// 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 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp -x c++ -std=c++11 -triple i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK4

// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp-simd -x c++ -std=c++11 -triple i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"

// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp -x c++ -std=c++11 -triple x86_64-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK12

// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp-simd -x c++ -std=c++11 -triple x86_64-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -fopenmp-simd -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"

// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17

// RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis -verify -fopenmp-simd -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// expected-no-diagnostics
#ifndef ARRAY
#ifndef HEADER
#define HEADER

enum omp_allocator_handle_t {
  omp_null_allocator = 0,
  omp_default_mem_alloc = 1,
  omp_large_cap_mem_alloc = 2,
  omp_const_mem_alloc = 3,
  omp_high_bw_mem_alloc = 4,
  omp_low_lat_mem_alloc = 5,
  omp_cgroup_mem_alloc = 6,
  omp_pteam_mem_alloc = 7,
  omp_thread_mem_alloc = 8,
  KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
};

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 __attribute__((aligned(128))) = 1212;

struct SS {
  int a;
  int b : 4;
  int &c;
  int e[4];
  SS(int &d) : a(0), b(0), c(d) {
#pragma omp parallel firstprivate(a, b, c, e)
#ifdef LAMBDA
    [&]() {
      ++this->a, --b, (this)->c /= 1;
#pragma omp parallel firstprivate(a, b, c)
      ++(this)->a, --b, this->c /= 1;
    }();
#elif defined(BLOCKS)
    ^{
      ++a;
      --this->b;
      (this)->c /= 1;
#pragma omp parallel firstprivate(a, b, c)
      ++(this)->a, --b, this->c /= 1;
    }();
#else
    ++this->a, --b, c /= 1, e[2] = 1111;
#endif
  }
};

template<typename T>
struct SST {
  T a;
  SST() : a(T()) {
#pragma omp parallel firstprivate(a)
#ifdef LAMBDA
    [&]() {
      [&]() {
        ++this->a;
#pragma omp parallel firstprivate(a)
        ++(this)->a;
      }();
    }();
#elif defined(BLOCKS)
    ^{
      ^{
        ++a;
#pragma omp parallel firstprivate(a)
        ++(this)->a;
      }();
    }();
#else
    ++(this)->a;
#endif
  }
};

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;
  SST<T> sst;
  T t_var __attribute__((aligned(128))) = T();
  T vec[] __attribute__((aligned(128))) = {1, 2};
  S<T> s_arr[] __attribute__((aligned(128))) = {1, 2};
  S<T> var __attribute__((aligned(128))) (3);
#pragma omp parallel firstprivate(t_var, vec, s_arr, var)
  {
    vec[0] = t_var;
    s_arr[0] = var;
  }
#pragma omp parallel firstprivate(t_var)
  {}
  return T();
}

int main() {
  static int sivar;
  SS ss(sivar);
#ifdef LAMBDA
  [&]() {
#pragma omp parallel firstprivate(g, sivar)
  {



    g = 1;
    sivar = 2;
    [&]() {
      g = 2;
      sivar = 4;
    }();
  }
  }();
  return 0;
#elif defined(BLOCKS)
  ^{
#pragma omp parallel firstprivate(g, sivar)
  {
    g = 1;
    sivar = 2;
    ^{
      g = 2;
      sivar = 4;
    }();
  }
  }();
  return 0;


#else
  S<float> test;
  int t_var = 0;
  int vec[] = {1, 2};
  S<float> s_arr[] = {1, 2};
  S<float> var(3);
#pragma omp parallel firstprivate(t_var, vec, s_arr, var, sivar)
  {
    vec[0] = t_var;
    s_arr[0] = var;
    sivar = 2;
  }
  const int a = 0;
#pragma omp parallel allocate(omp_default_mem_alloc: t_var) firstprivate(t_var, a)
  { t_var = a; }
  return tmain<int>();
#endif
}













#endif
#else

enum omp_allocator_handle_t {
  omp_null_allocator = 0,
  omp_default_mem_alloc = 1,
  omp_large_cap_mem_alloc = 2,
  omp_const_mem_alloc = 3,
  omp_high_bw_mem_alloc = 4,
  omp_low_lat_mem_alloc = 5,
  omp_cgroup_mem_alloc = 6,
  omp_pteam_mem_alloc = 7,
  omp_thread_mem_alloc = 8,
  KMP_ALLOCATOR_MAX_HANDLE = __UINTPTR_MAX__
};

struct St {
  int a, b;
  St() : a(0), b(0) {}
  St(const St &) { }
  ~St() {}
  void St_func(St s[2], int n, long double vla1[n]) {
    double vla2[n][n] __attribute__((aligned(128)));
    a = b;
#pragma omp parallel allocate(omp_thread_mem_alloc:vla2) firstprivate(s, vla1, vla2)
    vla1[b] = vla2[1][n - 1] = a = b;
  }
};

void array_func(float a[3], St s[2], int n, long double vla1[n]) {
  double vla2[n][n] __attribute__((aligned(128)));
#pragma omp parallel firstprivate(a, s, vla1, vla2)
  s[0].St_func(s, n, vla1);
  ;
}

#endif

// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[A:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR_CASTED1:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* 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.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i32 8, i1 false)
// CHECK1-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i32 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00)
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK1-NEXT:    store i32 [[TMP1]], i32* [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK1-NEXT:    store i32 [[TMP3]], i32* [[SIVAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], 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]* [[VEC]], i32 [[TMP2]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], i32 [[TMP4]])
// CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], i32* [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP6]])
// CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK1-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP8]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK1-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// 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 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
// 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 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// 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 {{[^@]+}}@.omp_outlined.
// CHECK1-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:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 4
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK1-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC1]] to i8*
// CHECK1-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 8, i1 false)
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], 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]], i32 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], 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]]) #[[ATTR4]]
// 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_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
// CHECK1-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]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 [[TMP7]], i32* [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8*
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP8]], i8* align 4 [[TMP9]], i32 4, i1 false)
// CHECK1-NEXT:    store i32 2, i32* [[SIVAR_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN7]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP10]], [[OMP_ARRAYCPY_DONE3]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_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 4
// CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
// 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 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 4
// 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 4
// CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// 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 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT:    [[DOTT_VAR__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP1]], i32 4, i8* inttoptr (i32 1 to i8*))
// CHECK1-NEXT:    [[DOTT_VAR__ADDR:%.*]] = bitcast i8* [[DOTT_VAR__VOID_ADDR]] to i32*
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTT_VAR__ADDR]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTT_VAR__ADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i32* [[DOTT_VAR__ADDR]] to i8*
// CHECK1-NEXT:    call void @__kmpc_free(i32 [[TMP1]], i8* [[TMP3]], i8* inttoptr (i32 1 to i8*))
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 128
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 128
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]])
// CHECK1-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK1-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 128 [[TMP0]], i8* align 128 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i32 8, i1 false)
// CHECK1-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i32 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]], i32 3)
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]])
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[T_VAR]])
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done1:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK1-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[A2:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[C7:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 4
// CHECK1-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[C_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK1-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK1-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK1-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK1-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK1-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK1-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[C]], align 4
// CHECK1-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32* [[A3]], i32** [[A2]], align 4
// CHECK1-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK1-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK1-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK1-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK1-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK1-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK1-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 4
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 4
// CHECK1-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK1-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    store i32 [[TMP3]], i32* [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], i32* [[B_CASTED]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK1-NEXT:    store i32 [[TMP8]], i32* [[C_CASTED]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[C_CASTED]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 4
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32, i32, i32, [4 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[C_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca [4 x i32]*, align 4
// CHECK1-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK1-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[B]], i32* [[B_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[C]], i32* [[C_ADDR]], align 4
// CHECK1-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK1-NEXT:    store i32* [[C_ADDR]], i32** [[_TMP1]], align 4
// CHECK1-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP2]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP2]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E3]] to i8*
// CHECK1-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 16, i1 false)
// CHECK1-NEXT:    store [4 x i32]* [[E3]], [4 x i32]** [[_TMP4]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP5]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK1-NEXT:    store i32 [[INC]], i32* [[TMP5]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK1-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK1-NEXT:    store i32 [[DEC]], i32* [[B_ADDR]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[_TMP1]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK1-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP9]], 1
// CHECK1-NEXT:    store i32 [[DIV]], i32* [[TMP8]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP4]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i32], [4 x i32]* [[TMP10]], i32 0, i32 2
// CHECK1-NEXT:    store i32 1111, i32* [[ARRAYIDX]], align 4
// 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 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 128
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK1-NEXT:    store float [[CONV]], float* [[F]], align 4
// 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 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// 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 128
// 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 {{[^@]+}}@_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 4
// CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 4
// 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 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// 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 4
// 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 4
// CHECK1-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], 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 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// 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 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
// CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 4
// CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN3SSTIiEC2Ev(%struct.SST* 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 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// 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 [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 4
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// CHECK1-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 128
// CHECK1-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK1-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128
// CHECK1-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP1]], align 128
// CHECK1-NEXT:    store i32 [[TMP4]], i32* [[T_VAR1]], align 128
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
// CHECK1-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 128 [[TMP5]], i8* align 128 [[TMP6]], i32 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:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
// 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* [ [[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]]) #[[ATTR4]]
// 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_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 @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.St* [[AGG_TMP6]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[T_VAR1]], align 128
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i32 0, i32 0
// CHECK1-NEXT:    store i32 [[TMP9]], i32* [[ARRAYIDX]], align 128
// CHECK1-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct.S.0* [[ARRAYIDX7]] to i8*
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast %struct.S.0* [[VAR5]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 128 [[TMP10]], i8* align 128 [[TMP11]], i32 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN8]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_ARRAYCPY_DONE4]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN8]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE9:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done9:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_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 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 4
// 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 {{[^@]+}}@_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 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 128
// CHECK1-NEXT:    store i32 [[TMP1]], i32* [[T_VAR1]], align 128
// 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 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// 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 128
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
// CHECK1-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 4
// CHECK1-NEXT:    [[A2:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK1-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SST]], %struct.SST* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32* [[A3]], i32** [[A2]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[A2]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT:    store i32 [[TMP1]], i32* [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SST*, i32)* @.omp_outlined..5 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]], i32 [[A:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK1-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// 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 128
// 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 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// 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 4
// 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 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 4
// CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP0]], align 4
// CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull align 4 dereferenceable(4) [[REF_TMP]])
// CHECK3-NEXT:    ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK3-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK3-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[A2:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C7:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 4
// CHECK3-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK3-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK3-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK3-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK3-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK3-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK3-NEXT:    store i32* [[TMP0]], i32** [[C]], align 4
// CHECK3-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    store i32* [[A3]], i32** [[A2]], align 4
// CHECK3-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK3-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK3-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK3-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK3-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK3-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 4
// CHECK3-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 4
// CHECK3-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK3-NEXT:    store i32 [[TMP3]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK3-NEXT:    store i32 [[TMP5]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK3-NEXT:    store i32 [[TMP8]], i32* [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 4
// CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32, i32, i32, [4 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[_TMP2:%.*]] = alloca [4 x i32]*, align 4
// CHECK3-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK3-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 4
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], 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 %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[B]], i32* [[B_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[C]], i32* [[C_ADDR]], align 4
// CHECK3-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 4
// CHECK3-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK3-NEXT:    store i32* [[C_ADDR]], i32** [[_TMP1]], align 4
// CHECK3-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E3]] to i8*
// CHECK3-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 16, i1 false)
// CHECK3-NEXT:    store [4 x i32]* [[E3]], [4 x i32]** [[_TMP4]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK3-NEXT:    store i32* [[TMP7]], i32** [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
// CHECK3-NEXT:    store i32* [[B_ADDR]], i32** [[TMP8]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[_TMP1]], align 4
// CHECK3-NEXT:    store i32* [[TMP10]], i32** [[TMP9]], align 4
// CHECK3-NEXT:    call void @_ZZN2SSC1ERiENKUlvE_clEv(%class.anon.0* nonnull align 4 dereferenceable(16) [[REF_TMP]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv
// CHECK3-SAME: (%class.anon.0* nonnull align 4 dereferenceable(16) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 4
// CHECK3-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP1:%.*]] = load %struct.SS*, %struct.SS** [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK3-NEXT:    store i32 [[INC]], i32* [[TMP3]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK3-NEXT:    store i32 [[DEC]], i32* [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[TMP8]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 1
// CHECK3-NEXT:    store i32 [[DIV]], i32* [[TMP9]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32*, i32** [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
// CHECK3-NEXT:    store i32 [[TMP13]], i32* [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP14:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32*, i32** [[TMP15]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[TMP16]], align 4
// CHECK3-NEXT:    store i32 [[TMP17]], i32* [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP20:%.*]] = load i32*, i32** [[TMP19]], align 4
// CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4
// CHECK3-NEXT:    store i32 [[TMP21]], i32* [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[C_CASTED]], align 4
// CHECK3-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32, i32, i32)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[TMP1]], i32 [[TMP14]], i32 [[TMP18]], i32 [[TMP22]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = 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 %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[B]], i32* [[B_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[C]], i32* [[C_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK3-NEXT:    store i32* [[C_ADDR]], i32** [[_TMP1]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK3-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK3-NEXT:    store i32 [[DEC]], i32* [[B_ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP1]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK3-NEXT:    store i32 [[DIV]], i32* [[TMP4]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[G_ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[G1:%.*]] = alloca i32, align 128
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], 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* [[G]], i32** [[G_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* [[TMP0]], align 128
// CHECK3-NEXT:    store i32 [[TMP1]], i32* [[G1]], align 128
// CHECK3-NEXT:    store i32 1, i32* [[G1]], align 128
// CHECK3-NEXT:    store i32 2, i32* [[SIVAR_ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT:    store i32* [[G1]], i32** [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 1
// CHECK3-NEXT:    store i32* [[SIVAR_ADDR]], i32** [[TMP3]], align 4
// CHECK3-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.1* nonnull align 4 dereferenceable(8) [[REF_TMP]])
// CHECK3-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@main
// CHECK4-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, align 4
// CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 4
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store i8* bitcast (void (i8*)* @__main_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i32, i32, i8*, i8* }* @__block_descriptor_tmp.1 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK4-NEXT:    store i32 [[TMP0]], i32* [[BLOCK_CAPTURED]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]] to void ()*
// CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP1]] to %struct.__block_literal_generic*
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK4-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[TMP2]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to void (i8*)*
// CHECK4-NEXT:    call void [[TMP5]](i8* [[TMP3]])
// CHECK4-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK4-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK4-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 4 dereferenceable(28) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*, align 4
// CHECK4-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*
// CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>** [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK4-NEXT:    store i32 [[TMP0]], i32* [[SIVAR_CASTED]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
// CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* @g, i32 [[TMP1]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[G_ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[G1:%.*]] = alloca i32, align 128
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, align 128
// 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* [[G]], i32** [[G_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* [[TMP0]], align 128
// CHECK4-NEXT:    store i32 [[TMP1]], i32* [[G1]], align 128
// CHECK4-NEXT:    store i32 1, i32* [[G1]], align 128
// CHECK4-NEXT:    store i32 2, i32* [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 128
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 8
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store i8* bitcast (void (i8*)* @g_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i32, i32, i8*, i8* }* @__block_descriptor_tmp to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 16
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP2:%.*]] = load volatile i32, i32* [[G1]], align 128
// CHECK4-NEXT:    store volatile i32 [[TMP2]], i32* [[BLOCK_CAPTURED]], align 128
// CHECK4-NEXT:    [[BLOCK_CAPTURED2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[TMP3]], i32* [[BLOCK_CAPTURED2]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]] to void ()*
// CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP4]] to %struct.__block_literal_generic*
// CHECK4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP6:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK4-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP5]], align 4
// CHECK4-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to void (i8*)*
// CHECK4-NEXT:    call void [[TMP8]](i8* [[TMP6]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g_block_invoke
// CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>*, align 4
// CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>*
// CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>** [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    store i32 2, i32* [[BLOCK_CAPTURE_ADDR]], align 128
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [104 x i8], i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    store i32 4, i32* [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK4-SAME: (%struct.SS* nonnull align 4 dereferenceable(28) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[A2:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C7:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 4
// CHECK4-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK4-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK4-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK4-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK4-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK4-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK4-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK4-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 4
// CHECK4-NEXT:    store i32* [[TMP0]], i32** [[C]], align 4
// CHECK4-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK4-NEXT:    store i32* [[A3]], i32** [[A2]], align 4
// CHECK4-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK4-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK4-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK4-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK4-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK4-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK4-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 4
// CHECK4-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 4
// CHECK4-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK4-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK4-NEXT:    store i32 [[TMP3]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK4-NEXT:    store i32 [[TMP5]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 4
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK4-NEXT:    store i32 [[TMP8]], i32* [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 4
// CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32, i32, i32, [4 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 4
// CHECK4-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[_TMP2:%.*]] = alloca [4 x i32]*, align 4
// CHECK4-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK4-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, 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 %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[B]], i32* [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[C]], i32* [[C_ADDR]], align 4
// CHECK4-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 4
// CHECK4-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK4-NEXT:    store i32* [[C_ADDR]], i32** [[_TMP1]], align 4
// CHECK4-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP2]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP2]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E3]] to i8*
// CHECK4-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i32 16, i1 false)
// CHECK4-NEXT:    store [4 x i32]* [[E3]], [4 x i32]** [[_TMP4]], align 4
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 4
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store i8* bitcast (void (i8*)* @g_block_invoke_2 to i8*), i8** [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i32, i32, i8*, i8* }* @__block_descriptor_tmp.4 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[BLOCK_CAPTURED_THIS_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK4-NEXT:    store i32* [[TMP5]], i32** [[BLOCK_CAPTURED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED5:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[TMP6]], i32* [[BLOCK_CAPTURED5]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED6:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP1]], align 4
// CHECK4-NEXT:    store i32* [[TMP7]], i32** [[BLOCK_CAPTURED6]], align 4
// CHECK4-NEXT:    [[TMP8:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]] to void ()*
// CHECK4-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP8]] to %struct.__block_literal_generic*
// CHECK4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP10:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK4-NEXT:    [[TMP11:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK4-NEXT:    [[TMP12:%.*]] = bitcast i8* [[TMP11]] to void (i8*)*
// CHECK4-NEXT:    call void [[TMP12]](i8* [[TMP10]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g_block_invoke_2
// CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>*, align 4
// CHECK4-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[B_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>*
// CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>** [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[THIS:%.*]] = load %struct.SS*, %struct.SS** [[BLOCK_CAPTURED_THIS]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK4-NEXT:    store i32 [[INC]], i32* [[TMP0]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP2]], -1
// CHECK4-NEXT:    store i32 [[DEC]], i32* [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR2]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP4]], 1
// CHECK4-NEXT:    store i32 [[DIV]], i32* [[TMP3]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR3]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP5]], align 4
// CHECK4-NEXT:    store i32 [[TMP6]], i32* [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[A_CASTED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR4]], align 4
// CHECK4-NEXT:    store i32 [[TMP8]], i32* [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[B_CASTED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR5:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32, i32* }>* [[BLOCK]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR5]], align 4
// CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
// CHECK4-NEXT:    store i32 [[TMP11]], i32* [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[C_CASTED]], align 4
// CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i32, i32, i32)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SS* [[THIS]], i32 [[TMP7]], i32 [[TMP9]], i32 [[TMP12]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[TMP:%.*]] = alloca i32*, align 4
// CHECK4-NEXT:    [[_TMP1:%.*]] = 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 %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[B]], i32* [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[C]], i32* [[C_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32* [[A_ADDR]], i32** [[TMP]], align 4
// CHECK4-NEXT:    store i32* [[C_ADDR]], i32** [[_TMP1]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK4-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[B_ADDR]], align 4
// CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK4-NEXT:    store i32 [[DEC]], i32* [[B_ADDR]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP1]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK4-NEXT:    store i32 [[DIV]], i32* [[TMP4]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
// CHECK9-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK9-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK9-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK9-NEXT:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[T_VAR_CASTED2:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK9-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00)
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP1]], i32* [[CONV]], align 4
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP3]], i32* [[CONV1]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], 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]* [[VEC]], i64 [[TMP2]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]], i64 [[TMP4]])
// CHECK9-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK9-NEXT:    [[CONV3:%.*]] = bitcast i64* [[T_VAR_CASTED2]] to i32*
// CHECK9-NEXT:    store i32 [[TMP5]], i32* [[CONV3]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i64, i64* [[T_VAR_CASTED2]], align 8
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP6]])
// CHECK9-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK9-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP7]], [[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]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE4:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done4:
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP8]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK9-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK9-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// 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 @_ZN1SIfEC2Ev(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]])
// 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 {{[^@]+}}@.omp_outlined.
// CHECK9-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:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK9-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S], align 4
// CHECK9-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK9-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK9-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
// 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 [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
// CHECK9-NEXT:    [[TMP1:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
// CHECK9-NEXT:    [[TMP3:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
// CHECK9-NEXT:    [[TMP4:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP3]], i8* align 4 [[TMP4]], i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP5:%.*]] = bitcast [2 x %struct.S]* [[TMP1]] to %struct.S*
// CHECK9-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK9:       omp.arraycpy.body:
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK9-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]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
// CHECK9:       omp.arraycpy.done4:
// CHECK9-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
// CHECK9-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]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i64 0, i64 0
// CHECK9-NEXT:    store i32 [[TMP7]], i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i64 0, i64 0
// CHECK9-NEXT:    [[TMP8:%.*]] = bitcast %struct.S* [[ARRAYIDX7]] to i8*
// CHECK9-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[VAR5]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP8]], i8* align 4 [[TMP9]], i64 4, i1 false)
// CHECK9-NEXT:    store i32 2, i32* [[CONV1]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(%struct.S* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN8]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP10]], [[OMP_ARRAYCPY_DONE4]] ], [ [[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]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN8]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE9:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done9:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StC1Ev
// CHECK9-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK9-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2StC2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
// CHECK9-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 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(%struct.S* nonnull align 4 dereferenceable(4) [[THIS1]], %struct.S* nonnull align 4 dereferenceable(4) [[TMP0]], %struct.St* [[T]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StD1Ev
// CHECK9-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK9-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2StD2Ev(%struct.St* nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// 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]]) #[[ATTR4]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = 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 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT:    [[DOTT_VAR__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP1]], i64 4, i8* inttoptr (i64 1 to i8*))
// CHECK9-NEXT:    [[DOTT_VAR__ADDR:%.*]] = bitcast i8* [[DOTT_VAR__VOID_ADDR]] to i32*
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK9-NEXT:    store i32 [[TMP2]], i32* [[DOTT_VAR__ADDR]], align 4
// CHECK9-NEXT:    store i32 0, i32* [[DOTT_VAR__ADDR]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = bitcast i32* [[DOTT_VAR__ADDR]] to i8*
// CHECK9-NEXT:    call void @__kmpc_free(i32 [[TMP1]], i8* [[TMP3]], i8* inttoptr (i64 1 to i8*))
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK9-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK9-NEXT:    [[SST:%.*]] = alloca [[STRUCT_SST:%.*]], align 4
// CHECK9-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK9-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 128
// CHECK9-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK9-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 128
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT:    call void @_ZN3SSTIiEC1Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[SST]])
// CHECK9-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK9-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP0]], i8* align 128 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK9-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]], i32 3)
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]])
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* [[T_VAR]])
// CHECK9-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done1:
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK9-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK9-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[A2:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[C7:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 8
// CHECK9-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[C_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, i32* [[A]], align 8
// CHECK9-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK9-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK9-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK9-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK9-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK9-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
// CHECK9-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32* [[A3]], i32** [[A2]], align 8
// CHECK9-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK9-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK9-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK9-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK9-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK9-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 8
// CHECK9-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 8
// CHECK9-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK9-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK9-NEXT:    [[CONV10:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP5]], i32* [[CONV10]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK9-NEXT:    [[CONV11:%.*]] = bitcast i64* [[C_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP8]], i32* [[CONV11]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i64, i64* [[C_CASTED]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 8
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64, i64, i64, [4 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 8
// CHECK9-NEXT:    [[E5:%.*]] = alloca [4 x i32], align 16
// CHECK9-NEXT:    [[_TMP6:%.*]] = alloca [4 x i32]*, 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 %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[B]], i64* [[B_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[C]], i64* [[C_ADDR]], align 8
// CHECK9-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK9-NEXT:    [[CONV1:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK9-NEXT:    [[CONV2:%.*]] = bitcast i64* [[C_ADDR]] to i32*
// CHECK9-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 8
// CHECK9-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK9-NEXT:    store i32* [[CONV2]], i32** [[_TMP3]], align 8
// CHECK9-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP4]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP4]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E5]] to i8*
// CHECK9-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP3]], i8* align 4 [[TMP4]], i64 16, i1 false)
// CHECK9-NEXT:    store [4 x i32]* [[E5]], [4 x i32]** [[_TMP6]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP5]], align 4
// CHECK9-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK9-NEXT:    store i32 [[INC]], i32* [[TMP5]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK9-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK9-NEXT:    store i32 [[DEC]], i32* [[CONV1]], align 4
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32*, i32** [[_TMP3]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP9]], 1
// CHECK9-NEXT:    store i32 [[DIV]], i32* [[TMP8]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP6]], align 8
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i32], [4 x i32]* [[TMP10]], i64 0, i64 2
// CHECK9-NEXT:    store i32 1111, i32* [[ARRAYIDX]], align 4
// 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 128
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK9-NEXT:    store float [[CONV]], float* [[F]], align 4
// 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 128
// 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 {{[^@]+}}@_ZN2StC2Ev
// CHECK9-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK9-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK9-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    store i32 0, i32* [[B]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
// CHECK9-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 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S* [[S]], %struct.S** [[S_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 %struct.S*, %struct.S** [[S_ADDR]], align 8
// CHECK9-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP0]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP1:%.*]] = load float, float* [[F2]], align 4
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
// CHECK9-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
// CHECK9-NEXT:    store float [[ADD]], float* [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StD2Ev
// CHECK9-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK9-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// 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 {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
// CHECK9-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
// CHECK9-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN3SSTIiEC2Ev(%struct.SST* nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK9-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
// CHECK9-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// CHECK9-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 128
// CHECK9-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK9-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK9-NEXT:    [[VAR5:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128
// CHECK9-NEXT:    [[AGG_TMP6:%.*]] = alloca [[STRUCT_ST]], align 4
// 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 [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK9-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP1]], align 128
// CHECK9-NEXT:    store i32 [[TMP4]], i32* [[T_VAR1]], align 128
// CHECK9-NEXT:    [[TMP5:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
// CHECK9-NEXT:    [[TMP6:%.*]] = bitcast [2 x i32]* [[TMP0]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP5]], i8* align 128 [[TMP6]], i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP7:%.*]] = bitcast [2 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK9-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP8]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK9:       omp.arraycpy.body:
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK9-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]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
// CHECK9:       omp.arraycpy.done4:
// CHECK9-NEXT:    call void @_ZN2StC1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]])
// CHECK9-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]], %struct.S.0* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.St* [[AGG_TMP6]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(%struct.St* nonnull align 4 dereferenceable(8) [[AGG_TMP6]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, i32* [[T_VAR1]], align 128
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i64 0, i64 0
// CHECK9-NEXT:    store i32 [[TMP9]], i32* [[ARRAYIDX]], align 128
// CHECK9-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i64 0, i64 0
// CHECK9-NEXT:    [[TMP10:%.*]] = bitcast %struct.S.0* [[ARRAYIDX7]] to i8*
// CHECK9-NEXT:    [[TMP11:%.*]] = bitcast %struct.S.0* [[VAR5]] to i8*
// CHECK9-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP10]], i8* align 128 [[TMP11]], i64 4, i1 false)
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[VAR5]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN8]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_ARRAYCPY_DONE4]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN8]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE9:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done9:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
// CHECK9-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 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
// CHECK9-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]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// 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 i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 128
// CHECK9-NEXT:    store i32 [[TMP1]], i32* [[T_VAR1]], align 128
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load volatile i32, i32* @g, align 128
// CHECK9-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
// CHECK9-SAME: (%struct.SST* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
// CHECK9-NEXT:    [[A2:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SST:%.*]], %struct.SST* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, i32* [[A]], align 4
// CHECK9-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SST]], %struct.SST* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32* [[A3]], i32** [[A2]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[A2]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK9-NEXT:    store i32 [[TMP1]], i32* [[CONV]], align 4
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK9-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SST*, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), %struct.SST* [[THIS1]], i64 [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SST* [[THIS:%.*]], i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SST*, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32*, 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 %struct.SST* [[THIS]], %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load %struct.SST*, %struct.SST** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK9-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK9-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK9-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* @g, align 128
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
// CHECK9-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
// CHECK9-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 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store %struct.S.0* [[S]], %struct.S.0** [[S_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load %struct.S.0*, %struct.S.0** [[S_ADDR]], align 8
// CHECK9-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP0]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, i32* [[F2]], align 4
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[T]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, i32* [[A]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
// CHECK9-NEXT:    store i32 [[ADD]], i32* [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK9-SAME: (%struct.S.0* nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@main
// CHECK11-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
// CHECK11-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK11-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON]], %class.anon* [[REF_TMP]], i32 0, i32 0
// CHECK11-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP0]], align 8
// CHECK11-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull align 8 dereferenceable(8) [[REF_TMP]])
// CHECK11-NEXT:    ret i32 0
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK11-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK11-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK11-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK11-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK11-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[A2:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[C7:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 8
// CHECK11-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[C_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    store i32 0, i32* [[A]], align 8
// CHECK11-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK11-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK11-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK11-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK11-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK11-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
// CHECK11-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    store i32* [[A3]], i32** [[A2]], align 8
// CHECK11-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK11-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK11-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK11-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK11-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK11-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 8
// CHECK11-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 8
// CHECK11-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK11-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK11-NEXT:    [[CONV10:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP5]], i32* [[CONV10]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 8
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK11-NEXT:    [[CONV11:%.*]] = bitcast i64* [[C_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP8]], i32* [[CONV11]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i64, i64* [[C_CASTED]], align 8
// CHECK11-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 8
// CHECK11-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64, i64, i64, [4 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 8
// CHECK11-NEXT:    [[E5:%.*]] = alloca [4 x i32], align 16
// CHECK11-NEXT:    [[_TMP6:%.*]] = alloca [4 x i32]*, align 8
// CHECK11-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK11-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK11-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK11-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[B]], i64* [[B_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[C]], i64* [[C_ADDR]], align 8
// CHECK11-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK11-NEXT:    [[CONV1:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK11-NEXT:    [[CONV2:%.*]] = bitcast i64* [[C_ADDR]] to i32*
// CHECK11-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 8
// CHECK11-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK11-NEXT:    store i32* [[CONV2]], i32** [[_TMP3]], align 8
// CHECK11-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP4]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP4]], align 8
// CHECK11-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E5]] to i8*
// CHECK11-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK11-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP3]], i8* align 4 [[TMP4]], i64 16, i1 false)
// CHECK11-NEXT:    store [4 x i32]* [[E5]], [4 x i32]** [[_TMP6]], align 8
// CHECK11-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK11-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[TMP5]], align 8
// CHECK11-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK11-NEXT:    store i32* [[TMP7]], i32** [[TMP6]], align 8
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
// CHECK11-NEXT:    store i32* [[CONV1]], i32** [[TMP8]], align 8
// CHECK11-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[_TMP3]], align 8
// CHECK11-NEXT:    store i32* [[TMP10]], i32** [[TMP9]], align 8
// CHECK11-NEXT:    call void @_ZZN2SSC1ERiENKUlvE_clEv(%class.anon.0* nonnull align 8 dereferenceable(32) [[REF_TMP]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv
// CHECK11-SAME: (%class.anon.0* nonnull align 8 dereferenceable(32) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca %class.anon.0*, align 8
// CHECK11-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[C_CASTED:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    store %class.anon.0* [[THIS]], %class.anon.0** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load %class.anon.0*, %class.anon.0** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], %class.anon.0* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP1:%.*]] = load %struct.SS*, %struct.SS** [[TMP0]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK11-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK11-NEXT:    store i32 [[INC]], i32* [[TMP3]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[TMP5]], align 8
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK11-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK11-NEXT:    store i32 [[DEC]], i32* [[TMP6]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32*, i32** [[TMP8]], align 8
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 1
// CHECK11-NEXT:    store i32 [[DIV]], i32* [[TMP9]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32*, i32** [[TMP11]], align 8
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
// CHECK11-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP13]], i32* [[CONV]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP16:%.*]] = load i32*, i32** [[TMP15]], align 8
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, i32* [[TMP16]], align 4
// CHECK11-NEXT:    [[CONV2:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP17]], i32* [[CONV2]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK11-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP20:%.*]] = load i32*, i32** [[TMP19]], align 8
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4
// CHECK11-NEXT:    [[CONV3:%.*]] = bitcast i64* [[C_CASTED]] to i32*
// CHECK11-NEXT:    store i32 [[TMP21]], i32* [[CONV3]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i64, i64* [[C_CASTED]], align 8
// CHECK11-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), %struct.SS* [[TMP1]], i64 [[TMP14]], i64 [[TMP18]], i64 [[TMP22]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK11-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK11-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[B]], i64* [[B_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[C]], i64* [[C_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK11-NEXT:    [[CONV1:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK11-NEXT:    [[CONV2:%.*]] = bitcast i64* [[C_ADDR]] to i32*
// CHECK11-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK11-NEXT:    store i32* [[CONV2]], i32** [[_TMP3]], align 8
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK11-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK11-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK11-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK11-NEXT:    store i32 [[DEC]], i32* [[CONV1]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP3]], align 8
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK11-NEXT:    store i32 [[DIV]], i32* [[TMP4]], align 4
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK11-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[G_ADDR:%.*]] = alloca i32*, align 8
// CHECK11-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[G1:%.*]] = alloca i32, align 128
// CHECK11-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_1:%.*]], align 8
// CHECK11-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK11-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK11-NEXT:    store i32* [[G]], i32** [[G_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 8
// CHECK11-NEXT:    [[CONV:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
// CHECK11-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* [[TMP0]], align 128
// CHECK11-NEXT:    store i32 [[TMP1]], i32* [[G1]], align 128
// CHECK11-NEXT:    store i32 1, i32* [[G1]], align 128
// CHECK11-NEXT:    store i32 2, i32* [[CONV]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 0
// CHECK11-NEXT:    store i32* [[G1]], i32** [[TMP2]], align 8
// CHECK11-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_1]], %class.anon.1* [[REF_TMP]], i32 0, i32 1
// CHECK11-NEXT:    store i32* [[CONV]], i32** [[TMP3]], align 8
// CHECK11-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.1* nonnull align 8 dereferenceable(16) [[REF_TMP]])
// CHECK11-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@main
// CHECK12-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[SS:%.*]] = alloca [[STRUCT_SS:%.*]], align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, align 8
// CHECK12-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK12-NEXT:    call void @_ZN2SSC1ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[SS]], i32* nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store i8* bitcast (void (i8*)* @__main_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 8
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.1 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK12-NEXT:    store i32 [[TMP0]], i32* [[BLOCK_CAPTURED]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]] to void ()*
// CHECK12-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP1]] to %struct.__block_literal_generic*
// CHECK12-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP3:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK12-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[TMP2]], align 8
// CHECK12-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to void (i8*)*
// CHECK12-NEXT:    call void [[TMP5]](i8* [[TMP3]])
// CHECK12-NEXT:    ret i32 0
//
//
// CHECK12-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK12-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK12-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK12-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK12-NEXT:    call void @_ZN2SSC2ERi(%struct.SS* nonnull align 8 dereferenceable(32) [[THIS1]], i32* nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK12-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*, align 8
// CHECK12-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>*
// CHECK12-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32 }>** [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP0]], i32* [[CONV]], align 4
// CHECK12-NEXT:    [[TMP1:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
// CHECK12-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* @g, i64 [[TMP1]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[G:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[G_ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[G1:%.*]] = alloca i32, align 128
// CHECK12-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, align 128
// CHECK12-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store i32* [[G]], i32** [[G_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[G_ADDR]], align 8
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
// CHECK12-NEXT:    [[TMP1:%.*]] = load volatile i32, i32* [[TMP0]], align 128
// CHECK12-NEXT:    store i32 [[TMP1]], i32* [[G1]], align 128
// CHECK12-NEXT:    store i32 1, i32* [[G1]], align 128
// CHECK12-NEXT:    store i32 2, i32* [[CONV]], align 4
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 128
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store i8* bitcast (void (i8*)* @g_block_invoke to i8*), i8** [[BLOCK_INVOKE]], align 16
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP2:%.*]] = load volatile i32, i32* [[G1]], align 128
// CHECK12-NEXT:    store volatile i32 [[TMP2]], i32* [[BLOCK_CAPTURED]], align 128
// CHECK12-NEXT:    [[BLOCK_CAPTURED2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV]], align 4
// CHECK12-NEXT:    store i32 [[TMP3]], i32* [[BLOCK_CAPTURED2]], align 32
// CHECK12-NEXT:    [[TMP4:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]] to void ()*
// CHECK12-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP4]] to %struct.__block_literal_generic*
// CHECK12-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP6:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK12-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP5]], align 8
// CHECK12-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to void (i8*)*
// CHECK12-NEXT:    call void [[TMP8]](i8* [[TMP6]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@g_block_invoke
// CHECK12-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>*, align 8
// CHECK12-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>*
// CHECK12-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>** [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    store i32 2, i32* [[BLOCK_CAPTURE_ADDR]], align 128
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, i32, [92 x i8], i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    store i32 4, i32* [[BLOCK_CAPTURE_ADDR1]], align 32
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK12-SAME: (%struct.SS* nonnull align 8 dereferenceable(32) [[THIS:%.*]], i32* nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK12-NEXT:    [[D_ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[A2:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[C7:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[E:%.*]] = alloca [4 x i32]*, align 8
// CHECK12-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[C_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i32* [[D]], i32** [[D_ADDR]], align 8
// CHECK12-NEXT:    [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK12-NEXT:    store i32 0, i32* [[A]], align 8
// CHECK12-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK12-NEXT:    [[BF_LOAD:%.*]] = load i8, i8* [[B]], align 4
// CHECK12-NEXT:    [[BF_CLEAR:%.*]] = and i8 [[BF_LOAD]], -16
// CHECK12-NEXT:    [[BF_SET:%.*]] = or i8 [[BF_CLEAR]], 0
// CHECK12-NEXT:    store i8 [[BF_SET]], i8* [[B]], align 4
// CHECK12-NEXT:    [[C:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[D_ADDR]], align 8
// CHECK12-NEXT:    store i32* [[TMP0]], i32** [[C]], align 8
// CHECK12-NEXT:    [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0
// CHECK12-NEXT:    store i32* [[A3]], i32** [[A2]], align 8
// CHECK12-NEXT:    [[B5:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 1
// CHECK12-NEXT:    [[BF_LOAD6:%.*]] = load i8, i8* [[B5]], align 4
// CHECK12-NEXT:    [[BF_SHL:%.*]] = shl i8 [[BF_LOAD6]], 4
// CHECK12-NEXT:    [[BF_ASHR:%.*]] = ashr i8 [[BF_SHL]], 4
// CHECK12-NEXT:    [[BF_CAST:%.*]] = sext i8 [[BF_ASHR]] to i32
// CHECK12-NEXT:    store i32 [[BF_CAST]], i32* [[B4]], align 4
// CHECK12-NEXT:    [[C8:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 2
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[C8]], align 8
// CHECK12-NEXT:    store i32* [[TMP1]], i32** [[C7]], align 8
// CHECK12-NEXT:    [[E9:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 3
// CHECK12-NEXT:    store [4 x i32]* [[E9]], [4 x i32]** [[E]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[A2]], align 8
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP3]], i32* [[CONV]], align 4
// CHECK12-NEXT:    [[TMP4:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32, i32* [[B4]], align 4
// CHECK12-NEXT:    [[CONV10:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP5]], i32* [[CONV10]], align 4
// CHECK12-NEXT:    [[TMP6:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK12-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[C7]], align 8
// CHECK12-NEXT:    [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4
// CHECK12-NEXT:    [[CONV11:%.*]] = bitcast i64* [[C_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP8]], i32* [[CONV11]], align 4
// CHECK12-NEXT:    [[TMP9:%.*]] = load i64, i64* [[C_CASTED]], align 8
// CHECK12-NEXT:    [[TMP10:%.*]] = load [4 x i32]*, [4 x i32]** [[E]], align 8
// CHECK12-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64, i64, i64, [4 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), %struct.SS* [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], [4 x i32]* [[TMP10]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], [4 x i32]* nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK12-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[E_ADDR:%.*]] = alloca [4 x i32]*, align 8
// CHECK12-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[_TMP4:%.*]] = alloca [4 x i32]*, align 8
// CHECK12-NEXT:    [[E5:%.*]] = alloca [4 x i32], align 16
// CHECK12-NEXT:    [[_TMP6:%.*]] = alloca [4 x i32]*, align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, align 8
// CHECK12-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[B]], i64* [[B_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[C]], i64* [[C_ADDR]], align 8
// CHECK12-NEXT:    store [4 x i32]* [[E]], [4 x i32]** [[E_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK12-NEXT:    [[CONV1:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK12-NEXT:    [[CONV2:%.*]] = bitcast i64* [[C_ADDR]] to i32*
// CHECK12-NEXT:    [[TMP1:%.*]] = load [4 x i32]*, [4 x i32]** [[E_ADDR]], align 8
// CHECK12-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK12-NEXT:    store i32* [[CONV2]], i32** [[_TMP3]], align 8
// CHECK12-NEXT:    store [4 x i32]* [[TMP1]], [4 x i32]** [[_TMP4]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load [4 x i32]*, [4 x i32]** [[_TMP4]], align 8
// CHECK12-NEXT:    [[TMP3:%.*]] = bitcast [4 x i32]* [[E5]] to i8*
// CHECK12-NEXT:    [[TMP4:%.*]] = bitcast [4 x i32]* [[TMP2]] to i8*
// CHECK12-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP3]], i8* align 4 [[TMP4]], i64 16, i1 false)
// CHECK12-NEXT:    store [4 x i32]* [[E5]], [4 x i32]** [[_TMP6]], align 8
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store i8* bitcast (void (i8*)* @g_block_invoke_2 to i8*), i8** [[BLOCK_INVOKE]], align 8
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.4 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    store %struct.SS* [[TMP0]], %struct.SS** [[BLOCK_CAPTURED_THIS_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK12-NEXT:    store i32* [[TMP5]], i32** [[BLOCK_CAPTURED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED7:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP6:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK12-NEXT:    store i32 [[TMP6]], i32* [[BLOCK_CAPTURED7]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED8:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP7:%.*]] = load i32*, i32** [[_TMP3]], align 8
// CHECK12-NEXT:    store i32* [[TMP7]], i32** [[BLOCK_CAPTURED8]], align 8
// CHECK12-NEXT:    [[TMP8:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]] to void ()*
// CHECK12-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP8]] to %struct.__block_literal_generic*
// CHECK12-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP10:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK12-NEXT:    [[TMP11:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK12-NEXT:    [[TMP12:%.*]] = bitcast i8* [[TMP11]] to void (i8*)*
// CHECK12-NEXT:    call void [[TMP12]](i8* [[TMP10]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@g_block_invoke_2
// CHECK12-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR1]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>*, align 8
// CHECK12-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[B_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[C_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>*
// CHECK12-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>** [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    [[THIS:%.*]] = load %struct.SS*, %struct.SS** [[BLOCK_CAPTURED_THIS]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK12-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK12-NEXT:    store i32 [[INC]], i32* [[TMP0]], align 4
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK12-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP2]], -1
// CHECK12-NEXT:    store i32 [[DEC]], i32* [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR2]], align 8
// CHECK12-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK12-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP4]], 1
// CHECK12-NEXT:    store i32 [[DIV]], i32* [[TMP3]], align 4
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR3]], align 8
// CHECK12-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP5]], align 4
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP6]], i32* [[CONV]], align 4
// CHECK12-NEXT:    [[TMP7:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP8:%.*]] = load i32, i32* [[BLOCK_CAPTURE_ADDR4]], align 8
// CHECK12-NEXT:    [[CONV5:%.*]] = bitcast i64* [[B_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP8]], i32* [[CONV5]], align 4
// CHECK12-NEXT:    [[TMP9:%.*]] = load i64, i64* [[B_CASTED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR6:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, %struct.SS*, i32*, i32*, i32 }>* [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP10:%.*]] = load i32*, i32** [[BLOCK_CAPTURE_ADDR6]], align 8
// CHECK12-NEXT:    [[TMP11:%.*]] = load i32, i32* [[TMP10]], align 4
// CHECK12-NEXT:    [[CONV7:%.*]] = bitcast i64* [[C_CASTED]] to i32*
// CHECK12-NEXT:    store i32 [[TMP11]], i32* [[CONV7]], align 4
// CHECK12-NEXT:    [[TMP12:%.*]] = load i64, i64* [[C_CASTED]], align 8
// CHECK12-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*, i64, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), %struct.SS* [[THIS]], i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP12]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK12-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.SS* [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8
// CHECK12-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[C_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[TMP:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    [[_TMP3:%.*]] = alloca i32*, align 8
// CHECK12-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[B]], i64* [[B_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[C]], i64* [[C_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i32*
// CHECK12-NEXT:    [[CONV1:%.*]] = bitcast i64* [[B_ADDR]] to i32*
// CHECK12-NEXT:    [[CONV2:%.*]] = bitcast i64* [[C_ADDR]] to i32*
// CHECK12-NEXT:    store i32* [[CONV]], i32** [[TMP]], align 8
// CHECK12-NEXT:    store i32* [[CONV2]], i32** [[_TMP3]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[TMP]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK12-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK12-NEXT:    store i32 [[INC]], i32* [[TMP1]], align 4
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4
// CHECK12-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK12-NEXT:    store i32 [[DEC]], i32* [[CONV1]], align 4
// CHECK12-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[_TMP3]], align 8
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK12-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK12-NEXT:    store i32 [[DIV]], i32* [[TMP4]], align 4
// CHECK12-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_Z10array_funcPfP2StiPe
// CHECK17-SAME: (float* [[A:%.*]], %struct.St* [[S:%.*]], i32 [[N:%.*]], x86_fp80* [[VLA1:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT:    [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK17-NEXT:    [[TMP4:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK17-NEXT:    [[TMP6:%.*]] = call i8* @llvm.stacksave()
// CHECK17-NEXT:    store i8* [[TMP6]], i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    [[TMP7:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]]
// CHECK17-NEXT:    [[VLA:%.*]] = alloca double, i64 [[TMP7]], align 128
// CHECK17-NEXT:    store i64 [[TMP3]], i64* [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP5]], i64* [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[TMP8:%.*]] = load %struct.St*, %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP9:%.*]] = load x86_fp80*, x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP10:%.*]] = load float*, float** [[A_ADDR]], align 8
// CHECK17-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 8, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.St*, i32*, i64, x86_fp80*, float*, i64, i64, double*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.St* [[TMP8]], i32* [[N_ADDR]], i64 [[TMP1]], x86_fp80* [[TMP9]], float* [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]])
// CHECK17-NEXT:    [[TMP11:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore(i8* [[TMP11]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.St* [[S:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], float* [[A:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 8
// CHECK17-NEXT:    [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK17-NEXT:    [[VLA_ADDR3:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA_ADDR5:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA2_ADDR:%.*]] = alloca double*, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    store i32* [[N]], i32** [[N_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT:    store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    store float* [[A]], float** [[A_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA2]], i64* [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    store i64 [[VLA4]], i64* [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    store double* [[VLA26]], double** [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8
// CHECK17-NEXT:    [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT:    [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    [[TMP3:%.*]] = load i64, i64* [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    [[TMP4:%.*]] = load double*, double** [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP5:%.*]] = call i8* @llvm.stacksave()
// CHECK17-NEXT:    store i8* [[TMP5]], i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    [[TMP6:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[VLA7:%.*]] = alloca double, i64 [[TMP6]], align 128
// CHECK17-NEXT:    store i64 [[TMP2]], i64* [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP3]], i64* [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[TMP7:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[TMP8:%.*]] = mul nuw i64 [[TMP7]], 8
// CHECK17-NEXT:    [[TMP9:%.*]] = bitcast double* [[VLA7]] to i8*
// CHECK17-NEXT:    [[TMP10:%.*]] = bitcast double* [[TMP4]] to i8*
// CHECK17-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP9]], i8* align 128 [[TMP10]], i64 [[TMP8]], i1 false)
// CHECK17-NEXT:    [[TMP11:%.*]] = load %struct.St*, %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[TMP11]], i64 0
// CHECK17-NEXT:    [[TMP12:%.*]] = load %struct.St*, %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP13:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK17-NEXT:    [[TMP14:%.*]] = load x86_fp80*, x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    call void @_ZN2St7St_funcEPS_iPe(%struct.St* nonnull align 4 dereferenceable(8) [[ARRAYIDX]], %struct.St* [[TMP12]], i32 [[TMP13]], x86_fp80* [[TMP14]])
// CHECK17-NEXT:    [[TMP15:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore(i8* [[TMP15]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_ZN2St7St_funcEPS_iPe
// CHECK17-SAME: (%struct.St* nonnull align 4 dereferenceable(8) [[THIS:%.*]], %struct.St* [[S:%.*]], i32 [[N:%.*]], x86_fp80* [[VLA1:%.*]]) #[[ATTR0]] align 2 {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[THIS1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT:    [[TMP2:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK17-NEXT:    [[TMP4:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK17-NEXT:    [[TMP6:%.*]] = call i8* @llvm.stacksave()
// CHECK17-NEXT:    store i8* [[TMP6]], i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    [[TMP7:%.*]] = mul nuw i64 [[TMP3]], [[TMP5]]
// CHECK17-NEXT:    [[VLA:%.*]] = alloca double, i64 [[TMP7]], align 128
// CHECK17-NEXT:    store i64 [[TMP3]], i64* [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP5]], i64* [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[THIS1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP8:%.*]] = load i32, i32* [[B]], align 4
// CHECK17-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[THIS1]], i32 0, i32 0
// CHECK17-NEXT:    store i32 [[TMP8]], i32* [[A]], align 4
// CHECK17-NEXT:    [[TMP9:%.*]] = load x86_fp80*, x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP10:%.*]] = load %struct.St*, %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 8, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, x86_fp80*, %struct.St*, i64, i64, double*, i32*, %struct.St*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP1]], x86_fp80* [[TMP9]], %struct.St* [[THIS1]], i64 [[TMP3]], i64 [[TMP5]], double* [[VLA]], i32* [[N_ADDR]], %struct.St* [[TMP10]])
// CHECK17-NEXT:    [[TMP11:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore(i8* [[TMP11]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK17-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], x86_fp80* [[VLA1:%.*]], %struct.St* [[THIS:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], double* nonnull align 8 dereferenceable(8) [[VLA26:%.*]], i32* nonnull align 4 dereferenceable(4) [[N:%.*]], %struct.St* [[S:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca x86_fp80*, align 8
// CHECK17-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    [[VLA_ADDR3:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA_ADDR5:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA2_ADDR:%.*]] = alloca double*, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32*, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK17-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT:    store x86_fp80* [[VLA1]], x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    store %struct.St* [[THIS]], %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA2]], i64* [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    store i64 [[VLA4]], i64* [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    store double* [[VLA26]], double** [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    store i32* [[N]], i32** [[N_ADDR]], align 8
// CHECK17-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK17-NEXT:    [[TMP1:%.*]] = load %struct.St*, %struct.St** [[THIS_ADDR]], align 8
// CHECK17-NEXT:    [[TMP2:%.*]] = load i64, i64* [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    [[TMP3:%.*]] = load i64, i64* [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    [[TMP4:%.*]] = load double*, double** [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP5:%.*]] = load i32*, i32** [[N_ADDR]], align 8
// CHECK17-NEXT:    [[TMP6:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[TMP7:%.*]] = mul nuw i64 [[TMP6]], 8
// CHECK17-NEXT:    [[TMP8:%.*]] = add nuw i64 [[TMP7]], 127
// CHECK17-NEXT:    [[TMP9:%.*]] = udiv i64 [[TMP8]], 128
// CHECK17-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], 128
// CHECK17-NEXT:    [[TMP11:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK17-NEXT:    [[DOTVLA2__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP12]], i64 [[TMP10]], i8* inttoptr (i64 8 to i8*))
// CHECK17-NEXT:    [[DOTVLA2__ADDR:%.*]] = bitcast i8* [[DOTVLA2__VOID_ADDR]] to double*
// CHECK17-NEXT:    [[TMP13:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8
// CHECK17-NEXT:    [[TMP15:%.*]] = bitcast double* [[DOTVLA2__ADDR]] to i8*
// CHECK17-NEXT:    [[TMP16:%.*]] = bitcast double* [[TMP4]] to i8*
// CHECK17-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 128 [[TMP15]], i8* align 128 [[TMP16]], i64 [[TMP14]], i1 false)
// CHECK17-NEXT:    [[B:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], %struct.St* [[TMP1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP17:%.*]] = load i32, i32* [[B]], align 4
// CHECK17-NEXT:    [[A:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[TMP1]], i32 0, i32 0
// CHECK17-NEXT:    store i32 [[TMP17]], i32* [[A]], align 4
// CHECK17-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP17]] to double
// CHECK17-NEXT:    [[TMP18:%.*]] = mul nsw i64 1, [[TMP3]]
// CHECK17-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds double, double* [[DOTVLA2__ADDR]], i64 [[TMP18]]
// CHECK17-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP5]], align 4
// CHECK17-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP19]], 1
// CHECK17-NEXT:    [[IDXPROM:%.*]] = sext i32 [[SUB]] to i64
// CHECK17-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds double, double* [[ARRAYIDX]], i64 [[IDXPROM]]
// CHECK17-NEXT:    store double [[CONV]], double* [[ARRAYIDX7]], align 8
// CHECK17-NEXT:    [[CONV8:%.*]] = fpext double [[CONV]] to x86_fp80
// CHECK17-NEXT:    [[TMP20:%.*]] = load x86_fp80*, x86_fp80** [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[B9:%.*]] = getelementptr inbounds [[STRUCT_ST]], %struct.St* [[TMP1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP21:%.*]] = load i32, i32* [[B9]], align 4
// CHECK17-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP21]] to i64
// CHECK17-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds x86_fp80, x86_fp80* [[TMP20]], i64 [[IDXPROM10]]
// CHECK17-NEXT:    store x86_fp80 [[CONV8]], x86_fp80* [[ARRAYIDX11]], align 16
// CHECK17-NEXT:    [[TMP22:%.*]] = bitcast double* [[DOTVLA2__ADDR]] to i8*
// CHECK17-NEXT:    call void @__kmpc_free(i32 [[TMP12]], i8* [[TMP22]], i8* inttoptr (i64 8 to i8*))
// CHECK17-NEXT:    ret void
//