llvm/clang/test/OpenMP/parallel_firstprivate_codegen.cpp

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -x c++ -std=c++11 -triple i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-enable-noundef-analysis -fopenmp -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -fopenmp -x c++ -fblocks -DBLOCKS -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK4

// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -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-enable-noundef-analysis -fopenmp-simd -x c++ -triple i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -fopenmp -x c++ -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -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-enable-noundef-analysis -fopenmp -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK12

// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -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-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-enable-noundef-analysis -fopenmp-simd -x c++ -triple x86_64-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -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-enable-noundef-analysis -verify -Wno-vla -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, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 4 dereferenceable(28) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const.main.vec, i32 8, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[S_ARR]], float 1.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[S_ARR]], i32 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00)
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK1-NEXT:    store i32 [[TMP2]], ptr [[SIVAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[SIVAR_CASTED]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 5, ptr @main.omp_outlined, ptr [[VEC]], i32 [[TMP1]], ptr [[S_ARR]], ptr [[VAR]], i32 [[TMP3]])
// CHECK1-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    store i32 [[TMP4]], ptr [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 1, ptr @main.omp_outlined.1, i32 [[TMP5]])
// CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK1-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 4 dereferenceable(28) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ef(ptr nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], ptr nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr nonnull align 4 dereferenceable(4) [[VAR:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, 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 ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[SIVAR]], ptr [[SIVAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VEC_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR_ADDR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC1]], ptr align 4 [[TMP0]], i32 8, i1 false)
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP3]]
// 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 ptr [ [[TMP1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], ptr [[AGG_TMP]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP3]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
// CHECK1-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[VAR4]], ptr nonnull align 4 dereferenceable(4) [[TMP2]], ptr [[AGG_TMP5]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 [[TMP4]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[ARRAYIDX6]], ptr align 4 [[VAR4]], i32 4, i1 false)
// CHECK1-NEXT:    store i32 2, ptr [[SIVAR_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN7]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[OMP_ARRAYCPY_DONE3]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2StC1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN2StC2Ev(ptr nonnull align 4 dereferenceable(8) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]], ptr [[T]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2StD1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN2StD2Ev(ptr nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfED2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.1
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i32 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[DOTT_VAR__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP1]], i32 4, ptr inttoptr (i32 1 to ptr))
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[TMP2]], ptr [[DOTT_VAR__VOID_ADDR]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTT_VAR__VOID_ADDR]], align 4
// CHECK1-NEXT:    call void @__kmpc_free(i32 [[TMP1]], ptr [[DOTT_VAR__VOID_ADDR]], ptr inttoptr (i32 1 to ptr))
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR1]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[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:    [[T_VAR_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR_CASTED1:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    call void @_ZN3SSTIiEC1Ev(ptr nonnull align 4 dereferenceable(4) [[SST]])
// CHECK1-NEXT:    store i32 0, ptr [[T_VAR]], align 128
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 128 [[VEC]], ptr align 128 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[S_ARR]], i32 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[S_ARR]], i32 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[VAR]], i32 3)
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 128
// CHECK1-NEXT:    store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[T_VAR_CASTED]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @_Z5tmainIiET_v.omp_outlined, ptr [[VEC]], i32 [[TMP1]], ptr [[S_ARR]], ptr [[VAR]])
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[T_VAR]], align 128
// CHECK1-NEXT:    store i32 [[TMP2]], ptr [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[T_VAR_CASTED1]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 1, ptr @_Z5tmainIiET_v.omp_outlined.2, i32 [[TMP3]])
// CHECK1-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP4]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP5]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A2:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[C7:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK1-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK1-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK1-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[C]], align 4
// CHECK1-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A3]], ptr [[A2]], align 4
// CHECK1-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK1-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK1-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 4
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 4
// CHECK1-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK1-NEXT:    store ptr [[E9]], ptr [[E]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK1-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[B_CASTED]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[C_CASTED]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], ptr [[TMP10]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK1-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[C]], ptr [[C_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK1-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 4
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[E3]], ptr align 4 [[TMP2]], i32 16, i1 false)
// CHECK1-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK1-NEXT:    store i32 [[INC]], ptr [[TMP3]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK1-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP5]], -1
// CHECK1-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[_TMP1]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK1-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP7]], 1
// CHECK1-NEXT:    store i32 [[DIV]], ptr [[TMP6]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[_TMP4]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i32], ptr [[TMP8]], i32 0, i32 2
// CHECK1-NEXT:    store i32 1111, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK1-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 128
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
// CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
// CHECK1-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2StC2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK1-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_ST]], ptr [[THIS1]], i32 0, i32 1
// CHECK1-NEXT:    store i32 0, ptr [[B]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = load float, ptr [[F2]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[T]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
// CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
// CHECK1-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2StD2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN3SSTIiEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(ptr nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v.omp_outlined
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32 [[T_VAR:%.*]], ptr nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 128
// CHECK1-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK1-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128
// CHECK1-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VEC_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR_ADDR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 128 [[VEC1]], ptr align 128 [[TMP0]], i32 8, i1 false)
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i32 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP3]]
// 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 ptr [ [[TMP1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], ptr [[AGG_TMP]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP3]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
// CHECK1-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[VAR4]], ptr nonnull align 4 dereferenceable(4) [[TMP2]], ptr [[AGG_TMP5]])
// CHECK1-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 [[TMP4]], ptr [[ARRAYIDX]], align 128
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 128 [[ARRAYIDX6]], ptr align 128 [[VAR4]], i32 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN7]], i32 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[OMP_ARRAYCPY_DONE3]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]], ptr [[T]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiED2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v.omp_outlined.2
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i32 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store i32 [[T_VAR]], ptr [[T_VAR_ADDR]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK1-NEXT:    store i32 [[TMP0]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A2:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SST:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK1-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SST]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A3]], ptr [[A2]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A2]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    store i32 [[TMP1]], ptr [[A_CASTED]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @_ZN3SSTIiEC2Ev.omp_outlined, ptr [[THIS1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev.omp_outlined
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK1-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 128
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 4
// CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[F2]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[T]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK1-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[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, ptr [[RETVAL]], align 4
// CHECK3-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 4 dereferenceable(28) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[CLASS_ANON]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT:    store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4
// CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(ptr nonnull align 4 dereferenceable(4) [[REF_TMP]])
// CHECK3-NEXT:    ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK3-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 4 dereferenceable(28) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK3-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[A2:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[C7:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK3-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK3-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[TMP0]], ptr [[C]], align 4
// CHECK3-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[A3]], ptr [[A2]], align 4
// CHECK3-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK3-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 4
// CHECK3-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 4
// CHECK3-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    store ptr [[E9]], ptr [[E]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK3-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK3-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, ptr [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], ptr [[TMP10]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK3-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[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 ptr, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK3-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[C]], ptr [[C_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK3-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 4
// CHECK3-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 4
// CHECK3-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[E3]], ptr align 4 [[TMP2]], i32 16, i1 false)
// CHECK3-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[TMP0]], ptr [[TMP3]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK3-NEXT:    store ptr [[TMP5]], ptr [[TMP4]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 2
// CHECK3-NEXT:    store ptr [[B_ADDR]], ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 4
// CHECK3-NEXT:    store ptr [[TMP8]], ptr [[TMP7]], align 4
// CHECK3-NEXT:    call void @_ZZN2SSC1ERiENKUlvE_clEv(ptr nonnull align 4 dereferenceable(16) [[REF_TMP]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv
// CHECK3-SAME: (ptr nonnull align 4 dereferenceable(16) [[THIS:%.*]]) #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK3-NEXT:    store i32 [[INC]], ptr [[TMP3]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK3-NEXT:    store i32 [[DEC]], ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 1
// CHECK3-NEXT:    store i32 [[DIV]], ptr [[TMP9]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK3-NEXT:    store i32 [[TMP13]], ptr [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP14:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK3-NEXT:    store i32 [[TMP17]], ptr [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP18:%.*]] = load i32, ptr [[B_CASTED]], align 4
// CHECK3-NEXT:    [[TMP19:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 3
// CHECK3-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[TMP19]], align 4
// CHECK3-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK3-NEXT:    store i32 [[TMP21]], ptr [[C_CASTED]], align 4
// CHECK3-NEXT:    [[TMP22:%.*]] = load i32, ptr [[C_CASTED]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @_ZZN2SSC1ERiENKUlvE_clEv.omp_outlined, ptr [[TMP1]], i32 [[TMP14]], i32 [[TMP18]], i32 [[TMP22]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv.omp_outlined
// CHECK3-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[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 ptr, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK3-NEXT:    store i32 [[C]], ptr [[C_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK3-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK3-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK3-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK3-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK3-NEXT:    store i32 [[DIV]], ptr [[TMP4]], align 4
// 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 <{ ptr, i32, i32, ptr, ptr, i32 }>, align 4
// CHECK4-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK4-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 4 dereferenceable(28) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 4
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store ptr @__main_block_invoke, ptr [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store ptr @__block_descriptor_tmp.1, ptr [[BLOCK_DESCRIPTOR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK4-NEXT:    store i32 [[TMP0]], ptr [[BLOCK_CAPTURED]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 4
// CHECK4-NEXT:    call void [[TMP2]](ptr [[BLOCK]])
// CHECK4-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK4-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK4-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 4 dereferenceable(28) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK4-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[G_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK4-NEXT:    store i32 [[TMP0]], ptr [[G_CASTED]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, ptr [[G_CASTED]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK4-NEXT:    store i32 [[TMP2]], ptr [[SIVAR_CASTED]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, ptr [[SIVAR_CASTED]], align 4
// CHECK4-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 2, ptr @__main_block_invoke.omp_outlined, i32 [[TMP1]], i32 [[TMP3]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke.omp_outlined
// CHECK4-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i32 [[G:%.*]], i32 [[SIVAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[G_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, align 128
// CHECK4-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK4-NEXT:    store i32 [[G]], ptr [[G_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[SIVAR]], ptr [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    store i32 1, ptr [[G_ADDR]], align 4
// CHECK4-NEXT:    store i32 2, ptr [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 128
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 8
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store ptr @g_block_invoke, ptr [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store ptr @__block_descriptor_tmp, ptr [[BLOCK_DESCRIPTOR]], align 16
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr [[G_ADDR]], align 4
// CHECK4-NEXT:    store volatile i32 [[TMP0]], ptr [[BLOCK_CAPTURED]], align 128
// CHECK4-NEXT:    [[BLOCK_CAPTURED1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, ptr [[SIVAR_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[TMP1]], ptr [[BLOCK_CAPTURED1]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 4
// CHECK4-NEXT:    call void [[TMP3]](ptr [[BLOCK]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g_block_invoke
// CHECK4-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK4-NEXT:    store i32 2, ptr [[BLOCK_CAPTURE_ADDR]], align 128
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [104 x i8], i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5
// CHECK4-NEXT:    store i32 4, ptr [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK4-SAME: (ptr nonnull align 4 dereferenceable(28) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[A2:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[C7:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK4-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK4-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK4-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK4-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[TMP0]], ptr [[C]], align 4
// CHECK4-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK4-NEXT:    store ptr [[A3]], ptr [[A2]], align 4
// CHECK4-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK4-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK4-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 4
// CHECK4-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 4
// CHECK4-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK4-NEXT:    store ptr [[E9]], ptr [[E]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK4-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK4-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, ptr [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 4
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK4-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, ptr [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 4
// CHECK4-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP9]], ptr [[TMP10]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK4-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 4
// CHECK4-NEXT:    [[TMP:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 4
// CHECK4-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, align 4
// CHECK4-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK4-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[C]], ptr [[C_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK4-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 4
// CHECK4-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 4
// CHECK4-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[E3]], ptr align 4 [[TMP2]], i32 16, i1 false)
// CHECK4-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 4
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 4
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 4
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store ptr @g_block_invoke_2, ptr [[BLOCK_INVOKE]], align 4
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store ptr @__block_descriptor_tmp.2, ptr [[BLOCK_DESCRIPTOR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    store ptr [[TMP0]], ptr [[BLOCK_CAPTURED_THIS_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK4-NEXT:    store ptr [[TMP3]], ptr [[BLOCK_CAPTURED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED5:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[TMP4]], ptr [[BLOCK_CAPTURED5]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED6:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[BLOCK]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 4
// CHECK4-NEXT:    store ptr [[TMP5]], ptr [[BLOCK_CAPTURED6]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK4-NEXT:    call void [[TMP7]](ptr [[BLOCK]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g_block_invoke_2
// CHECK4-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, 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 ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5
// CHECK4-NEXT:    [[THIS:%.*]] = load ptr, ptr [[BLOCK_CAPTURED_THIS]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK4-NEXT:    store i32 [[INC]], ptr [[TMP0]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, ptr [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP2]], -1
// CHECK4-NEXT:    store i32 [[DEC]], ptr [[BLOCK_CAPTURE_ADDR1]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR2]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP4]], 1
// CHECK4-NEXT:    store i32 [[DIV]], ptr [[TMP3]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR3]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK4-NEXT:    store i32 [[TMP6]], ptr [[A_CASTED]], align 4
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, ptr [[BLOCK_CAPTURE_ADDR4]], align 4
// CHECK4-NEXT:    store i32 [[TMP8]], ptr [[B_CASTED]], align 4
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, ptr [[B_CASTED]], align 4
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR5:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, i32, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR5]], align 4
// CHECK4-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK4-NEXT:    store i32 [[TMP11]], ptr [[C_CASTED]], align 4
// CHECK4-NEXT:    [[TMP12:%.*]] = load i32, ptr [[C_CASTED]], align 4
// CHECK4-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @g_block_invoke_2.omp_outlined, ptr [[THIS]], i32 [[TMP7]], i32 [[TMP9]], i32 [[TMP12]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g_block_invoke_2.omp_outlined
// CHECK4-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i32 [[A:%.*]], i32 [[B:%.*]], i32 [[C:%.*]]) #[[ATTR3]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 4
// CHECK4-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 4
// CHECK4-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK4-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK4-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK4-NEXT:    store i32 [[C]], ptr [[C_ADDR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK4-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 4
// CHECK4-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 4
// CHECK4-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK4-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK4-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK4-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK4-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK4-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK4-NEXT:    store i32 [[DIV]], ptr [[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_CASTED1:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 8 dereferenceable(32) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT:    store i32 0, ptr [[T_VAR]], align 4
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const.main.vec, i64 8, i1 false)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[S_ARR]], float 1.000000e+00)
// CHECK9-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[S_ARR]], i64 1
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK9-NEXT:    call void @_ZN1SIfEC1Ef(ptr nonnull align 4 dereferenceable(4) [[VAR]], float 3.000000e+00)
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK9-NEXT:    store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[T_VAR_CASTED]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK9-NEXT:    store i32 [[TMP2]], ptr [[SIVAR_CASTED]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load i64, ptr [[SIVAR_CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 5, ptr @main.omp_outlined, ptr [[VEC]], i64 [[TMP1]], ptr [[S_ARR]], ptr [[VAR]], i64 [[TMP3]])
// CHECK9-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[T_VAR_CASTED1]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i64, ptr [[T_VAR_CASTED1]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 1, ptr @main.omp_outlined.1, i64 [[TMP5]])
// CHECK9-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK9-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done2:
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK9-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 8 dereferenceable(32) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfEC2Ef(ptr nonnull align 4 dereferenceable(4) [[THIS1]], float [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main.omp_outlined
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], ptr nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr nonnull align 4 dereferenceable(4) [[VAR:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S], align 4
// CHECK9-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK9-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK9-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[SIVAR]], ptr [[SIVAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC1]], ptr align 4 [[TMP0]], i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP3]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK9:       omp.arraycpy.body:
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK9-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], ptr [[AGG_TMP]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP3]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK9:       omp.arraycpy.done3:
// CHECK9-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
// CHECK9-NEXT:    call void @_ZN1SIfEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[VAR4]], ptr nonnull align 4 dereferenceable(4) [[TMP2]], ptr [[AGG_TMP5]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC1]], i64 0, i64 0
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i64 0, i64 0
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX6]], ptr align 4 [[VAR4]], i64 4, i1 false)
// CHECK9-NEXT:    store i32 2, ptr [[SIVAR_ADDR]], align 4
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[S_ARR2]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN7]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[OMP_ARRAYCPY_DONE3]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIfED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done8:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StC1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2StC2Ev(ptr nonnull align 4 dereferenceable(8) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1ERKS0_2St
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfEC2ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]], ptr [[T]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StD1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN2StD2Ev(ptr nonnull align 4 dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIfED2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main.omp_outlined.1
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i64 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    [[DOTT_VAR__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP1]], i64 4, ptr inttoptr (i64 1 to ptr))
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP2]], ptr [[DOTT_VAR__VOID_ADDR]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[DOTT_VAR__VOID_ADDR]], align 4
// CHECK9-NEXT:    call void @__kmpc_free(i32 [[TMP1]], ptr [[DOTT_VAR__VOID_ADDR]], ptr inttoptr (i64 1 to ptr))
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK9-SAME: () #[[ATTR1]] 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:    [[T_VAR_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[T_VAR_CASTED1:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT:    call void @_ZN3SSTIiEC1Ev(ptr nonnull align 4 dereferenceable(4) [[SST]])
// CHECK9-NEXT:    store i32 0, ptr [[T_VAR]], align 128
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[VEC]], ptr align 128 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[S_ARR]], i32 1)
// CHECK9-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[S_ARR]], i64 1
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK9-NEXT:    call void @_ZN1SIiEC1Ei(ptr nonnull align 4 dereferenceable(4) [[VAR]], i32 3)
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 128
// CHECK9-NEXT:    store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[T_VAR_CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @_Z5tmainIiET_v.omp_outlined, ptr [[VEC]], i64 [[TMP1]], ptr [[S_ARR]], ptr [[VAR]])
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[T_VAR]], align 128
// CHECK9-NEXT:    store i32 [[TMP2]], ptr [[T_VAR_CASTED1]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = load i64, ptr [[T_VAR_CASTED1]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 1, ptr @_Z5tmainIiET_v.omp_outlined.2, i64 [[TMP3]])
// CHECK9-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP4]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done2:
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP5]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK9-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A2:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[C7:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, ptr [[A]], align 8
// CHECK9-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK9-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[TMP0]], ptr [[C]], align 8
// CHECK9-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A3]], ptr [[A2]], align 8
// CHECK9-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK9-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 8
// CHECK9-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 8
// CHECK9-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[E9]], ptr [[E]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK9-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i64, ptr [[B_CASTED]], align 8
// CHECK9-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK9-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i64, ptr [[C_CASTED]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], ptr [[TMP10]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 16
// CHECK9-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[C]], ptr [[C_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK9-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 8
// CHECK9-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 16 [[E3]], ptr align 4 [[TMP2]], i64 16, i1 false)
// CHECK9-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK9-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK9-NEXT:    store i32 [[INC]], ptr [[TMP3]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK9-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP5]], -1
// CHECK9-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP7]], 1
// CHECK9-NEXT:    store i32 [[DIV]], ptr [[TMP6]], align 4
// CHECK9-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x i32], ptr [[TMP8]], i64 0, i64 2
// CHECK9-NEXT:    store i32 1111, ptr [[ARRAYIDX]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to float
// CHECK9-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 128
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP1]] to float
// CHECK9-NEXT:    [[ADD:%.*]] = fadd float [[TMP0]], [[CONV]]
// CHECK9-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StC2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK9-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_ST]], ptr [[THIS1]], i32 0, i32 1
// CHECK9-NEXT:    store i32 0, ptr [[B]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2ERKS0_2St
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    [[F2:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP0]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP1:%.*]] = load float, ptr [[F2]], align 4
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[T]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A]], align 4
// CHECK9-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP2]] to float
// CHECK9-NEXT:    [[ADD:%.*]] = fadd float [[TMP1]], [[CONV]]
// CHECK9-NEXT:    store float [[ADD]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN2StD2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIiEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN3SSTIiEC1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN3SSTIiEC2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    call void @_ZN1SIiEC2Ei(ptr nonnull align 4 dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v.omp_outlined
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(8) [[VEC:%.*]], i64 [[T_VAR:%.*]], ptr nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[VEC1:%.*]] = alloca [2 x i32], align 128
// CHECK9-NEXT:    [[S_ARR2:%.*]] = alloca [2 x %struct.S.0], align 128
// CHECK9-NEXT:    [[AGG_TMP:%.*]] = alloca [[STRUCT_ST:%.*]], align 4
// CHECK9-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0:%.*]], align 128
// CHECK9-NEXT:    [[AGG_TMP5:%.*]] = alloca [[STRUCT_ST]], align 4
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[VEC1]], ptr align 128 [[TMP0]], i64 8, i1 false)
// CHECK9-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP3]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK9:       omp.arraycpy.body:
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK9-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]])
// CHECK9-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]], ptr [[AGG_TMP]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP]]) #[[ATTR4]]
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK9-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP3]]
// CHECK9-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK9:       omp.arraycpy.done3:
// CHECK9-NEXT:    call void @_ZN2StC1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]])
// CHECK9-NEXT:    call void @_ZN1SIiEC1ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[VAR4]], ptr nonnull align 4 dereferenceable(4) [[TMP2]], ptr [[AGG_TMP5]])
// CHECK9-NEXT:    call void @_ZN2StD1Ev(ptr nonnull align 4 dereferenceable(8) [[AGG_TMP5]]) #[[ATTR4]]
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[T_VAR_ADDR]], align 4
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC1]], i64 0, i64 0
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[ARRAYIDX]], align 128
// CHECK9-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i64 0, i64 0
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[ARRAYIDX6]], ptr align 128 [[VAR4]], i64 4, i1 false)
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR2]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN7]], i64 2
// CHECK9-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9:       arraydestroy.body:
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[OMP_ARRAYCPY_DONE3]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT:    call void @_ZN1SIiED1Ev(ptr nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK9-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9:       arraydestroy.done8:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1ERKS0_2St
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIiEC2ERKS0_2St(ptr nonnull align 4 dereferenceable(4) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]], ptr [[T]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    call void @_ZN1SIiED2Ev(ptr nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v.omp_outlined.2
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i64 [[T_VAR:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK9-NEXT:    store i32 [[TMP0]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A2:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SST:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store i32 0, ptr [[A]], align 4
// CHECK9-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SST]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A3]], ptr [[A2]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A2]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT:    store i32 [[TMP1]], ptr [[A_CASTED]], align 4
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 2, ptr @_ZN3SSTIiEC2Ev.omp_outlined, ptr [[THIS1]], i64 [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN3SSTIiEC2Ev.omp_outlined
// CHECK9-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK9-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK9-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load volatile i32, ptr @g, align 128
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP0]], [[TMP1]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2ERKS0_2St
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[S:%.*]], ptr [[T:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[T_INDIRECT_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    store ptr [[T]], ptr [[T_INDIRECT_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK9-NEXT:    [[F2:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0]], ptr [[TMP0]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP1:%.*]] = load i32, ptr [[F2]], align 4
// CHECK9-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[T]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[A]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[F]], align 4
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK9-SAME: (ptr nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK9-NEXT:    ret void
//
//
// 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, ptr [[RETVAL]], align 4
// CHECK11-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 8 dereferenceable(32) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[CLASS_ANON]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK11-NEXT:    store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8
// CHECK11-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(ptr nonnull align 8 dereferenceable(8) [[REF_TMP]])
// CHECK11-NEXT:    ret i32 0
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK11-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK11-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 8 dereferenceable(32) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK11-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[A2:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[C7:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    store i32 0, ptr [[A]], align 8
// CHECK11-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK11-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[TMP0]], ptr [[C]], align 8
// CHECK11-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A3]], ptr [[A2]], align 8
// CHECK11-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK11-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 8
// CHECK11-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 8
// CHECK11-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[E9]], ptr [[E]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 8
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i64, ptr [[B_CASTED]], align 8
// CHECK11-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 8
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK11-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i64, ptr [[C_CASTED]], align 8
// CHECK11-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 8
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], ptr [[TMP10]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK11-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 16
// CHECK11-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK11-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[C]], ptr [[C_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK11-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 8
// CHECK11-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK11-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 16 [[E3]], ptr align 4 [[TMP2]], i64 16, i1 false)
// CHECK11-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 8
// CHECK11-NEXT:    [[TMP3:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[TMP0]], ptr [[TMP3]], align 8
// CHECK11-NEXT:    [[TMP4:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK11-NEXT:    store ptr [[TMP5]], ptr [[TMP4]], align 8
// CHECK11-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[B_ADDR]], ptr [[TMP6]], align 8
// CHECK11-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK11-NEXT:    store ptr [[TMP8]], ptr [[TMP7]], align 8
// CHECK11-NEXT:    call void @_ZZN2SSC1ERiENKUlvE_clEv(ptr nonnull align 8 dereferenceable(32) [[REF_TMP]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv
// CHECK11-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]]) #[[ATTR1]] align 2 {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK11-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK11-NEXT:    store i32 [[INC]], ptr [[TMP3]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK11-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP7]], -1
// CHECK11-NEXT:    store i32 [[DEC]], ptr [[TMP6]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 1
// CHECK11-NEXT:    store i32 [[DIV]], ptr [[TMP9]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 1
// CHECK11-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 8
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK11-NEXT:    store i32 [[TMP13]], ptr [[A_CASTED]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 2
// CHECK11-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 8
// CHECK11-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK11-NEXT:    store i32 [[TMP17]], ptr [[B_CASTED]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = load i64, ptr [[B_CASTED]], align 8
// CHECK11-NEXT:    [[TMP19:%.*]] = getelementptr inbounds nuw [[CLASS_ANON_0]], ptr [[THIS1]], i32 0, i32 3
// CHECK11-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[TMP19]], align 8
// CHECK11-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK11-NEXT:    store i32 [[TMP21]], ptr [[C_CASTED]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = load i64, ptr [[C_CASTED]], align 8
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @_ZZN2SSC1ERiENKUlvE_clEv.omp_outlined, ptr [[TMP1]], i64 [[TMP14]], i64 [[TMP18]], i64 [[TMP22]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZZN2SSC1ERiENKUlvE_clEv.omp_outlined
// CHECK11-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 8
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK11-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK11-NEXT:    store i64 [[C]], ptr [[C_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK11-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK11-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 8
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK11-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK11-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK11-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK11-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK11-NEXT:    store i32 [[DIV]], ptr [[TMP4]], align 4
// 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 <{ ptr, i32, i32, ptr, ptr, i32 }>, align 8
// CHECK12-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK12-NEXT:    call void @_ZN2SSC1ERi(ptr nonnull align 8 dereferenceable(32) [[SS]], ptr nonnull align 4 dereferenceable(4) @_ZZ4mainE5sivar)
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 8
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store ptr @__main_block_invoke, ptr [[BLOCK_INVOKE]], align 8
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store ptr @__block_descriptor_tmp.1, ptr [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    [[TMP0:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK12-NEXT:    store i32 [[TMP0]], ptr [[BLOCK_CAPTURED]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 8
// CHECK12-NEXT:    call void [[TMP2]](ptr [[BLOCK]])
// CHECK12-NEXT:    ret i32 0
//
//
// CHECK12-LABEL: define {{[^@]+}}@_ZN2SSC1ERi
// CHECK12-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK12-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK12-NEXT:    call void @_ZN2SSC2ERi(ptr nonnull align 8 dereferenceable(32) [[THIS1]], ptr nonnull align 4 dereferenceable(4) [[TMP0]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK12-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[G_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[SIVAR_CASTED:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr @g, align 128
// CHECK12-NEXT:    store i32 [[TMP0]], ptr [[G_CASTED]], align 4
// CHECK12-NEXT:    [[TMP1:%.*]] = load i64, ptr [[G_CASTED]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK12-NEXT:    store i32 [[TMP2]], ptr [[SIVAR_CASTED]], align 4
// CHECK12-NEXT:    [[TMP3:%.*]] = load i64, ptr [[SIVAR_CASTED]], align 8
// CHECK12-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 2, ptr @__main_block_invoke.omp_outlined, i64 [[TMP1]], i64 [[TMP3]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@__main_block_invoke.omp_outlined
// CHECK12-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i64 [[G:%.*]], i64 [[SIVAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[G_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = alloca <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, align 128
// CHECK12-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store i64 [[G]], ptr [[G_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[SIVAR]], ptr [[SIVAR_ADDR]], align 8
// CHECK12-NEXT:    store i32 1, ptr [[G_ADDR]], align 4
// CHECK12-NEXT:    store i32 2, ptr [[SIVAR_ADDR]], align 4
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 128
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store ptr @g_block_invoke, ptr [[BLOCK_INVOKE]], align 16
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store ptr @__block_descriptor_tmp, ptr [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP0:%.*]] = load volatile i32, ptr [[G_ADDR]], align 4
// CHECK12-NEXT:    store volatile i32 [[TMP0]], ptr [[BLOCK_CAPTURED]], align 128
// CHECK12-NEXT:    [[BLOCK_CAPTURED1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32, ptr [[SIVAR_ADDR]], align 4
// CHECK12-NEXT:    store i32 [[TMP1]], ptr [[BLOCK_CAPTURED1]], align 32
// CHECK12-NEXT:    [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP2]], align 8
// CHECK12-NEXT:    call void [[TMP3]](ptr [[BLOCK]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@g_block_invoke
// CHECK12-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK12-NEXT:    store i32 2, ptr [[BLOCK_CAPTURE_ADDR]], align 128
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, i32, [92 x i8], i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5
// CHECK12-NEXT:    store i32 4, ptr [[BLOCK_CAPTURE_ADDR1]], align 32
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@_ZN2SSC2ERi
// CHECK12-SAME: (ptr nonnull align 8 dereferenceable(32) [[THIS:%.*]], ptr nonnull align 4 dereferenceable(4) [[D:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[D_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[A2:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[B4:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[C7:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[E:%.*]] = alloca ptr, 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 ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK12-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK12-NEXT:    store i32 0, ptr [[A]], align 8
// CHECK12-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK12-NEXT:    [[BF_LOAD:%.*]] = load i8, ptr [[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]], ptr [[B]], align 4
// CHECK12-NEXT:    [[C:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK12-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[TMP0]], ptr [[C]], align 8
// CHECK12-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK12-NEXT:    store ptr [[A3]], ptr [[A2]], align 8
// CHECK12-NEXT:    [[B5:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1
// CHECK12-NEXT:    [[BF_LOAD6:%.*]] = load i8, ptr [[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]], ptr [[B4]], align 4
// CHECK12-NEXT:    [[C8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 2
// CHECK12-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[C8]], align 8
// CHECK12-NEXT:    store ptr [[TMP1]], ptr [[C7]], align 8
// CHECK12-NEXT:    [[E9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 3
// CHECK12-NEXT:    store ptr [[E9]], ptr [[E]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A2]], align 8
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK12-NEXT:    store i32 [[TMP3]], ptr [[A_CASTED]], align 4
// CHECK12-NEXT:    [[TMP4:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32, ptr [[B4]], align 4
// CHECK12-NEXT:    store i32 [[TMP5]], ptr [[B_CASTED]], align 4
// CHECK12-NEXT:    [[TMP6:%.*]] = load i64, ptr [[B_CASTED]], align 8
// CHECK12-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[C7]], align 8
// CHECK12-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK12-NEXT:    store i32 [[TMP8]], ptr [[C_CASTED]], align 4
// CHECK12-NEXT:    [[TMP9:%.*]] = load i64, ptr [[C_CASTED]], align 8
// CHECK12-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[E]], align 8
// CHECK12-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 5, ptr @_ZN2SSC2ERi.omp_outlined, ptr [[THIS1]], i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP9]], ptr [[TMP10]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@_ZN2SSC2ERi.omp_outlined
// CHECK12-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]], ptr nonnull align 4 dereferenceable(16) [[E:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 8
// CHECK12-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[E3:%.*]] = alloca [4 x i32], align 16
// CHECK12-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[BLOCK:%.*]] = alloca <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, align 8
// CHECK12-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[C]], ptr [[C_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[E]], ptr [[E_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[E_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK12-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 8
// CHECK12-NEXT:    store ptr [[TMP1]], ptr [[_TMP2]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK12-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 16 [[E3]], ptr align 4 [[TMP2]], i64 16, i1 false)
// CHECK12-NEXT:    store ptr [[E3]], ptr [[_TMP4]], align 8
// CHECK12-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK12-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 8
// CHECK12-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK12-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 8
// CHECK12-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK12-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK12-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    store ptr @g_block_invoke_2, ptr [[BLOCK_INVOKE]], align 8
// CHECK12-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK12-NEXT:    store ptr @__block_descriptor_tmp.2, ptr [[BLOCK_DESCRIPTOR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED_THIS_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK12-NEXT:    store ptr [[TMP0]], ptr [[BLOCK_CAPTURED_THIS_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK12-NEXT:    store ptr [[TMP3]], ptr [[BLOCK_CAPTURED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED5:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP4:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK12-NEXT:    store i32 [[TMP4]], ptr [[BLOCK_CAPTURED5]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED6:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[BLOCK]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK12-NEXT:    store ptr [[TMP5]], ptr [[BLOCK_CAPTURED6]], align 8
// CHECK12-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK12-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK12-NEXT:    call void [[TMP7]](ptr [[BLOCK]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@g_block_invoke_2
// CHECK12-SAME: (ptr [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, 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 ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURED_THIS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5
// CHECK12-NEXT:    [[THIS:%.*]] = load ptr, ptr [[BLOCK_CAPTURED_THIS]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK12-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK12-NEXT:    store i32 [[INC]], ptr [[TMP0]], align 4
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, ptr [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK12-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP2]], -1
// CHECK12-NEXT:    store i32 [[DEC]], ptr [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR2:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR2]], align 8
// CHECK12-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK12-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP4]], 1
// CHECK12-NEXT:    store i32 [[DIV]], ptr [[TMP3]], align 4
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR3:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6
// CHECK12-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR3]], align 8
// CHECK12-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK12-NEXT:    store i32 [[TMP6]], ptr [[A_CASTED]], align 4
// CHECK12-NEXT:    [[TMP7:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR4:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 8
// CHECK12-NEXT:    [[TMP8:%.*]] = load i32, ptr [[BLOCK_CAPTURE_ADDR4]], align 8
// CHECK12-NEXT:    store i32 [[TMP8]], ptr [[B_CASTED]], align 4
// CHECK12-NEXT:    [[TMP9:%.*]] = load i64, ptr [[B_CASTED]], align 8
// CHECK12-NEXT:    [[BLOCK_CAPTURE_ADDR5:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, ptr, ptr, ptr, i32 }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 7
// CHECK12-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR5]], align 8
// CHECK12-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK12-NEXT:    store i32 [[TMP11]], ptr [[C_CASTED]], align 4
// CHECK12-NEXT:    [[TMP12:%.*]] = load i64, ptr [[C_CASTED]], align 8
// CHECK12-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @g_block_invoke_2.omp_outlined, ptr [[THIS]], i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP12]])
// CHECK12-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@g_block_invoke_2.omp_outlined
// CHECK12-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[THIS:%.*]], i64 [[A:%.*]], i64 [[B:%.*]], i64 [[C:%.*]]) #[[ATTR3]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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 ptr, align 8
// CHECK12-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK12-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK12-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK12-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK12-NEXT:    store i64 [[C]], ptr [[C_ADDR]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK12-NEXT:    store ptr [[A_ADDR]], ptr [[TMP]], align 8
// CHECK12-NEXT:    store ptr [[C_ADDR]], ptr [[_TMP1]], align 8
// CHECK12-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK12-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK12-NEXT:    store i32 [[INC]], ptr [[TMP1]], align 4
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK12-NEXT:    [[DEC:%.*]] = add nsw i32 [[TMP3]], -1
// CHECK12-NEXT:    store i32 [[DEC]], ptr [[B_ADDR]], align 4
// CHECK12-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK12-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP5]], 1
// CHECK12-NEXT:    store i32 [[DIV]], ptr [[TMP4]], align 4
// CHECK12-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_Z10array_funcPfP2StiPe
// CHECK17-SAME: (ptr [[A:%.*]], ptr [[S:%.*]], i32 [[N:%.*]], ptr [[VLA1:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    store ptr [[VLA1]], ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT:    [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK17-NEXT:    [[TMP4:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK17-NEXT:    [[TMP6:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK17-NEXT:    store ptr [[TMP6]], ptr [[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]], ptr [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP5]], ptr [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK17-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1:[0-9]+]], i32 8, ptr @_Z10array_funcPfP2StiPe.omp_outlined, ptr [[TMP8]], ptr [[N_ADDR]], i64 [[TMP1]], ptr [[TMP9]], ptr [[TMP10]], i64 [[TMP3]], i64 [[TMP5]], ptr [[VLA]])
// CHECK17-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP11]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_Z10array_funcPfP2StiPe.omp_outlined
// CHECK17-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr [[S:%.*]], ptr nonnull align 4 dereferenceable(4) [[N:%.*]], i64 [[VLA:%.*]], ptr [[VLA1:%.*]], ptr [[A:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], ptr nonnull align 8 dereferenceable(8) [[VLA26:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[VLA_ADDR3:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA_ADDR5:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA2_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[VLA1]], ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA2]], ptr [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    store i64 [[VLA4]], ptr [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    store ptr [[VLA26]], ptr [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8
// CHECK17-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK17-NEXT:    [[TMP2:%.*]] = load i64, ptr [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    [[TMP3:%.*]] = load i64, ptr [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP5:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK17-NEXT:    store ptr [[TMP5]], ptr [[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]], ptr [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[TMP7:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[TMP8:%.*]] = mul nuw i64 [[TMP7]], 8
// CHECK17-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[VLA7]], ptr align 128 [[TMP4]], i64 [[TMP8]], i1 false)
// CHECK17-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [[STRUCT_ST:%.*]], ptr [[TMP9]], i64 0
// CHECK17-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK17-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    call void @_ZN2St7St_funcEPS_iPe(ptr nonnull align 4 dereferenceable(8) [[ARRAYIDX]], ptr [[TMP10]], i32 [[TMP11]], ptr [[TMP12]])
// CHECK17-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP13]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_ZN2St7St_funcEPS_iPe
// CHECK17-SAME: (ptr nonnull align 4 dereferenceable(8) [[THIS:%.*]], ptr [[S:%.*]], i32 [[N:%.*]], ptr [[VLA1:%.*]]) #[[ATTR0]] align 2 {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    store ptr [[VLA1]], ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK17-NEXT:    [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK17-NEXT:    [[TMP4:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK17-NEXT:    [[TMP5:%.*]] = zext i32 [[TMP4]] to i64
// CHECK17-NEXT:    [[TMP6:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK17-NEXT:    store ptr [[TMP6]], ptr [[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]], ptr [[__VLA_EXPR0]], align 8
// CHECK17-NEXT:    store i64 [[TMP5]], ptr [[__VLA_EXPR1]], align 8
// CHECK17-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[THIS1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP8:%.*]] = load i32, ptr [[B]], align 4
// CHECK17-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST]], ptr [[THIS1]], i32 0, i32 0
// CHECK17-NEXT:    store i32 [[TMP8]], ptr [[A]], align 4
// CHECK17-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 8, ptr @_ZN2St7St_funcEPS_iPe.omp_outlined, i64 [[TMP1]], ptr [[TMP9]], ptr [[THIS1]], i64 [[TMP3]], i64 [[TMP5]], ptr [[VLA]], ptr [[N_ADDR]], ptr [[TMP10]])
// CHECK17-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK17-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP11]])
// CHECK17-NEXT:    ret void
//
//
// CHECK17-LABEL: define {{[^@]+}}@_ZN2St7St_funcEPS_iPe.omp_outlined
// CHECK17-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], ptr [[VLA1:%.*]], ptr [[THIS:%.*]], i64 [[VLA2:%.*]], i64 [[VLA4:%.*]], ptr nonnull align 8 dereferenceable(8) [[VLA26:%.*]], ptr nonnull align 4 dereferenceable(4) [[N:%.*]], ptr [[S:%.*]]) #[[ATTR2]] {
// CHECK17-NEXT:  entry:
// CHECK17-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA1_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[VLA_ADDR3:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA_ADDR5:%.*]] = alloca i64, align 8
// CHECK17-NEXT:    [[VLA2_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[N_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    [[S_ADDR:%.*]] = alloca ptr, align 8
// CHECK17-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[VLA1]], ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK17-NEXT:    store i64 [[VLA2]], ptr [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    store i64 [[VLA4]], ptr [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    store ptr [[VLA26]], ptr [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[N]], ptr [[N_ADDR]], align 8
// CHECK17-NEXT:    store ptr [[S]], ptr [[S_ADDR]], align 8
// CHECK17-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK17-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK17-NEXT:    [[TMP2:%.*]] = load i64, ptr [[VLA_ADDR3]], align 8
// CHECK17-NEXT:    [[TMP3:%.*]] = load i64, ptr [[VLA_ADDR5]], align 8
// CHECK17-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[VLA2_ADDR]], align 8
// CHECK17-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[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 ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK17-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK17-NEXT:    [[DOTVLA2__VOID_ADDR:%.*]] = call ptr @__kmpc_alloc(i32 [[TMP12]], i64 [[TMP10]], ptr inttoptr (i64 8 to ptr))
// CHECK17-NEXT:    [[TMP13:%.*]] = mul nuw i64 [[TMP2]], [[TMP3]]
// CHECK17-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], 8
// CHECK17-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 128 [[DOTVLA2__VOID_ADDR]], ptr align 128 [[TMP4]], i64 [[TMP14]], i1 false)
// CHECK17-NEXT:    [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_ST:%.*]], ptr [[TMP1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP15:%.*]] = load i32, ptr [[B]], align 4
// CHECK17-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_ST]], ptr [[TMP1]], i32 0, i32 0
// CHECK17-NEXT:    store i32 [[TMP15]], ptr [[A]], align 4
// CHECK17-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP15]] to double
// CHECK17-NEXT:    [[TMP16:%.*]] = mul nsw i64 1, [[TMP3]]
// CHECK17-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[DOTVLA2__VOID_ADDR]], i64 [[TMP16]]
// CHECK17-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK17-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP17]], 1
// CHECK17-NEXT:    [[IDXPROM:%.*]] = sext i32 [[SUB]] to i64
// CHECK17-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds double, ptr [[ARRAYIDX]], i64 [[IDXPROM]]
// CHECK17-NEXT:    store double [[CONV]], ptr [[ARRAYIDX7]], align 8
// CHECK17-NEXT:    [[CONV8:%.*]] = fpext double [[CONV]] to x86_fp80
// CHECK17-NEXT:    [[TMP18:%.*]] = load ptr, ptr [[VLA1_ADDR]], align 8
// CHECK17-NEXT:    [[B9:%.*]] = getelementptr inbounds nuw [[STRUCT_ST]], ptr [[TMP1]], i32 0, i32 1
// CHECK17-NEXT:    [[TMP19:%.*]] = load i32, ptr [[B9]], align 4
// CHECK17-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP19]] to i64
// CHECK17-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds x86_fp80, ptr [[TMP18]], i64 [[IDXPROM10]]
// CHECK17-NEXT:    store x86_fp80 [[CONV8]], ptr [[ARRAYIDX11]], align 16
// CHECK17-NEXT:    call void @__kmpc_free(i32 [[TMP12]], ptr [[DOTVLA2__VOID_ADDR]], ptr inttoptr (i64 8 to ptr))
// CHECK17-NEXT:    ret void
//