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK9-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK9-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK9-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK9-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK9-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[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(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 0), float 1.000000e+00)
 // CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 1), float 2.000000e+00)
 // CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK9-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK9-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK9-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
 // CHECK9-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
 // CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK9:       arraydestroy.body:
 // CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK9-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // 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: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK9-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK9-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK9-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK9-NEXT:    store float [[ADD]], float* [[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(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK9-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__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, i32* [[RETVAL]], align 4
 // CHECK9-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* 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 [[G:%.*]], i64 [[SIVAR:%.*]], i64 [[G1:%.*]]) #[[ATTR5:[0-9]+]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[G_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G1_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV2:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK9-NEXT:    store i32* [[CONV2]], i32** [[TMP]], align 8
 // CHECK9-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK9-NEXT:    [[CONV3:%.*]] = bitcast i64* [[G_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP0]], i32* [[CONV3]], align 4
 // CHECK9-NEXT:    [[TMP1:%.*]] = load i64, i64* [[G_CASTED]], align 8
 // CHECK9-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK9-NEXT:    [[TMP3:%.*]] = load volatile i32, i32* [[TMP2]], align 4
 // CHECK9-NEXT:    [[CONV4:%.*]] = bitcast i64* [[G1_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP3]], i32* [[CONV4]], align 4
 // CHECK9-NEXT:    [[TMP4:%.*]] = load i64, i64* [[G1_CASTED]], align 8
 // CHECK9-NEXT:    [[TMP5:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK9-NEXT:    [[CONV5:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP5]], i32* [[CONV5]], align 4
 // CHECK9-NEXT:    [[TMP6:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i64 [[TMP1]], i64 [[TMP4]], i64 [[TMP6]])
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[G:%.*]], i64 [[G1:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR5]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[G_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G1_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK9-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV2:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK9-NEXT:    store i32* [[CONV1]], i32** [[TMP]], align 8
 // CHECK9-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK9-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK9-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK9-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
 // CHECK9-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[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, i32* [[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]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK9-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK9-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK9:       omp.inner.for.cond:
 // CHECK9-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK9-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
 // CHECK9-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 // CHECK9:       omp.inner.for.body:
 // CHECK9-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK9-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
 // CHECK9-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK9-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
 // CHECK9-NEXT:    [[TMP11:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK9-NEXT:    [[CONV5:%.*]] = bitcast i64* [[G_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP11]], i32* [[CONV5]], align 4
 // CHECK9-NEXT:    [[TMP12:%.*]] = load i64, i64* [[G_CASTED]], align 8
 // CHECK9-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK9-NEXT:    [[TMP14:%.*]] = load volatile i32, i32* [[TMP13]], align 4
 // CHECK9-NEXT:    [[CONV6:%.*]] = bitcast i64* [[G1_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP14]], i32* [[CONV6]], align 4
 // CHECK9-NEXT:    [[TMP15:%.*]] = load i64, i64* [[G1_CASTED]], align 8
 // CHECK9-NEXT:    [[TMP16:%.*]] = load i32, i32* [[CONV2]], align 8
 // CHECK9-NEXT:    [[CONV7:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK9-NEXT:    store i32 [[TMP16]], i32* [[CONV7]], align 4
 // CHECK9-NEXT:    [[TMP17:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]], i64 [[TMP12]], i64 [[TMP15]], i64 [[TMP17]])
 // CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK9:       omp.inner.for.inc:
 // CHECK9-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
 // CHECK9-NEXT:    store i32 [[ADD]], i32* [[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(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[G:%.*]], i64 [[G1:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR5]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK9-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK9-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
 // CHECK9-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK9-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK9-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK9-NEXT:    [[CONV2:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK9-NEXT:    store i32* [[CONV1]], i32** [[TMP]], align 8
 // CHECK9-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK9-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK9-NEXT:    [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK9-NEXT:    [[CONV4:%.*]] = trunc i64 [[TMP0]] to i32
 // CHECK9-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK9-NEXT:    [[CONV5:%.*]] = trunc i64 [[TMP1]] to i32
 // CHECK9-NEXT:    store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4
 // CHECK9-NEXT:    store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4
 // CHECK9-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK9-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK9-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK9-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
 // CHECK9-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK9-NEXT:    [[TMP4:%.*]] = load i32, i32* [[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, i32* [[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]], i32* [[DOTOMP_UB]], align 4
 // CHECK9-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK9-NEXT:    store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK9:       omp.inner.for.cond:
 // CHECK9-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK9-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
 // CHECK9-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 // CHECK9:       omp.inner.for.body:
 // CHECK9-NEXT:    [[TMP9:%.*]] = load i32, i32* [[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]], i32* [[I]], align 4
 // CHECK9-NEXT:    store i32 1, i32* [[CONV]], align 8
 // CHECK9-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK9-NEXT:    store volatile i32 1, i32* [[TMP10]], align 4
 // CHECK9-NEXT:    store i32 2, i32* [[CONV2]], align 8
 // CHECK9-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
 // CHECK9-NEXT:    store i32* [[CONV]], i32** [[TMP11]], align 8
 // CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
 // CHECK9-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK9-NEXT:    store i32* [[TMP13]], i32** [[TMP12]], align 8
 // CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
 // CHECK9-NEXT:    store i32* [[CONV2]], i32** [[TMP14]], align 8
 // CHECK9-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* 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, i32* [[DOTOMP_IV]], align 4
 // CHECK9-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP15]], 1
 // CHECK9-NEXT:    store i32 [[ADD7]], i32* [[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(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_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
 //
 //
 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK9-SAME: () #[[ATTR0]] {
 // CHECK9-NEXT:  entry:
 // CHECK9-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK9-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@__cxx_global_var_init
 // CHECK10-SAME: () #[[ATTR0:[0-9]+]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) @test)
 // CHECK10-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @test to i8*), i8* @__dso_handle) #[[ATTR2:[0-9]+]]
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR2]]
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK10-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK10-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
 // CHECK10-NEXT:    store float [[CONV]], float* [[F]], align 4
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@__cxx_global_var_init.1
 // CHECK10-SAME: () #[[ATTR0]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 0), float 1.000000e+00)
 // CHECK10-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i64 0, i64 1), float 2.000000e+00)
 // CHECK10-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* @__cxx_global_array_dtor, i8* null, i8* @__dso_handle) #[[ATTR2]]
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK10-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@__cxx_global_array_dtor
 // CHECK10-SAME: (i8* [[TMP0:%.*]]) #[[ATTR0]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
 // CHECK10-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
 // CHECK10-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
 // CHECK10:       arraydestroy.body:
 // CHECK10-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ getelementptr inbounds ([[STRUCT_S:%.*]], %struct.S* getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0), i64 2), [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
 // CHECK10-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
 // CHECK10-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR2]]
 // CHECK10-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], getelementptr inbounds ([2 x %struct.S], [2 x %struct.S]* @s_arr, i32 0, i32 0)
 // CHECK10-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
 // CHECK10:       arraydestroy.done1:
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
 // CHECK10-SAME: (%struct.S* nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
 // CHECK10-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
 // CHECK10-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
 // CHECK10-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
 // CHECK10-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
 // CHECK10-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
 // CHECK10-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 4
 // CHECK10-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
 // CHECK10-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
 // CHECK10-NEXT:    store float [[ADD]], float* [[F]], align 4
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@__cxx_global_var_init.2
 // CHECK10-SAME: () #[[ATTR0]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) @var, float 3.000000e+00)
 // CHECK10-NEXT:    [[TMP0:%.*]] = call i32 @__cxa_atexit(void (i8*)* bitcast (void (%struct.S*)* @_ZN1SIfED1Ev to void (i8*)*), i8* bitcast (%struct.S* @var to i8*), i8* @__dso_handle) #[[ATTR2]]
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@main
 // CHECK10-SAME: () #[[ATTR3:[0-9]+]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
 // CHECK10-NEXT:    store i32 0, i32* [[RETVAL]], align 4
 // CHECK10-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull align 1 dereferenceable(1) [[REF_TMP]])
 // CHECK10-NEXT:    ret i32 0
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l75
 // CHECK10-SAME: (i64 [[G:%.*]], i64 [[SIVAR:%.*]], i64 [[G1:%.*]]) #[[ATTR5:[0-9]+]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[G_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G1_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK10-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV2:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK10-NEXT:    store i32* [[CONV2]], i32** [[TMP]], align 8
 // CHECK10-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK10-NEXT:    [[CONV3:%.*]] = bitcast i64* [[G_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP0]], i32* [[CONV3]], align 4
 // CHECK10-NEXT:    [[TMP1:%.*]] = load i64, i64* [[G_CASTED]], align 8
 // CHECK10-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK10-NEXT:    [[TMP3:%.*]] = load volatile i32, i32* [[TMP2]], align 4
 // CHECK10-NEXT:    [[CONV4:%.*]] = bitcast i64* [[G1_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP3]], i32* [[CONV4]], align 4
 // CHECK10-NEXT:    [[TMP4:%.*]] = load i64, i64* [[G1_CASTED]], align 8
 // CHECK10-NEXT:    [[TMP5:%.*]] = load i32, i32* [[CONV1]], align 8
 // CHECK10-NEXT:    [[CONV5:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP5]], i32* [[CONV5]], align 4
 // CHECK10-NEXT:    [[TMP6:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK10-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i64 [[TMP1]], i64 [[TMP4]], i64 [[TMP6]])
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined.
 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[G:%.*]], i64 [[G1:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR5]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[G_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G1_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK10-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK10-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV1:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV2:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK10-NEXT:    store i32* [[CONV1]], i32** [[TMP]], align 8
 // CHECK10-NEXT:    store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK10-NEXT:    store i32 1, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK10-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK10-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK10-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
 // CHECK10-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK10-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
 // CHECK10-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK10:       cond.true:
 // CHECK10-NEXT:    br label [[COND_END:%.*]]
 // CHECK10:       cond.false:
 // CHECK10-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    br label [[COND_END]]
 // CHECK10:       cond.end:
 // CHECK10-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
 // CHECK10-NEXT:    store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK10-NEXT:    store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK10:       omp.inner.for.cond:
 // CHECK10-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
 // CHECK10-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 // CHECK10:       omp.inner.for.body:
 // CHECK10-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
 // CHECK10-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
 // CHECK10-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
 // CHECK10-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
 // CHECK10-NEXT:    [[TMP11:%.*]] = load i32, i32* [[CONV]], align 8
 // CHECK10-NEXT:    [[CONV5:%.*]] = bitcast i64* [[G_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP11]], i32* [[CONV5]], align 4
 // CHECK10-NEXT:    [[TMP12:%.*]] = load i64, i64* [[G_CASTED]], align 8
 // CHECK10-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK10-NEXT:    [[TMP14:%.*]] = load volatile i32, i32* [[TMP13]], align 4
 // CHECK10-NEXT:    [[CONV6:%.*]] = bitcast i64* [[G1_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP14]], i32* [[CONV6]], align 4
 // CHECK10-NEXT:    [[TMP15:%.*]] = load i64, i64* [[G1_CASTED]], align 8
 // CHECK10-NEXT:    [[TMP16:%.*]] = load i32, i32* [[CONV2]], align 8
 // CHECK10-NEXT:    [[CONV7:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
 // CHECK10-NEXT:    store i32 [[TMP16]], i32* [[CONV7]], align 4
 // CHECK10-NEXT:    [[TMP17:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
 // CHECK10-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i64, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]], i64 [[TMP12]], i64 [[TMP15]], i64 [[TMP17]])
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK10:       omp.inner.for.inc:
 // CHECK10-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK10-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
 // CHECK10-NEXT:    store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK10:       omp.inner.for.end:
 // CHECK10-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK10:       omp.loop.exit:
 // CHECK10-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3
 // CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]], i64 [[G:%.*]], i64 [[G1:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR5]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[G1_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
 // CHECK10-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
 // CHECK10-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[I:%.*]] = alloca i32, align 4
 // CHECK10-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
 // CHECK10-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK10-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[G]], i64* [[G_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[G1]], i64* [[G1_ADDR]], align 8
 // CHECK10-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
 // CHECK10-NEXT:    [[CONV:%.*]] = bitcast i64* [[G_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV1:%.*]] = bitcast i64* [[G1_ADDR]] to i32*
 // CHECK10-NEXT:    [[CONV2:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
 // CHECK10-NEXT:    store i32* [[CONV1]], i32** [[TMP]], align 8
 // CHECK10-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
 // CHECK10-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
 // CHECK10-NEXT:    [[CONV4:%.*]] = trunc i64 [[TMP0]] to i32
 // CHECK10-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
 // CHECK10-NEXT:    [[CONV5:%.*]] = trunc i64 [[TMP1]] to i32
 // CHECK10-NEXT:    store i32 [[CONV4]], i32* [[DOTOMP_LB]], align 4
 // CHECK10-NEXT:    store i32 [[CONV5]], i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
 // CHECK10-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
 // CHECK10-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
 // CHECK10-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
 // CHECK10-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
 // CHECK10-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
 // CHECK10-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
 // CHECK10:       cond.true:
 // CHECK10-NEXT:    br label [[COND_END:%.*]]
 // CHECK10:       cond.false:
 // CHECK10-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    br label [[COND_END]]
 // CHECK10:       cond.end:
 // CHECK10-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
 // CHECK10-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
 // CHECK10-NEXT:    store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
 // CHECK10:       omp.inner.for.cond:
 // CHECK10-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
 // CHECK10-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
 // CHECK10-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
 // CHECK10:       omp.inner.for.body:
 // CHECK10-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
 // CHECK10-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
 // CHECK10-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
 // CHECK10-NEXT:    store i32 1, i32* [[CONV]], align 8
 // CHECK10-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK10-NEXT:    store volatile i32 1, i32* [[TMP10]], align 4
 // CHECK10-NEXT:    store i32 2, i32* [[CONV2]], align 8
 // CHECK10-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
 // CHECK10-NEXT:    store i32* [[CONV]], i32** [[TMP11]], align 8
 // CHECK10-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
 // CHECK10-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[TMP]], align 8
 // CHECK10-NEXT:    store i32* [[TMP13]], i32** [[TMP12]], align 8
 // CHECK10-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
 // CHECK10-NEXT:    store i32* [[CONV2]], i32** [[TMP14]], align 8
 // CHECK10-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* nonnull align 8 dereferenceable(24) [[REF_TMP]])
 // CHECK10-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
 // CHECK10:       omp.body.continue:
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
 // CHECK10:       omp.inner.for.inc:
 // CHECK10-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP15]], 1
 // CHECK10-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
 // CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND]]
 // CHECK10:       omp.inner.for.end:
 // CHECK10-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
 // CHECK10:       omp.loop.exit:
 // CHECK10-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@_GLOBAL__sub_I_teams_distribute_parallel_for_firstprivate_codegen.cpp
 // CHECK10-SAME: () #[[ATTR0]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    call void @__cxx_global_var_init()
 // CHECK10-NEXT:    call void @__cxx_global_var_init.1()
 // CHECK10-NEXT:    call void @__cxx_global_var_init.2()
 // CHECK10-NEXT:    ret void
 //
 //
 // CHECK10-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
 // CHECK10-SAME: () #[[ATTR0]] {
 // CHECK10-NEXT:  entry:
 // CHECK10-NEXT:    call void @__tgt_register_requires(i64 1)
 // CHECK10-NEXT:    ret void
 //
 //