// 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 -verify -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -verify -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK4
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -verify -fopenmp -x c++ -std=c++11 -DLOOP -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK6

// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK8
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK10
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DARRAY -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -std=c++11 -DLOOP -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK12
// expected-no-diagnostics

#if !defined(ARRAY) && !defined(LOOP)
#ifndef HEADER
#define HEADER

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

volatile double g;

template <typename T>
T tmain() {
  S<T> ttt;
  S<T> test;
  T t_var __attribute__((aligned(128))) = T();
  T vec[] = {1, 2};
  S<T> s_arr[] = {1, 2};
  S<T> var(3);
#pragma omp parallel master taskloop simd lastprivate(t_var, vec, s_arr, s_arr, var, var)
  for (int i = 0; i < 10; ++i) {
    vec[0] = t_var;
    s_arr[0] = var;
  }
  return T();
}

int main() {
  static int sivar;
#ifdef LAMBDA
  [&]() {

#pragma omp parallel master taskloop simd lastprivate(g, sivar)
  for (int i = 0; i < 10; ++i) {

    g = 1;
    sivar = 11;
    [&]() {
      g = 2;
      sivar = 22;
    }();
  }
  }();
  return 0;
#elif defined(BLOCKS)
  ^{
#pragma omp parallel master taskloop simd lastprivate(g, sivar)
  for (int i = 0; i < 10; ++i) {

    g = 1;
    sivar = 11;
    ^{
      g = 2;
      sivar = 22;
    }();
  }
  }();
  return 0;
#else
  S<double> ttt;
  S<double> test;
  int t_var = 0;
  int vec[] = {1, 2};
  S<double> s_arr[] = {1, 2};
  S<double> var(3);
#pragma omp parallel master taskloop simd lastprivate(var, t_var, s_arr, vec, s_arr, var, sivar)
  for (int i = 0; i < 10; ++i) {
    vec[0] = t_var;
    s_arr[0] = var;
    sivar = 33;
  }
  return tmain<int>();
#endif
}


// Store original variables in capture struct.

// Allocate task.
// Returns struct kmp_task_t {
//         [[KMP_TASK_T]] task_data;
//         [[KMP_TASK_MAIN_TY]] privates;
//       };

// Fill kmp_task_t->shareds by copying from original capture argument.

// Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).
// Also copy address of private copy to the corresponding shareds reference.

// Constructors for s_arr and var.
// s_arr;

// var;

// t_var;
// vec;
// sivar;

// Provide pointer to destructor function, which will destroy private variables at the end of the task.

// Start task.


// Privates actually are used.


// Store original variables in capture struct.

// Allocate task.
// Returns struct kmp_task_t {
//         [[KMP_TASK_T_TY]] task_data;
//         [[KMP_TASK_TMAIN_TY]] privates;
//       };

// Fill kmp_task_t->shareds by copying from original capture argument.

// Initialize kmp_task_t->privates with default values (no init for simple types, default constructors for classes).

// t_var;
// vec;

// Constructors for s_arr and var.
// a_arr;

// var;

// Provide pointer to destructor function, which will destroy private variables at the end of the task.

// Start task.

// No destructors must be called for private copies of s_arr and var.


// Privates actually are used.


#endif
#elif defined(ARRAY)
struct St {
  int a, b;
  St() : a(0), b(0) {}
  St(const St &) {}
  ~St() {}
};

void array_func(int n, float a[n], St s[2]) {
#pragma omp parallel master taskloop simd lastprivate(a, s)
  for (int i = 0; i < 10; ++i)
    ;
}
#else

void loop() {
  int i, j;
#pragma omp parallel master taskloop simd linear(i, j)
  for (i = 0; i < 10; ++i)
    ++j;
}
#endif

// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S]], align 8
// 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 16
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]])
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]])
// CHECK1-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK1-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK1-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_BEGIN]], double 1.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_ELEMENT]], double 2.000000e+00)
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[VAR]], double 3.000000e+00)
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]])
// CHECK1-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK1-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done1:
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ev
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIdEC2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ed
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIdEC2Ed(%struct.S* nonnull dereferenceable(8) [[THIS1]], double [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 8 dereferenceable(16) [[S_ARR:%.*]], %struct.S* nonnull align 8 dereferenceable(8) [[VAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    [[TMP7:%.*]] = icmp ne i32 [[TMP6]], 0
// CHECK1-NEXT:    br i1 [[TMP7]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK1:       omp_if.then:
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT:    store [2 x i32]* [[TMP0]], [2 x i32]** [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT:    store [2 x %struct.S]* [[TMP2]], [2 x %struct.S]** [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 3
// CHECK1-NEXT:    store %struct.S* [[TMP3]], %struct.S** [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 4
// CHECK1-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP12]], align 8
// CHECK1-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    [[TMP13:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 9, i64 120, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.kmp_task_t_with_privates*
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP14]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP17]], i8* align 8 [[TMP18]], i64 40, i1 false)
// CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP14]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP19]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP20]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[OMP_IF_THEN]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP19]], i32 0, i32 1
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TMP21]])
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP23:%.*]] = bitcast %union.kmp_cmplrdata_t* [[TMP22]] to i32 (i32, i8*)**
// CHECK1-NEXT:    store i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_destructor. to i32 (i32, i8*)*), i32 (i32, i8*)** [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 5
// CHECK1-NEXT:    store i64 0, i64* [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 6
// CHECK1-NEXT:    store i64 9, i64* [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 7
// CHECK1-NEXT:    store i64 1, i64* [[TMP26]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 9
// CHECK1-NEXT:    [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i8*
// CHECK1-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP28]], i8 0, i64 8, i1 false)
// CHECK1-NEXT:    [[TMP29:%.*]] = load i64, i64* [[TMP26]], align 8
// CHECK1-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i8* [[TMP13]], i32 1, i64* [[TMP24]], i64* [[TMP25]], i64 [[TMP29]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates*, i32)* @.omp_task_dup. to i8*))
// CHECK1-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    br label [[OMP_IF_END]]
// CHECK1:       omp_if.end:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK1-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], %struct.S** noalias [[TMP1:%.*]], i32** noalias [[TMP2:%.*]], [2 x %struct.S]** noalias [[TMP3:%.*]], [2 x i32]** noalias [[TMP4:%.*]], i32** noalias [[TMP5:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S**, align 8
// CHECK1-NEXT:    [[DOTADDR2:%.*]] = alloca i32**, align 8
// CHECK1-NEXT:    [[DOTADDR3:%.*]] = alloca [2 x %struct.S]**, align 8
// CHECK1-NEXT:    [[DOTADDR4:%.*]] = alloca [2 x i32]**, align 8
// CHECK1-NEXT:    [[DOTADDR5:%.*]] = alloca i32**, align 8
// CHECK1-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S** [[TMP1]], %struct.S*** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store i32** [[TMP2]], i32*** [[DOTADDR2]], align 8
// CHECK1-NEXT:    store [2 x %struct.S]** [[TMP3]], [2 x %struct.S]*** [[DOTADDR3]], align 8
// CHECK1-NEXT:    store [2 x i32]** [[TMP4]], [2 x i32]*** [[DOTADDR4]], align 8
// CHECK1-NEXT:    store i32** [[TMP5]], i32*** [[DOTADDR5]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP8:%.*]] = load [2 x %struct.S]**, [2 x %struct.S]*** [[DOTADDR3]], align 8
// CHECK1-NEXT:    store [2 x %struct.S]* [[TMP7]], [2 x %struct.S]** [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP10:%.*]] = load %struct.S**, %struct.S*** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP9]], %struct.S** [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32**, i32*** [[DOTADDR2]], align 8
// CHECK1-NEXT:    store i32* [[TMP11]], i32** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP14:%.*]] = load [2 x i32]**, [2 x i32]*** [[DOTADDR4]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[TMP13]], [2 x i32]** [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32**, i32*** [[DOTADDR5]], align 8
// CHECK1-NEXT:    store i32* [[TMP15]], i32** [[TMP16]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR7:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK1-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR2_I:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR3_I:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR4_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK1-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK1-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK1-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK1-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK1-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, %struct.S**, i32**, [2 x %struct.S]**, [2 x i32]**, i32**)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK1-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP22:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, %struct.S**, i32**, [2 x %struct.S]**, [2 x i32]**, i32**)*
// CHECK1-NEXT:    call void [[TMP25]](i8* [[TMP24]], %struct.S** [[DOTLASTPRIV_PTR_ADDR_I]], i32** [[DOTLASTPRIV_PTR_ADDR1_I]], [2 x %struct.S]** [[DOTLASTPRIV_PTR_ADDR2_I]], [2 x i32]** [[DOTLASTPRIV_PTR_ADDR3_I]], i32** [[DOTLASTPRIV_PTR_ADDR4_I]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP22]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP27:%.*]] = load %struct.S*, %struct.S** [[TMP26]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32*, i32** [[TMP28]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP31:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP33:%.*]] = load [2 x i32]*, [2 x i32]** [[TMP32]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP35:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[TMP34]], align 8
// CHECK1-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP37:%.*]] = load %struct.S*, %struct.S** [[TMP36]], align 8
// CHECK1-NEXT:    [[TMP38:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 4
// CHECK1-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[TMP38]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = load %struct.S*, %struct.S** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP41:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP42:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[DOTLASTPRIV_PTR_ADDR2_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP43:%.*]] = load [2 x i32]*, [2 x i32]** [[DOTLASTPRIV_PTR_ADDR3_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP44:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR4_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[TMP45:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK1-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP45]] to i32
// CHECK1-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK1:       omp.inner.for.cond.i:
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    [[CONV5_I:%.*]] = sext i32 [[TMP46]] to i64
// CHECK1-NEXT:    [[TMP47:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV5_I]], [[TMP47]]
// CHECK1-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK1:       omp.inner.for.body.i:
// CHECK1-NEXT:    [[TMP48:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    store i32 [[TMP48]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32, i32* [[TMP41]], align 4, !llvm.access.group !14
// CHECK1-NEXT:    [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP43]], i64 0, i64 0
// CHECK1-NEXT:    store i32 [[TMP49]], i32* [[ARRAYIDX_I]], align 4, !llvm.access.group !14
// CHECK1-NEXT:    [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP42]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP50:%.*]] = bitcast %struct.S* [[ARRAYIDX6_I]] to i8*
// CHECK1-NEXT:    [[TMP51:%.*]] = bitcast %struct.S* [[TMP40]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP50]], i8* align 8 [[TMP51]], i64 8, i1 false) #[[ATTR4]], !llvm.access.group !14
// CHECK1-NEXT:    store i32 33, i32* [[TMP44]], align 4, !llvm.access.group !14
// CHECK1-NEXT:    [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    [[ADD7_I:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK1-NEXT:    store i32 [[ADD7_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK1:       omp.inner.for.end.i:
// CHECK1-NEXT:    [[TMP53:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK1-NEXT:    [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK1-NEXT:    br i1 [[TMP54]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK1:       .omp.lastprivate.then.i:
// CHECK1-NEXT:    [[TMP55:%.*]] = bitcast %struct.S* [[TMP27]] to i8*
// CHECK1-NEXT:    [[TMP56:%.*]] = bitcast %struct.S* [[TMP40]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP55]], i8* align 8 [[TMP56]], i64 8, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP57:%.*]] = load i32, i32* [[TMP41]], align 4
// CHECK1-NEXT:    store i32 [[TMP57]], i32* [[TMP29]], align 4
// CHECK1-NEXT:    [[ARRAY_BEGIN_I:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP31]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP58:%.*]] = bitcast [2 x %struct.S]* [[TMP42]] to %struct.S*
// CHECK1-NEXT:    [[TMP59:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN_I]], i64 2
// CHECK1-NEXT:    br label [[OMP_ARRAYCPY_BODY_I:%.*]]
// CHECK1:       omp.arraycpy.body.i:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST_I:%.*]] = phi %struct.S* [ [[TMP58]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST_I:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN_I]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK1-NEXT:    [[TMP60:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]] to i8*
// CHECK1-NEXT:    [[TMP61:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP60]], i8* align 8 [[TMP61]], i64 8, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT_I]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT_I]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE_I:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT_I]], [[TMP59]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE_I]], label [[OMP_ARRAYCPY_DONE8_I:%.*]], label [[OMP_ARRAYCPY_BODY_I]]
// CHECK1:       omp.arraycpy.done8.i:
// CHECK1-NEXT:    [[TMP62:%.*]] = bitcast [2 x i32]* [[TMP33]] to i8*
// CHECK1-NEXT:    [[TMP63:%.*]] = bitcast [2 x i32]* [[TMP43]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP62]], i8* align 4 [[TMP63]], i64 8, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP64:%.*]] = load i32, i32* [[TMP44]], align 4
// CHECK1-NEXT:    store i32 [[TMP64]], i32* [[TMP39]], align 4
// CHECK1-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK1:       .omp_outlined..1.exit:
// CHECK1-NEXT:    ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_dup.
// CHECK1-SAME: (%struct.kmp_task_t_with_privates* [[TMP0:%.*]], %struct.kmp_task_t_with_privates* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR7]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK1-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP0]], %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK1-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP7]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP8]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP7]], i32 0, i32 1
// CHECK1-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TMP9]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_destructor.
// CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP2]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 1
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TMP5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdED1Ev
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIdED2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]])
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK1-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK1-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[VAR]], i32 3)
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]])
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done1:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ev
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store double 0.000000e+00, double* [[F]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ed
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK1-NEXT:    store double [[TMP0]], double* [[F]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIdED2Ev
// CHECK1-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_1:%.*]], align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    [[TMP7:%.*]] = icmp ne i32 [[TMP6]], 0
// CHECK1-NEXT:    br i1 [[TMP7]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK1:       omp_if.then:
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT:    store [2 x i32]* [[TMP0]], [2 x i32]** [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT:    store [2 x %struct.S.0]* [[TMP2]], [2 x %struct.S.0]** [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 3
// CHECK1-NEXT:    store %struct.S.0* [[TMP3]], %struct.S.0** [[TMP11]], align 8
// CHECK1-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    [[TMP12:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 9, i64 256, i64 32, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates.2*)* @.omp_task_entry..5 to i32 (i32, i8*)*))
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast i8* [[TMP12]] to %struct.kmp_task_t_with_privates.2*
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP13]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 128
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast %struct.anon.1* [[AGG_CAPTURED]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP16]], i8* align 8 [[TMP17]], i64 32, i1 false)
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP13]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP18]], i32 0, i32 2
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP19]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[OMP_IF_THEN]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP18]], i32 0, i32 3
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP20]])
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP22:%.*]] = bitcast %union.kmp_cmplrdata_t* [[TMP21]] to i32 (i32, i8*)**
// CHECK1-NEXT:    store i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates.2*)* @.omp_task_destructor..7 to i32 (i32, i8*)*), i32 (i32, i8*)** [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 5
// CHECK1-NEXT:    store i64 0, i64* [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 6
// CHECK1-NEXT:    store i64 9, i64* [[TMP24]], align 16
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 7
// CHECK1-NEXT:    store i64 1, i64* [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 9
// CHECK1-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i8*
// CHECK1-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP27]], i8 0, i64 8, i1 false)
// CHECK1-NEXT:    [[TMP28:%.*]] = load i64, i64* [[TMP25]], align 8
// CHECK1-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i8* [[TMP12]], i32 1, i64* [[TMP23]], i64* [[TMP24]], i64 [[TMP28]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2*, i32)* @.omp_task_dup..6 to i8*))
// CHECK1-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK1-NEXT:    br label [[OMP_IF_END]]
// CHECK1:       omp_if.end:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map..4
// CHECK1-SAME: (%struct..kmp_privates.t.3* noalias [[TMP0:%.*]], i32** noalias [[TMP1:%.*]], [2 x i32]** noalias [[TMP2:%.*]], [2 x %struct.S.0]** noalias [[TMP3:%.*]], %struct.S.0** noalias [[TMP4:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t.3*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32**, align 8
// CHECK1-NEXT:    [[DOTADDR2:%.*]] = alloca [2 x i32]**, align 8
// CHECK1-NEXT:    [[DOTADDR3:%.*]] = alloca [2 x %struct.S.0]**, align 8
// CHECK1-NEXT:    [[DOTADDR4:%.*]] = alloca %struct.S.0**, align 8
// CHECK1-NEXT:    store %struct..kmp_privates.t.3* [[TMP0]], %struct..kmp_privates.t.3** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32** [[TMP1]], i32*** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store [2 x i32]** [[TMP2]], [2 x i32]*** [[DOTADDR2]], align 8
// CHECK1-NEXT:    store [2 x %struct.S.0]** [[TMP3]], [2 x %struct.S.0]*** [[DOTADDR3]], align 8
// CHECK1-NEXT:    store %struct.S.0** [[TMP4]], %struct.S.0*** [[DOTADDR4]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load %struct..kmp_privates.t.3*, %struct..kmp_privates.t.3** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32**, i32*** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store i32* [[TMP6]], i32** [[TMP7]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load [2 x i32]**, [2 x i32]*** [[DOTADDR2]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[TMP8]], [2 x i32]** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP11:%.*]] = load [2 x %struct.S.0]**, [2 x %struct.S.0]*** [[DOTADDR3]], align 8
// CHECK1-NEXT:    store [2 x %struct.S.0]* [[TMP10]], [2 x %struct.S.0]** [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP13:%.*]] = load %struct.S.0**, %struct.S.0*** [[DOTADDR4]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP12]], %struct.S.0** [[TMP13]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry..5
// CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK1-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.1*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR2_I:%.*]] = alloca [2 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[DOTLASTPRIV_PTR_ADDR3_I:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 128
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.1*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t.3* [[TMP9]] to i8*
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates.2* [[TMP3]] to i8*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK1-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK1-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 16
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK1-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 64
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK1-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META20:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META23:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META25:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META27:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META29:![0-9]+]])
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !31
// CHECK1-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t.3*, i32**, [2 x i32]**, [2 x %struct.S.0]**, %struct.S.0**)* @.omp_task_privates_map..4 to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !31
// CHECK1-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    store %struct.anon.1* [[TMP8]], %struct.anon.1** [[__CONTEXT_ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP22:%.*]] = load %struct.anon.1*, %struct.anon.1** [[__CONTEXT_ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, i32**, [2 x i32]**, [2 x %struct.S.0]**, %struct.S.0**)*
// CHECK1-NEXT:    call void [[TMP25]](i8* [[TMP24]], i32** [[DOTLASTPRIV_PTR_ADDR_I]], [2 x i32]** [[DOTLASTPRIV_PTR_ADDR1_I]], [2 x %struct.S.0]** [[DOTLASTPRIV_PTR_ADDR2_I]], %struct.S.0** [[DOTLASTPRIV_PTR_ADDR3_I]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_1:%.*]], %struct.anon.1* [[TMP22]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP29:%.*]] = load [2 x i32]*, [2 x i32]** [[TMP28]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP31:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP33:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[TMP32]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP35:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP34]], align 8
// CHECK1-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 3
// CHECK1-NEXT:    [[TMP37:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP36]], align 8
// CHECK1-NEXT:    [[TMP38:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP39:%.*]] = load [2 x i32]*, [2 x i32]** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP40:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[DOTLASTPRIV_PTR_ADDR2_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP41:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTLASTPRIV_PTR_ADDR3_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[TMP42:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !31
// CHECK1-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP42]] to i32
// CHECK1-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !31
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK1:       omp.inner.for.cond.i:
// CHECK1-NEXT:    [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    [[CONV4_I:%.*]] = sext i32 [[TMP43]] to i64
// CHECK1-NEXT:    [[TMP44:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV4_I]], [[TMP44]]
// CHECK1-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK1:       omp.inner.for.body.i:
// CHECK1-NEXT:    [[TMP45:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    store i32 [[TMP45]], i32* [[I_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32, i32* [[TMP38]], align 128, !llvm.access.group !32
// CHECK1-NEXT:    [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP39]], i64 0, i64 0
// CHECK1-NEXT:    store i32 [[TMP46]], i32* [[ARRAYIDX_I]], align 4, !llvm.access.group !32
// CHECK1-NEXT:    [[ARRAYIDX5_I:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP40]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP47:%.*]] = bitcast %struct.S.0* [[ARRAYIDX5_I]] to i8*
// CHECK1-NEXT:    [[TMP48:%.*]] = bitcast %struct.S.0* [[TMP41]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP47]], i8* align 4 [[TMP48]], i64 4, i1 false) #[[ATTR4]], !llvm.access.group !32
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    [[ADD6_I:%.*]] = add nsw i32 [[TMP49]], 1
// CHECK1-NEXT:    store i32 [[ADD6_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP33:![0-9]+]]
// CHECK1:       omp.inner.for.end.i:
// CHECK1-NEXT:    [[TMP50:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !31
// CHECK1-NEXT:    [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
// CHECK1-NEXT:    br i1 [[TMP51]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__3_EXIT:%.*]]
// CHECK1:       .omp.lastprivate.then.i:
// CHECK1-NEXT:    [[TMP52:%.*]] = load i32, i32* [[TMP38]], align 128
// CHECK1-NEXT:    store i32 [[TMP52]], i32* [[TMP27]], align 128
// CHECK1-NEXT:    [[TMP53:%.*]] = bitcast [2 x i32]* [[TMP29]] to i8*
// CHECK1-NEXT:    [[TMP54:%.*]] = bitcast [2 x i32]* [[TMP39]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP53]], i8* align 4 [[TMP54]], i64 8, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN_I:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP31]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP55:%.*]] = bitcast [2 x %struct.S.0]* [[TMP40]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP56:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN_I]], i64 2
// CHECK1-NEXT:    br label [[OMP_ARRAYCPY_BODY_I:%.*]]
// CHECK1:       omp.arraycpy.body.i:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST_I:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST_I:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN_I]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK1-NEXT:    [[TMP57:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]] to i8*
// CHECK1-NEXT:    [[TMP58:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP57]], i8* align 4 [[TMP58]], i64 4, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT_I]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT_I]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE_I:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT_I]], [[TMP56]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE_I]], label [[OMP_ARRAYCPY_DONE7_I:%.*]], label [[OMP_ARRAYCPY_BODY_I]]
// CHECK1:       omp.arraycpy.done7.i:
// CHECK1-NEXT:    [[TMP59:%.*]] = bitcast %struct.S.0* [[TMP35]] to i8*
// CHECK1-NEXT:    [[TMP60:%.*]] = bitcast %struct.S.0* [[TMP41]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP59]], i8* align 4 [[TMP60]], i64 4, i1 false) #[[ATTR4]]
// CHECK1-NEXT:    br label [[DOTOMP_OUTLINED__3_EXIT]]
// CHECK1:       .omp_outlined..3.exit:
// CHECK1-NEXT:    ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_dup..6
// CHECK1-SAME: (%struct.kmp_task_t_with_privates.2* [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR7]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK1-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP0]], %struct.kmp_task_t_with_privates.2** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK1-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 64
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP7]], i32 0, i32 2
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP8]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP7]], i32 0, i32 3
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP9]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_destructor..7
// CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP2]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP3]], i32 0, i32 3
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP5]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    store i32 0, i32* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK1-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@main
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK2-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK2-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 16
// CHECK2-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]])
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]])
// CHECK2-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK2-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK2-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_BEGIN]], double 1.000000e+00)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_ELEMENT]], double 2.000000e+00)
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[VAR]], double 3.000000e+00)
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined. to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[VAR]])
// CHECK2-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK2-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR]]) #[[ATTR4:[0-9]+]]
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done1:
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]]) #[[ATTR4]]
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ev
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIdEC2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ed
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIdEC2Ed(%struct.S* nonnull dereferenceable(8) [[THIS1]], double [[TMP0]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 8 dereferenceable(16) [[S_ARR:%.*]], %struct.S* nonnull align 8 dereferenceable(8) [[VAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK2-NEXT:    [[TMP6:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    [[TMP7:%.*]] = icmp ne i32 [[TMP6]], 0
// CHECK2-NEXT:    br i1 [[TMP7]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK2:       omp_if.then:
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK2-NEXT:    store [2 x i32]* [[TMP0]], [2 x i32]** [[TMP8]], align 8
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK2-NEXT:    store [2 x %struct.S]* [[TMP2]], [2 x %struct.S]** [[TMP10]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 3
// CHECK2-NEXT:    store %struct.S* [[TMP3]], %struct.S** [[TMP11]], align 8
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 4
// CHECK2-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP12]], align 8
// CHECK2-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    [[TMP13:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 9, i64 120, i64 40, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.kmp_task_t_with_privates*
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP14]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP17]], i8* align 8 [[TMP18]], i64 40, i1 false)
// CHECK2-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP14]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP19]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP20]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK2:       arrayctor.loop:
// CHECK2-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[OMP_IF_THEN]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK2-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK2-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK2-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK2:       arrayctor.cont:
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP19]], i32 0, i32 1
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TMP21]])
// CHECK2-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP23:%.*]] = bitcast %union.kmp_cmplrdata_t* [[TMP22]] to i32 (i32, i8*)**
// CHECK2-NEXT:    store i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_destructor. to i32 (i32, i8*)*), i32 (i32, i8*)** [[TMP23]], align 8
// CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 5
// CHECK2-NEXT:    store i64 0, i64* [[TMP24]], align 8
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 6
// CHECK2-NEXT:    store i64 9, i64* [[TMP25]], align 8
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 7
// CHECK2-NEXT:    store i64 1, i64* [[TMP26]], align 8
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP15]], i32 0, i32 9
// CHECK2-NEXT:    [[TMP28:%.*]] = bitcast i8** [[TMP27]] to i8*
// CHECK2-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP28]], i8 0, i64 8, i1 false)
// CHECK2-NEXT:    [[TMP29:%.*]] = load i64, i64* [[TMP26]], align 8
// CHECK2-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i8* [[TMP13]], i32 1, i64* [[TMP24]], i64* [[TMP25]], i64 [[TMP29]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates*, i32)* @.omp_task_dup. to i8*))
// CHECK2-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    br label [[OMP_IF_END]]
// CHECK2:       omp_if.end:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK2-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], %struct.S** noalias [[TMP1:%.*]], i32** noalias [[TMP2:%.*]], [2 x %struct.S]** noalias [[TMP3:%.*]], [2 x i32]** noalias [[TMP4:%.*]], i32** noalias [[TMP5:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S**, align 8
// CHECK2-NEXT:    [[DOTADDR2:%.*]] = alloca i32**, align 8
// CHECK2-NEXT:    [[DOTADDR3:%.*]] = alloca [2 x %struct.S]**, align 8
// CHECK2-NEXT:    [[DOTADDR4:%.*]] = alloca [2 x i32]**, align 8
// CHECK2-NEXT:    [[DOTADDR5:%.*]] = alloca i32**, align 8
// CHECK2-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S** [[TMP1]], %struct.S*** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store i32** [[TMP2]], i32*** [[DOTADDR2]], align 8
// CHECK2-NEXT:    store [2 x %struct.S]** [[TMP3]], [2 x %struct.S]*** [[DOTADDR3]], align 8
// CHECK2-NEXT:    store [2 x i32]** [[TMP4]], [2 x i32]*** [[DOTADDR4]], align 8
// CHECK2-NEXT:    store i32** [[TMP5]], i32*** [[DOTADDR5]], align 8
// CHECK2-NEXT:    [[TMP6:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP8:%.*]] = load [2 x %struct.S]**, [2 x %struct.S]*** [[DOTADDR3]], align 8
// CHECK2-NEXT:    store [2 x %struct.S]* [[TMP7]], [2 x %struct.S]** [[TMP8]], align 8
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP10:%.*]] = load %struct.S**, %struct.S*** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP9]], %struct.S** [[TMP10]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP12:%.*]] = load i32**, i32*** [[DOTADDR2]], align 8
// CHECK2-NEXT:    store i32* [[TMP11]], i32** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP14:%.*]] = load [2 x i32]**, [2 x i32]*** [[DOTADDR4]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[TMP13]], [2 x i32]** [[TMP14]], align 8
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP6]], i32 0, i32 4
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32**, i32*** [[DOTADDR5]], align 8
// CHECK2-NEXT:    store i32* [[TMP15]], i32** [[TMP16]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR7:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK2-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR2_I:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR3_I:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR4_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK2-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK2-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK2-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK2-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK2-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK2-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, %struct.S**, i32**, [2 x %struct.S]**, [2 x i32]**, i32**)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK2-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP22:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, %struct.S**, i32**, [2 x %struct.S]**, [2 x i32]**, i32**)*
// CHECK2-NEXT:    call void [[TMP25]](i8* [[TMP24]], %struct.S** [[DOTLASTPRIV_PTR_ADDR_I]], i32** [[DOTLASTPRIV_PTR_ADDR1_I]], [2 x %struct.S]** [[DOTLASTPRIV_PTR_ADDR2_I]], [2 x i32]** [[DOTLASTPRIV_PTR_ADDR3_I]], i32** [[DOTLASTPRIV_PTR_ADDR4_I]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP22]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP27:%.*]] = load %struct.S*, %struct.S** [[TMP26]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP29:%.*]] = load i32*, i32** [[TMP28]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP31:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[TMP30]], align 8
// CHECK2-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP33:%.*]] = load [2 x i32]*, [2 x i32]** [[TMP32]], align 8
// CHECK2-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP35:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[TMP34]], align 8
// CHECK2-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP37:%.*]] = load %struct.S*, %struct.S** [[TMP36]], align 8
// CHECK2-NEXT:    [[TMP38:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 4
// CHECK2-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[TMP38]], align 8
// CHECK2-NEXT:    [[TMP40:%.*]] = load %struct.S*, %struct.S** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP41:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP42:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[DOTLASTPRIV_PTR_ADDR2_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP43:%.*]] = load [2 x i32]*, [2 x i32]** [[DOTLASTPRIV_PTR_ADDR3_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP44:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR4_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[TMP45:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK2-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP45]] to i32
// CHECK2-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK2:       omp.inner.for.cond.i:
// CHECK2-NEXT:    [[TMP46:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    [[CONV5_I:%.*]] = sext i32 [[TMP46]] to i64
// CHECK2-NEXT:    [[TMP47:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV5_I]], [[TMP47]]
// CHECK2-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK2:       omp.inner.for.body.i:
// CHECK2-NEXT:    [[TMP48:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    store i32 [[TMP48]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32, i32* [[TMP41]], align 4, !llvm.access.group !14
// CHECK2-NEXT:    [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP43]], i64 0, i64 0
// CHECK2-NEXT:    store i32 [[TMP49]], i32* [[ARRAYIDX_I]], align 4, !llvm.access.group !14
// CHECK2-NEXT:    [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP42]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP50:%.*]] = bitcast %struct.S* [[ARRAYIDX6_I]] to i8*
// CHECK2-NEXT:    [[TMP51:%.*]] = bitcast %struct.S* [[TMP40]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP50]], i8* align 8 [[TMP51]], i64 8, i1 false) #[[ATTR4]], !llvm.access.group !14
// CHECK2-NEXT:    store i32 33, i32* [[TMP44]], align 4, !llvm.access.group !14
// CHECK2-NEXT:    [[TMP52:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    [[ADD7_I:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK2-NEXT:    store i32 [[ADD7_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK2:       omp.inner.for.end.i:
// CHECK2-NEXT:    [[TMP53:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK2-NEXT:    [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0
// CHECK2-NEXT:    br i1 [[TMP54]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK2:       .omp.lastprivate.then.i:
// CHECK2-NEXT:    [[TMP55:%.*]] = bitcast %struct.S* [[TMP27]] to i8*
// CHECK2-NEXT:    [[TMP56:%.*]] = bitcast %struct.S* [[TMP40]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP55]], i8* align 8 [[TMP56]], i64 8, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP57:%.*]] = load i32, i32* [[TMP41]], align 4
// CHECK2-NEXT:    store i32 [[TMP57]], i32* [[TMP29]], align 4
// CHECK2-NEXT:    [[ARRAY_BEGIN_I:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP31]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP58:%.*]] = bitcast [2 x %struct.S]* [[TMP42]] to %struct.S*
// CHECK2-NEXT:    [[TMP59:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN_I]], i64 2
// CHECK2-NEXT:    br label [[OMP_ARRAYCPY_BODY_I:%.*]]
// CHECK2:       omp.arraycpy.body.i:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST_I:%.*]] = phi %struct.S* [ [[TMP58]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST_I:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN_I]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK2-NEXT:    [[TMP60:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]] to i8*
// CHECK2-NEXT:    [[TMP61:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP60]], i8* align 8 [[TMP61]], i64 8, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT_I]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT_I]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE_I:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT_I]], [[TMP59]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE_I]], label [[OMP_ARRAYCPY_DONE8_I:%.*]], label [[OMP_ARRAYCPY_BODY_I]]
// CHECK2:       omp.arraycpy.done8.i:
// CHECK2-NEXT:    [[TMP62:%.*]] = bitcast [2 x i32]* [[TMP33]] to i8*
// CHECK2-NEXT:    [[TMP63:%.*]] = bitcast [2 x i32]* [[TMP43]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP62]], i8* align 4 [[TMP63]], i64 8, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP64:%.*]] = load i32, i32* [[TMP44]], align 4
// CHECK2-NEXT:    store i32 [[TMP64]], i32* [[TMP39]], align 4
// CHECK2-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK2:       .omp_outlined..1.exit:
// CHECK2-NEXT:    ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_dup.
// CHECK2-SAME: (%struct.kmp_task_t_with_privates* [[TMP0:%.*]], %struct.kmp_task_t_with_privates* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR7]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK2-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP0]], %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK2-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 8
// CHECK2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP7]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP8]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK2:       arrayctor.loop:
// CHECK2-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK2-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK2-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK2-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK2:       arrayctor.cont:
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP7]], i32 0, i32 1
// CHECK2-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TMP9]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_destructor.
// CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP2]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 1
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TMP5]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done2:
// CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdED1Ev
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIdED2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK2-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK2-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK2-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK2-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK2-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]])
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]])
// CHECK2-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK2-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK2-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[VAR]], i32 3)
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [2 x i32]*, i32*, [2 x %struct.S.0]*, %struct.S.0*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S.0]* [[S_ARR]], %struct.S.0* [[VAR]])
// CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP1]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE1:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done1:
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ev
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT:    store double 0.000000e+00, double* [[F]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ed
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK2-NEXT:    store double [[TMP0]], double* [[F]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIdED2Ev
// CHECK2-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S.0]* nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_1:%.*]], align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    store [2 x %struct.S.0]* [[S_ARR]], [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
// CHECK2-NEXT:    [[TMP6:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    [[TMP7:%.*]] = icmp ne i32 [[TMP6]], 0
// CHECK2-NEXT:    br i1 [[TMP7]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK2:       omp_if.then:
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK2-NEXT:    store [2 x i32]* [[TMP0]], [2 x i32]** [[TMP8]], align 8
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK2-NEXT:    store [2 x %struct.S.0]* [[TMP2]], [2 x %struct.S.0]** [[TMP10]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[AGG_CAPTURED]], i32 0, i32 3
// CHECK2-NEXT:    store %struct.S.0* [[TMP3]], %struct.S.0** [[TMP11]], align 8
// CHECK2-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    [[TMP12:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i32 9, i64 256, i64 32, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates.2*)* @.omp_task_entry..5 to i32 (i32, i8*)*))
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast i8* [[TMP12]] to %struct.kmp_task_t_with_privates.2*
// CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP13]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 128
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast %struct.anon.1* [[AGG_CAPTURED]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP16]], i8* align 8 [[TMP17]], i64 32, i1 false)
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP13]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP18]], i32 0, i32 2
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP19]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK2:       arrayctor.loop:
// CHECK2-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[OMP_IF_THEN]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK2-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK2-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK2-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK2:       arrayctor.cont:
// CHECK2-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP18]], i32 0, i32 3
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP20]])
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP22:%.*]] = bitcast %union.kmp_cmplrdata_t* [[TMP21]] to i32 (i32, i8*)**
// CHECK2-NEXT:    store i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates.2*)* @.omp_task_destructor..7 to i32 (i32, i8*)*), i32 (i32, i8*)** [[TMP22]], align 8
// CHECK2-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 5
// CHECK2-NEXT:    store i64 0, i64* [[TMP23]], align 8
// CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 6
// CHECK2-NEXT:    store i64 9, i64* [[TMP24]], align 16
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 7
// CHECK2-NEXT:    store i64 1, i64* [[TMP25]], align 8
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP14]], i32 0, i32 9
// CHECK2-NEXT:    [[TMP27:%.*]] = bitcast i8** [[TMP26]] to i8*
// CHECK2-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP27]], i8 0, i64 8, i1 false)
// CHECK2-NEXT:    [[TMP28:%.*]] = load i64, i64* [[TMP25]], align 8
// CHECK2-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]], i8* [[TMP12]], i32 1, i64* [[TMP23]], i64* [[TMP24]], i64 [[TMP28]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2*, i32)* @.omp_task_dup..6 to i8*))
// CHECK2-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP5]])
// CHECK2-NEXT:    br label [[OMP_IF_END]]
// CHECK2:       omp_if.end:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map..4
// CHECK2-SAME: (%struct..kmp_privates.t.3* noalias [[TMP0:%.*]], i32** noalias [[TMP1:%.*]], [2 x i32]** noalias [[TMP2:%.*]], [2 x %struct.S.0]** noalias [[TMP3:%.*]], %struct.S.0** noalias [[TMP4:%.*]]) #[[ATTR6]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t.3*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32**, align 8
// CHECK2-NEXT:    [[DOTADDR2:%.*]] = alloca [2 x i32]**, align 8
// CHECK2-NEXT:    [[DOTADDR3:%.*]] = alloca [2 x %struct.S.0]**, align 8
// CHECK2-NEXT:    [[DOTADDR4:%.*]] = alloca %struct.S.0**, align 8
// CHECK2-NEXT:    store %struct..kmp_privates.t.3* [[TMP0]], %struct..kmp_privates.t.3** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32** [[TMP1]], i32*** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store [2 x i32]** [[TMP2]], [2 x i32]*** [[DOTADDR2]], align 8
// CHECK2-NEXT:    store [2 x %struct.S.0]** [[TMP3]], [2 x %struct.S.0]*** [[DOTADDR3]], align 8
// CHECK2-NEXT:    store %struct.S.0** [[TMP4]], %struct.S.0*** [[DOTADDR4]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load %struct..kmp_privates.t.3*, %struct..kmp_privates.t.3** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i32**, i32*** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store i32* [[TMP6]], i32** [[TMP7]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP9:%.*]] = load [2 x i32]**, [2 x i32]*** [[DOTADDR2]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[TMP8]], [2 x i32]** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP11:%.*]] = load [2 x %struct.S.0]**, [2 x %struct.S.0]*** [[DOTADDR3]], align 8
// CHECK2-NEXT:    store [2 x %struct.S.0]* [[TMP10]], [2 x %struct.S.0]** [[TMP11]], align 8
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP5]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP13:%.*]] = load %struct.S.0**, %struct.S.0*** [[DOTADDR4]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP12]], %struct.S.0** [[TMP13]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry..5
// CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK2-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.1*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR2_I:%.*]] = alloca [2 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[DOTLASTPRIV_PTR_ADDR3_I:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 128
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.1*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t.3* [[TMP9]] to i8*
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates.2* [[TMP3]] to i8*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK2-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK2-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 16
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK2-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 64
// CHECK2-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK2-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META20:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META23:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META25:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META27:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META29:![0-9]+]])
// CHECK2-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !31
// CHECK2-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t.3*, i32**, [2 x i32]**, [2 x %struct.S.0]**, %struct.S.0**)* @.omp_task_privates_map..4 to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !31
// CHECK2-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    store %struct.anon.1* [[TMP8]], %struct.anon.1** [[__CONTEXT_ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP22:%.*]] = load %struct.anon.1*, %struct.anon.1** [[__CONTEXT_ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, i32**, [2 x i32]**, [2 x %struct.S.0]**, %struct.S.0**)*
// CHECK2-NEXT:    call void [[TMP25]](i8* [[TMP24]], i32** [[DOTLASTPRIV_PTR_ADDR_I]], [2 x i32]** [[DOTLASTPRIV_PTR_ADDR1_I]], [2 x %struct.S.0]** [[DOTLASTPRIV_PTR_ADDR2_I]], %struct.S.0** [[DOTLASTPRIV_PTR_ADDR3_I]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_1:%.*]], %struct.anon.1* [[TMP22]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP29:%.*]] = load [2 x i32]*, [2 x i32]** [[TMP28]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP31:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[TMP30]], align 8
// CHECK2-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP33:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[TMP32]], align 8
// CHECK2-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP35:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP34]], align 8
// CHECK2-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON_1]], %struct.anon.1* [[TMP22]], i32 0, i32 3
// CHECK2-NEXT:    [[TMP37:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP36]], align 8
// CHECK2-NEXT:    [[TMP38:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP39:%.*]] = load [2 x i32]*, [2 x i32]** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP40:%.*]] = load [2 x %struct.S.0]*, [2 x %struct.S.0]** [[DOTLASTPRIV_PTR_ADDR2_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP41:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTLASTPRIV_PTR_ADDR3_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[TMP42:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !31
// CHECK2-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP42]] to i32
// CHECK2-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !31
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK2:       omp.inner.for.cond.i:
// CHECK2-NEXT:    [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    [[CONV4_I:%.*]] = sext i32 [[TMP43]] to i64
// CHECK2-NEXT:    [[TMP44:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV4_I]], [[TMP44]]
// CHECK2-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK2:       omp.inner.for.body.i:
// CHECK2-NEXT:    [[TMP45:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    store i32 [[TMP45]], i32* [[I_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    [[TMP46:%.*]] = load i32, i32* [[TMP38]], align 128, !llvm.access.group !32
// CHECK2-NEXT:    [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP39]], i64 0, i64 0
// CHECK2-NEXT:    store i32 [[TMP46]], i32* [[ARRAYIDX_I]], align 4, !llvm.access.group !32
// CHECK2-NEXT:    [[ARRAYIDX5_I:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP40]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP47:%.*]] = bitcast %struct.S.0* [[ARRAYIDX5_I]] to i8*
// CHECK2-NEXT:    [[TMP48:%.*]] = bitcast %struct.S.0* [[TMP41]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP47]], i8* align 4 [[TMP48]], i64 4, i1 false) #[[ATTR4]], !llvm.access.group !32
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    [[ADD6_I:%.*]] = add nsw i32 [[TMP49]], 1
// CHECK2-NEXT:    store i32 [[ADD6_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !31, !llvm.access.group !32
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP33:![0-9]+]]
// CHECK2:       omp.inner.for.end.i:
// CHECK2-NEXT:    [[TMP50:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !31
// CHECK2-NEXT:    [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
// CHECK2-NEXT:    br i1 [[TMP51]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__3_EXIT:%.*]]
// CHECK2:       .omp.lastprivate.then.i:
// CHECK2-NEXT:    [[TMP52:%.*]] = load i32, i32* [[TMP38]], align 128
// CHECK2-NEXT:    store i32 [[TMP52]], i32* [[TMP27]], align 128
// CHECK2-NEXT:    [[TMP53:%.*]] = bitcast [2 x i32]* [[TMP29]] to i8*
// CHECK2-NEXT:    [[TMP54:%.*]] = bitcast [2 x i32]* [[TMP39]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP53]], i8* align 4 [[TMP54]], i64 8, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN_I:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP31]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP55:%.*]] = bitcast [2 x %struct.S.0]* [[TMP40]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP56:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN_I]], i64 2
// CHECK2-NEXT:    br label [[OMP_ARRAYCPY_BODY_I:%.*]]
// CHECK2:       omp.arraycpy.body.i:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST_I:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST_I:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN_I]], [[DOTOMP_LASTPRIVATE_THEN_I]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT_I:%.*]], [[OMP_ARRAYCPY_BODY_I]] ]
// CHECK2-NEXT:    [[TMP57:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]] to i8*
// CHECK2-NEXT:    [[TMP58:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP57]], i8* align 4 [[TMP58]], i64 4, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT_I]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST_I]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT_I]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST_I]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE_I:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT_I]], [[TMP56]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE_I]], label [[OMP_ARRAYCPY_DONE7_I:%.*]], label [[OMP_ARRAYCPY_BODY_I]]
// CHECK2:       omp.arraycpy.done7.i:
// CHECK2-NEXT:    [[TMP59:%.*]] = bitcast %struct.S.0* [[TMP35]] to i8*
// CHECK2-NEXT:    [[TMP60:%.*]] = bitcast %struct.S.0* [[TMP41]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP59]], i8* align 4 [[TMP60]], i64 4, i1 false) #[[ATTR4]]
// CHECK2-NEXT:    br label [[DOTOMP_OUTLINED__3_EXIT]]
// CHECK2:       .omp_outlined..3.exit:
// CHECK2-NEXT:    ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_dup..6
// CHECK2-SAME: (%struct.kmp_task_t_with_privates.2* [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR7]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK2-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP0]], %struct.kmp_task_t_with_privates.2** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK2-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 64
// CHECK2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2]], %struct.kmp_task_t_with_privates.2* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP7]], i32 0, i32 2
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP8]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK2:       arrayctor.loop:
// CHECK2-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK2-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK2-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK2-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK2:       arrayctor.cont:
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP7]], i32 0, i32 3
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP9]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_destructor..7
// CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates.2* noalias [[TMP1:%.*]]) #[[ATTR7]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates.2*, align 8
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates.2* [[TMP1]], %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.kmp_task_t_with_privates.2*, %struct.kmp_task_t_with_privates.2** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES_2:%.*]], %struct.kmp_task_t_with_privates.2* [[TMP2]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3:%.*]], %struct..kmp_privates.t.3* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T_3]], %struct..kmp_privates.t.3* [[TMP3]], i32 0, i32 3
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TMP5]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[TMP4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP6]], [[ENTRY:%.*]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done2:
// CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP7]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT:    store i32 0, i32* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK2-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull dereferenceable(1) [[REF_TMP]])
// CHECK3-NEXT:    ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK3-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK3-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]])
// CHECK3-NEXT:    [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK3-NEXT:    br i1 [[TMP3]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK3:       omp_if.then:
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK3-NEXT:    store double* @g, double** [[TMP4]], align 8
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK3-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP5]], align 8
// CHECK3-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK3-NEXT:    [[TMP6:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 96, i64 16, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK3-NEXT:    [[TMP7:%.*]] = bitcast i8* [[TMP6]] to %struct.kmp_task_t_with_privates*
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP7]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK3-NEXT:    [[TMP11:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP10]], i8* align 8 [[TMP11]], i64 16, i1 false)
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP7]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 5
// CHECK3-NEXT:    store i64 0, i64* [[TMP13]], align 8
// CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 6
// CHECK3-NEXT:    store i64 9, i64* [[TMP14]], align 8
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 7
// CHECK3-NEXT:    store i64 1, i64* [[TMP15]], align 8
// CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 9
// CHECK3-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i8*
// CHECK3-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP17]], i8 0, i64 8, i1 false)
// CHECK3-NEXT:    [[TMP18:%.*]] = load i64, i64* [[TMP15]], align 8
// CHECK3-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP6]], i32 1, i64* [[TMP13]], i64* [[TMP14]], i64 [[TMP18]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates*, i32)* @.omp_task_dup. to i8*))
// CHECK3-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK3-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK3-NEXT:    br label [[OMP_IF_END]]
// CHECK3:       omp_if.end:
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK3-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], double** noalias [[TMP1:%.*]], i32** noalias [[TMP2:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK3-NEXT:    [[DOTADDR1:%.*]] = alloca double**, align 8
// CHECK3-NEXT:    [[DOTADDR2:%.*]] = alloca i32**, align 8
// CHECK3-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK3-NEXT:    store double** [[TMP1]], double*** [[DOTADDR1]], align 8
// CHECK3-NEXT:    store i32** [[TMP2]], i32*** [[DOTADDR2]], align 8
// CHECK3-NEXT:    [[TMP3:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP5:%.*]] = load double**, double*** [[DOTADDR1]], align 8
// CHECK3-NEXT:    store double* [[TMP4]], double** [[TMP5]], align 8
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32**, i32*** [[DOTADDR2]], align 8
// CHECK3-NEXT:    store i32* [[TMP6]], i32** [[TMP7]], align 8
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK3-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK3-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK3-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK3-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK3-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK3-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK3-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK3-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK3-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK3-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca double*, align 8
// CHECK3-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca i32*, align 8
// CHECK3-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[REF_TMP_I:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK3-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK3-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK3-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK3-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK3-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK3-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK3-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK3-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 8
// CHECK3-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK3-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK3-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK3-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK3-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK3-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK3-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK3-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK3-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, double**, i32**)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK3-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP22:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, double**, i32**)*
// CHECK3-NEXT:    call void [[TMP25]](i8* [[TMP24]], double** [[DOTLASTPRIV_PTR_ADDR_I]], i32** [[DOTLASTPRIV_PTR_ADDR1_I]]) #[[ATTR3:[0-9]+]]
// CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP22]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP27:%.*]] = load double*, double** [[TMP26]], align 8
// CHECK3-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 1
// CHECK3-NEXT:    [[TMP29:%.*]] = load i32*, i32** [[TMP28]], align 8
// CHECK3-NEXT:    [[TMP30:%.*]] = load double*, double** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[TMP32:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK3-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP32]] to i32
// CHECK3-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK3:       omp.inner.for.cond.i:
// CHECK3-NEXT:    [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    [[CONV2_I:%.*]] = sext i32 [[TMP33]] to i64
// CHECK3-NEXT:    [[TMP34:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV2_I]], [[TMP34]]
// CHECK3-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK3:       omp.inner.for.body.i:
// CHECK3-NEXT:    [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    store i32 [[TMP35]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    store double 1.000000e+00, double* [[TMP30]], align 8, !llvm.access.group !14
// CHECK3-NEXT:    store i32 11, i32* [[TMP31]], align 4, !llvm.access.group !14
// CHECK3-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP_I]], i32 0, i32 0
// CHECK3-NEXT:    store double* [[TMP30]], double** [[TMP36]], align 8, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP_I]], i32 0, i32 1
// CHECK3-NEXT:    store i32* [[TMP31]], i32** [[TMP37]], align 8, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* nonnull dereferenceable(16) [[REF_TMP_I]]) #[[ATTR3]], !llvm.access.group !14
// CHECK3-NEXT:    [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    [[ADD3_I:%.*]] = add nsw i32 [[TMP38]], 1
// CHECK3-NEXT:    store i32 [[ADD3_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK3:       omp.inner.for.end.i:
// CHECK3-NEXT:    [[TMP39:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK3-NEXT:    [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0
// CHECK3-NEXT:    br i1 [[TMP40]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK3:       .omp.lastprivate.then.i:
// CHECK3-NEXT:    [[TMP41:%.*]] = load double, double* [[TMP30]], align 8
// CHECK3-NEXT:    store volatile double [[TMP41]], double* [[TMP27]], align 8
// CHECK3-NEXT:    [[TMP42:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK3-NEXT:    store i32 [[TMP42]], i32* [[TMP29]], align 4
// CHECK3-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK3:       .omp_outlined..1.exit:
// CHECK3-NEXT:    ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_task_dup.
// CHECK3-SAME: (%struct.kmp_task_t_with_privates* [[TMP0:%.*]], %struct.kmp_task_t_with_privates* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR6]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK3-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK3-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP0]], %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK3-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK3-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK3-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK3-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 8
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK3-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@main
// CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load i8*, i8** getelementptr inbounds ([[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to %struct.__block_literal_generic*), i32 0, i32 3), align 8
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast i8* [[TMP0]] to void (i8*)*
// CHECK4-NEXT:    call void [[TMP1]](i8* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*))
// CHECK4-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*, align 8
// CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*
// CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>** [[BLOCK_ADDR]], align 8
// CHECK4-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK4-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK4-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK4-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK4-NEXT:    [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK4-NEXT:    br i1 [[TMP3]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK4:       omp_if.then:
// CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK4-NEXT:    store double* @g, double** [[TMP4]], align 8
// CHECK4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK4-NEXT:    store i32* @_ZZ4mainE5sivar, i32** [[TMP5]], align 8
// CHECK4-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK4-NEXT:    [[TMP6:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 96, i64 16, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK4-NEXT:    [[TMP7:%.*]] = bitcast i8* [[TMP6]] to %struct.kmp_task_t_with_privates*
// CHECK4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP7]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK4-NEXT:    [[TMP11:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP10]], i8* align 8 [[TMP11]], i64 16, i1 false)
// CHECK4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP7]], i32 0, i32 1
// CHECK4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 5
// CHECK4-NEXT:    store i64 0, i64* [[TMP13]], align 8
// CHECK4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 6
// CHECK4-NEXT:    store i64 9, i64* [[TMP14]], align 8
// CHECK4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 7
// CHECK4-NEXT:    store i64 1, i64* [[TMP15]], align 8
// CHECK4-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP8]], i32 0, i32 9
// CHECK4-NEXT:    [[TMP17:%.*]] = bitcast i8** [[TMP16]] to i8*
// CHECK4-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP17]], i8 0, i64 8, i1 false)
// CHECK4-NEXT:    [[TMP18:%.*]] = load i64, i64* [[TMP15]], align 8
// CHECK4-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP6]], i32 1, i64* [[TMP13]], i64* [[TMP14]], i64 [[TMP18]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates*, i32)* @.omp_task_dup. to i8*))
// CHECK4-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK4-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK4-NEXT:    br label [[OMP_IF_END]]
// CHECK4:       omp_if.end:
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_block_invoke
// CHECK4-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>*, align 8
// CHECK4-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>*
// CHECK4-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>** [[BLOCK_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    store double 2.000000e+00, double* [[BLOCK_CAPTURE_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    store i32 22, i32* [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK4-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], double** noalias [[TMP1:%.*]], i32** noalias [[TMP2:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca double**, align 8
// CHECK4-NEXT:    [[DOTADDR2:%.*]] = alloca i32**, align 8
// CHECK4-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK4-NEXT:    store double** [[TMP1]], double*** [[DOTADDR1]], align 8
// CHECK4-NEXT:    store i32** [[TMP2]], i32*** [[DOTADDR2]], align 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP5:%.*]] = load double**, double*** [[DOTADDR1]], align 8
// CHECK4-NEXT:    store double* [[TMP4]], double** [[TMP5]], align 8
// CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 1
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32**, i32*** [[DOTADDR2]], align 8
// CHECK4-NEXT:    store i32* [[TMP6]], i32** [[TMP7]], align 8
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK4-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR7:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK4-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK4-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK4-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK4-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK4-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK4-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca double*, align 8
// CHECK4-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca i32*, align 8
// CHECK4-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[BLOCK_I:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, align 8
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK4-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK4-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK4-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK4-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK4-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK4-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK4-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK4-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 8
// CHECK4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK4-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK4-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK4-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK4-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK4-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK4-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK4-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK4-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, double**, i32**)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK4-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP22:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP23:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP24:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP25:%.*]] = bitcast void (i8*, ...)* [[TMP23]] to void (i8*, double**, i32**)*
// CHECK4-NEXT:    call void [[TMP25]](i8* [[TMP24]], double** [[DOTLASTPRIV_PTR_ADDR_I]], i32** [[DOTLASTPRIV_PTR_ADDR1_I]]) #[[ATTR4:[0-9]+]]
// CHECK4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP22]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP27:%.*]] = load double*, double** [[TMP26]], align 8
// CHECK4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 1
// CHECK4-NEXT:    [[TMP29:%.*]] = load i32*, i32** [[TMP28]], align 8
// CHECK4-NEXT:    [[TMP30:%.*]] = load double*, double** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[TMP32:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK4-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP32]] to i32
// CHECK4-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK4:       omp.inner.for.cond.i:
// CHECK4-NEXT:    [[TMP33:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[CONV2_I:%.*]] = sext i32 [[TMP33]] to i64
// CHECK4-NEXT:    [[TMP34:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV2_I]], [[TMP34]]
// CHECK4-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK4:       omp.inner.for.body.i:
// CHECK4-NEXT:    [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    store i32 [[TMP35]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    store double 1.000000e+00, double* [[TMP30]], align 8, !llvm.access.group !14
// CHECK4-NEXT:    store i32 11, i32* [[TMP31]], align 4, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_ISA_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 0
// CHECK4-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_FLAGS_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_RESERVED_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, i32* [[BLOCK_RESERVED_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_INVOKE_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 3
// CHECK4-NEXT:    store i8* bitcast (void (i8*)* @_block_invoke to i8*), i8** [[BLOCK_INVOKE_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 4
// CHECK4-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.2 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_CAPTURED_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP36:%.*]] = load volatile double, double* [[TMP30]], align 8, !llvm.access.group !14
// CHECK4-NEXT:    store volatile double [[TMP36]], double* [[BLOCK_CAPTURED_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[BLOCK_CAPTURED3_I:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP37:%.*]] = load i32, i32* [[TMP31]], align 4, !llvm.access.group !14
// CHECK4-NEXT:    store i32 [[TMP37]], i32* [[BLOCK_CAPTURED3_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[TMP38:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK_I]] to void ()*
// CHECK4-NEXT:    [[BLOCK_LITERAL_I:%.*]] = bitcast void ()* [[TMP38]] to %struct.__block_literal_generic*
// CHECK4-NEXT:    [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL_I]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP40:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL_I]] to i8*
// CHECK4-NEXT:    [[TMP41:%.*]] = load i8*, i8** [[TMP39]], align 8, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[TMP42:%.*]] = bitcast i8* [[TMP41]] to void (i8*)*
// CHECK4-NEXT:    call void [[TMP42]](i8* [[TMP40]]) #[[ATTR4]], !llvm.access.group !14
// CHECK4-NEXT:    [[TMP43:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    [[ADD4_I:%.*]] = add nsw i32 [[TMP43]], 1
// CHECK4-NEXT:    store i32 [[ADD4_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK4:       omp.inner.for.end.i:
// CHECK4-NEXT:    [[TMP44:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK4-NEXT:    [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK4-NEXT:    br i1 [[TMP45]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK4:       .omp.lastprivate.then.i:
// CHECK4-NEXT:    [[TMP46:%.*]] = load double, double* [[TMP30]], align 8
// CHECK4-NEXT:    store volatile double [[TMP46]], double* [[TMP27]], align 8
// CHECK4-NEXT:    [[TMP47:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK4-NEXT:    store i32 [[TMP47]], i32* [[TMP29]], align 4
// CHECK4-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK4:       .omp_outlined..1.exit:
// CHECK4-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_task_dup.
// CHECK4-SAME: (%struct.kmp_task_t_with_privates* [[TMP0:%.*]], %struct.kmp_task_t_with_privates* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR7]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK4-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP0]], %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK4-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK4-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK4-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK4-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 8
// CHECK4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK4-NEXT:    ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@_Z10array_funciPfP2St
// CHECK5-SAME: (i32 [[N:%.*]], float* [[A:%.*]], %struct.St* [[S:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK5-NEXT:  entry:
// CHECK5-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK5-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK5-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK5-NEXT:    store float* [[A]], float** [[A_ADDR]], align 8
// CHECK5-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK5-NEXT:    [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK5-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK5-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, float**, %struct.St**)* @.omp_outlined. to void (i32*, i32*, ...)*), i64 [[TMP1]], float** [[A_ADDR]], %struct.St** [[S_ADDR]])
// CHECK5-NEXT:    ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], float** nonnull align 8 dereferenceable(8) [[A:%.*]], %struct.St** nonnull align 8 dereferenceable(8) [[S:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK5-NEXT:  entry:
// CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT:    [[A_ADDR:%.*]] = alloca float**, align 8
// CHECK5-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St**, align 8
// CHECK5-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK5-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK5-NEXT:    store float** [[A]], float*** [[A_ADDR]], align 8
// CHECK5-NEXT:    store %struct.St** [[S]], %struct.St*** [[S_ADDR]], align 8
// CHECK5-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK5-NEXT:    [[TMP1:%.*]] = load float**, float*** [[A_ADDR]], align 8
// CHECK5-NEXT:    [[TMP2:%.*]] = load %struct.St**, %struct.St*** [[S_ADDR]], align 8
// CHECK5-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK5-NEXT:    [[TMP5:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT:    [[TMP6:%.*]] = icmp ne i32 [[TMP5]], 0
// CHECK5-NEXT:    br i1 [[TMP6]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK5:       omp_if.then:
// CHECK5-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK5-NEXT:    store i64 [[TMP0]], i64* [[TMP7]], align 8
// CHECK5-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK5-NEXT:    store float** [[TMP1]], float*** [[TMP8]], align 8
// CHECK5-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK5-NEXT:    store %struct.St** [[TMP2]], %struct.St*** [[TMP9]], align 8
// CHECK5-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT:    [[TMP10:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], i32 1, i64 96, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK5-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.kmp_task_t_with_privates*
// CHECK5-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP11]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP12]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP14:%.*]] = load i8*, i8** [[TMP13]], align 8
// CHECK5-NEXT:    [[TMP15:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK5-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP14]], i8* align 8 [[TMP15]], i64 24, i1 false)
// CHECK5-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP11]], i32 0, i32 1
// CHECK5-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP12]], i32 0, i32 5
// CHECK5-NEXT:    store i64 0, i64* [[TMP17]], align 8
// CHECK5-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP12]], i32 0, i32 6
// CHECK5-NEXT:    store i64 9, i64* [[TMP18]], align 8
// CHECK5-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP12]], i32 0, i32 7
// CHECK5-NEXT:    store i64 1, i64* [[TMP19]], align 8
// CHECK5-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP12]], i32 0, i32 9
// CHECK5-NEXT:    [[TMP21:%.*]] = bitcast i8** [[TMP20]] to i8*
// CHECK5-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP21]], i8 0, i64 8, i1 false)
// CHECK5-NEXT:    [[TMP22:%.*]] = load i64, i64* [[TMP19]], align 8
// CHECK5-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], i8* [[TMP10]], i32 1, i64* [[TMP17]], i64* [[TMP18]], i64 [[TMP22]], i32 1, i32 0, i64 0, i8* bitcast (void (%struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates*, i32)* @.omp_task_dup. to i8*))
// CHECK5-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT:    br label [[OMP_IF_END]]
// CHECK5:       omp_if.end:
// CHECK5-NEXT:    ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK5-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], float*** noalias [[TMP1:%.*]], %struct.St*** noalias [[TMP2:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK5-NEXT:  entry:
// CHECK5-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK5-NEXT:    [[DOTADDR1:%.*]] = alloca float***, align 8
// CHECK5-NEXT:    [[DOTADDR2:%.*]] = alloca %struct.St***, align 8
// CHECK5-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK5-NEXT:    store float*** [[TMP1]], float**** [[DOTADDR1]], align 8
// CHECK5-NEXT:    store %struct.St*** [[TMP2]], %struct.St**** [[DOTADDR2]], align 8
// CHECK5-NEXT:    [[TMP3:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK5-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP5:%.*]] = load float***, float**** [[DOTADDR1]], align 8
// CHECK5-NEXT:    store float** [[TMP4]], float*** [[TMP5]], align 8
// CHECK5-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T]], %struct..kmp_privates.t* [[TMP3]], i32 0, i32 1
// CHECK5-NEXT:    [[TMP7:%.*]] = load %struct.St***, %struct.St**** [[DOTADDR2]], align 8
// CHECK5-NEXT:    store %struct.St** [[TMP6]], %struct.St*** [[TMP7]], align 8
// CHECK5-NEXT:    ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK5-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK5-NEXT:  entry:
// CHECK5-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK5-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK5-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK5-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK5-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK5-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK5-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK5-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK5-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK5-NEXT:    [[DOTLASTPRIV_PTR_ADDR_I:%.*]] = alloca float**, align 8
// CHECK5-NEXT:    [[DOTLASTPRIV_PTR_ADDR1_I:%.*]] = alloca %struct.St**, align 8
// CHECK5-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK5-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK5-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK5-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK5-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK5-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK5-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK5-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK5-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK5-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK5-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK5-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK5-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK5-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK5-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK5-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK5-NEXT:    [[TMP17:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK5-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK5-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 8
// CHECK5-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK5-NEXT:    [[TMP21:%.*]] = load i8*, i8** [[TMP20]], align 8
// CHECK5-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK5-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK5-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK5-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK5-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK5-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK5-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, float***, %struct.St***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i64 [[TMP13]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i64 [[TMP15]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i64 [[TMP17]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store i32 [[TMP19]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK5-NEXT:    store i8* [[TMP21]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP22:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP22]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP24:%.*]] = load i64, i64* [[TMP23]], align 8
// CHECK5-NEXT:    [[TMP25:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP26:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP27:%.*]] = bitcast void (i8*, ...)* [[TMP25]] to void (i8*, float***, %struct.St***)*
// CHECK5-NEXT:    call void [[TMP27]](i8* [[TMP26]], float*** [[DOTLASTPRIV_PTR_ADDR_I]], %struct.St*** [[DOTLASTPRIV_PTR_ADDR1_I]]) #[[ATTR2:[0-9]+]]
// CHECK5-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 1
// CHECK5-NEXT:    [[TMP29:%.*]] = load float**, float*** [[TMP28]], align 8
// CHECK5-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP22]], i32 0, i32 2
// CHECK5-NEXT:    [[TMP31:%.*]] = load %struct.St**, %struct.St*** [[TMP30]], align 8
// CHECK5-NEXT:    [[TMP32:%.*]] = load float**, float*** [[DOTLASTPRIV_PTR_ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP33:%.*]] = load %struct.St**, %struct.St*** [[DOTLASTPRIV_PTR_ADDR1_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[TMP34:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK5-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP34]] to i32
// CHECK5-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK5-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK5:       omp.inner.for.cond.i:
// CHECK5-NEXT:    [[TMP35:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    [[CONV2_I:%.*]] = sext i32 [[TMP35]] to i64
// CHECK5-NEXT:    [[TMP36:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV2_I]], [[TMP36]]
// CHECK5-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK5:       omp.inner.for.body.i:
// CHECK5-NEXT:    [[TMP37:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    store i32 [[TMP37]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    [[TMP38:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    [[ADD3_I:%.*]] = add nsw i32 [[TMP38]], 1
// CHECK5-NEXT:    store i32 [[ADD3_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK5-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK5:       omp.inner.for.end.i:
// CHECK5-NEXT:    [[TMP39:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK5-NEXT:    [[TMP40:%.*]] = icmp ne i32 [[TMP39]], 0
// CHECK5-NEXT:    br i1 [[TMP40]], label [[DOTOMP_LASTPRIVATE_THEN_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK5:       .omp.lastprivate.then.i:
// CHECK5-NEXT:    [[TMP41:%.*]] = load float*, float** [[TMP32]], align 8
// CHECK5-NEXT:    store float* [[TMP41]], float** [[TMP29]], align 8
// CHECK5-NEXT:    [[TMP42:%.*]] = load %struct.St*, %struct.St** [[TMP33]], align 8
// CHECK5-NEXT:    store %struct.St* [[TMP42]], %struct.St** [[TMP31]], align 8
// CHECK5-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK5:       .omp_outlined..1.exit:
// CHECK5-NEXT:    ret i32 0
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_task_dup.
// CHECK5-SAME: (%struct.kmp_task_t_with_privates* [[TMP0:%.*]], %struct.kmp_task_t_with_privates* [[TMP1:%.*]], i32 [[TMP2:%.*]]) #[[ATTR5]] {
// CHECK5-NEXT:  entry:
// CHECK5-NEXT:    [[DOTADDR:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK5-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK5-NEXT:    [[DOTADDR2:%.*]] = alloca i32, align 4
// CHECK5-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP0]], %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK5-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK5-NEXT:    store i32 [[TMP2]], i32* [[DOTADDR2]], align 4
// CHECK5-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR]], align 8
// CHECK5-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK5-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK5-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTADDR2]], align 4
// CHECK5-NEXT:    store i32 [[TMP6]], i32* [[TMP5]], align 8
// CHECK5-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK5-NEXT:    ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@_Z4loopv
// CHECK6-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK6-NEXT:  entry:
// CHECK6-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[I]], i32* [[J]])
// CHECK6-NEXT:    ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[I:%.*]], i32* nonnull align 4 dereferenceable(4) [[J:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK6-NEXT:  entry:
// CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT:    [[I_ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT:    [[J_ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK6-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT:    store i32* [[I]], i32** [[I_ADDR]], align 8
// CHECK6-NEXT:    store i32* [[J]], i32** [[J_ADDR]], align 8
// CHECK6-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[I_ADDR]], align 8
// CHECK6-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[J_ADDR]], align 8
// CHECK6-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT:    [[TMP4:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT:    [[TMP5:%.*]] = icmp ne i32 [[TMP4]], 0
// CHECK6-NEXT:    br i1 [[TMP5]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_END:%.*]]
// CHECK6:       omp_if.then:
// CHECK6-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK6-NEXT:    store i32* [[TMP0]], i32** [[TMP6]], align 8
// CHECK6-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK6-NEXT:    store i32* [[TMP1]], i32** [[TMP7]], align 8
// CHECK6-NEXT:    call void @__kmpc_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT:    [[TMP8:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i32 1, i64 80, i64 16, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK6-NEXT:    [[TMP9:%.*]] = bitcast i8* [[TMP8]] to %struct.kmp_task_t_with_privates*
// CHECK6-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP9]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP10]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP12:%.*]] = load i8*, i8** [[TMP11]], align 8
// CHECK6-NEXT:    [[TMP13:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK6-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP12]], i8* align 8 [[TMP13]], i64 16, i1 false)
// CHECK6-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP10]], i32 0, i32 5
// CHECK6-NEXT:    store i64 0, i64* [[TMP14]], align 8
// CHECK6-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP10]], i32 0, i32 6
// CHECK6-NEXT:    store i64 9, i64* [[TMP15]], align 8
// CHECK6-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP10]], i32 0, i32 7
// CHECK6-NEXT:    store i64 1, i64* [[TMP16]], align 8
// CHECK6-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP10]], i32 0, i32 9
// CHECK6-NEXT:    [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i8*
// CHECK6-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP18]], i8 0, i64 8, i1 false)
// CHECK6-NEXT:    [[TMP19:%.*]] = load i64, i64* [[TMP16]], align 8
// CHECK6-NEXT:    call void @__kmpc_taskloop(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]], i8* [[TMP8]], i32 1, i64* [[TMP14]], i64* [[TMP15]], i64 [[TMP19]], i32 1, i32 0, i64 0, i8* null)
// CHECK6-NEXT:    call void @__kmpc_end_taskgroup(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT:    br label [[OMP_IF_END]]
// CHECK6:       omp_if.end:
// CHECK6-NEXT:    ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK6-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK6-NEXT:  entry:
// CHECK6-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK6-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK6-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK6-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK6-NEXT:    [[DOTLB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK6-NEXT:    [[DOTUB__ADDR_I:%.*]] = alloca i64, align 8
// CHECK6-NEXT:    [[DOTST__ADDR_I:%.*]] = alloca i64, align 8
// CHECK6-NEXT:    [[DOTLITER__ADDR_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTREDUCTIONS__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK6-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK6-NEXT:    [[DOTLINEAR_START_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTLINEAR_START1_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[I_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[J_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTOMP_IV_I:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK6-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK6-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK6-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK6-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK6-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK6-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK6-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK6-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK6-NEXT:    [[TMP9:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK6-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 5
// CHECK6-NEXT:    [[TMP11:%.*]] = load i64, i64* [[TMP10]], align 8
// CHECK6-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 6
// CHECK6-NEXT:    [[TMP13:%.*]] = load i64, i64* [[TMP12]], align 8
// CHECK6-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 7
// CHECK6-NEXT:    [[TMP15:%.*]] = load i64, i64* [[TMP14]], align 8
// CHECK6-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 8
// CHECK6-NEXT:    [[TMP17:%.*]] = load i32, i32* [[TMP16]], align 8
// CHECK6-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 9
// CHECK6-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK6-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK6-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK6-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK6-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK6-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META11:![0-9]+]])
// CHECK6-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !13
// CHECK6-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i8* null, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store void (i8*, ...)* null, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i8* [[TMP9]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i64 [[TMP11]], i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i64 [[TMP13]], i64* [[DOTUB__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i64 [[TMP15]], i64* [[DOTST__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store i32 [[TMP17]], i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK6-NEXT:    store i8* [[TMP19]], i8** [[DOTREDUCTIONS__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    [[TMP20:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP20]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP22:%.*]] = load i32*, i32** [[TMP21]], align 8
// CHECK6-NEXT:    [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4
// CHECK6-NEXT:    store i32 [[TMP23]], i32* [[DOTLINEAR_START_I]], align 4, !noalias !13
// CHECK6-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP20]], i32 0, i32 1
// CHECK6-NEXT:    [[TMP25:%.*]] = load i32*, i32** [[TMP24]], align 8
// CHECK6-NEXT:    [[TMP26:%.*]] = load i32, i32* [[TMP25]], align 4
// CHECK6-NEXT:    store i32 [[TMP26]], i32* [[DOTLINEAR_START1_I]], align 4, !noalias !13
// CHECK6-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP20]], i32 0, i32 0
// CHECK6-NEXT:    [[TMP28:%.*]] = load i32*, i32** [[TMP27]], align 8
// CHECK6-NEXT:    [[TMP29:%.*]] = load i64, i64* [[DOTLB__ADDR_I]], align 8, !noalias !13
// CHECK6-NEXT:    [[CONV_I:%.*]] = trunc i64 [[TMP29]] to i32
// CHECK6-NEXT:    store i32 [[CONV_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13
// CHECK6-NEXT:    br label [[OMP_INNER_FOR_COND_I:%.*]]
// CHECK6:       omp.inner.for.cond.i:
// CHECK6-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[CONV3_I:%.*]] = sext i32 [[TMP30]] to i64
// CHECK6-NEXT:    [[TMP31:%.*]] = load i64, i64* [[DOTUB__ADDR_I]], align 8, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[CMP_I:%.*]] = icmp ule i64 [[CONV3_I]], [[TMP31]]
// CHECK6-NEXT:    br i1 [[CMP_I]], label [[OMP_INNER_FOR_BODY_I:%.*]], label [[OMP_INNER_FOR_END_I:%.*]]
// CHECK6:       omp.inner.for.body.i:
// CHECK6-NEXT:    [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    store i32 [[TMP32]], i32* [[I_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[TMP33:%.*]] = load i32, i32* [[DOTLINEAR_START1_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[TMP34:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[ADD5_I:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
// CHECK6-NEXT:    store i32 [[ADD5_I]], i32* [[J_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[TMP35:%.*]] = load i32, i32* [[J_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[INC_I:%.*]] = add nsw i32 [[TMP35]], 1
// CHECK6-NEXT:    store i32 [[INC_I]], i32* [[J_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[TMP36:%.*]] = load i32, i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    [[ADD6_I:%.*]] = add nsw i32 [[TMP36]], 1
// CHECK6-NEXT:    store i32 [[ADD6_I]], i32* [[DOTOMP_IV_I]], align 4, !noalias !13, !llvm.access.group !14
// CHECK6-NEXT:    br label [[OMP_INNER_FOR_COND_I]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK6:       omp.inner.for.end.i:
// CHECK6-NEXT:    [[TMP37:%.*]] = load i32, i32* [[DOTLITER__ADDR_I]], align 4, !noalias !13
// CHECK6-NEXT:    [[TMP38:%.*]] = icmp ne i32 [[TMP37]], 0
// CHECK6-NEXT:    br i1 [[TMP38]], label [[DOTOMP_LINEAR_PU_I:%.*]], label [[DOTOMP_OUTLINED__1_EXIT:%.*]]
// CHECK6:       .omp.linear.pu.i:
// CHECK6-NEXT:    [[TMP39:%.*]] = load i32, i32* [[DOTLINEAR_START1_I]], align 4, !noalias !13
// CHECK6-NEXT:    [[ADD7_I:%.*]] = add nsw i32 [[TMP39]], 10
// CHECK6-NEXT:    store i32 [[ADD7_I]], i32* [[J_I]], align 4, !noalias !13
// CHECK6-NEXT:    br label [[DOTOMP_OUTLINED__1_EXIT]]
// CHECK6:       .omp_outlined..1.exit:
// CHECK6-NEXT:    ret i32 0
//
//
// CHECK7-LABEL: define {{[^@]+}}@main
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK7-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK7-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK7-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 16
// CHECK7-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK7-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK7-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK7-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK7-NEXT:    [[T_VAR2:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S], align 16
// CHECK7-NEXT:    [[VEC4:%.*]] = alloca [2 x i32], align 4
// CHECK7-NEXT:    [[SIVAR:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]])
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]])
// CHECK7-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK7-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK7-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_BEGIN]], double 1.000000e+00)
// CHECK7-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_ELEMENT]], double 2.000000e+00)
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[VAR]], double 3.000000e+00)
// CHECK7-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK7-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK7-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK7-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK7-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[VAR1]])
// CHECK7-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK7-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK7-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK7:       arrayctor.loop:
// CHECK7-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK7-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK7-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK7-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK7-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK7:       arrayctor.cont:
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7:       omp.inner.for.cond:
// CHECK7-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
// CHECK7-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !2
// CHECK7-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV5]], [[TMP3]]
// CHECK7-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK7:       omp.inner.for.cond.cleanup:
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK7:       omp.inner.for.body:
// CHECK7-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1
// CHECK7-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR2]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 0
// CHECK7-NEXT:    store i32 [[TMP5]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i64 0, i64 0
// CHECK7-NEXT:    [[TMP6:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8*
// CHECK7-NEXT:    [[TMP7:%.*]] = bitcast %struct.S* [[VAR1]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP6]], i8* align 8 [[TMP7]], i64 8, i1 false), !llvm.access.group !2
// CHECK7-NEXT:    store i32 33, i32* [[SIVAR]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7:       omp.body.continue:
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7:       omp.inner.for.inc:
// CHECK7-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK7-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK7:       omp.inner.for.end:
// CHECK7-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK7-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[VAR]] to i8*
// CHECK7-NEXT:    [[TMP10:%.*]] = bitcast %struct.S* [[VAR1]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP9]], i8* align 8 [[TMP10]], i64 8, i1 false)
// CHECK7-NEXT:    [[TMP11:%.*]] = load i32, i32* [[T_VAR2]], align 4
// CHECK7-NEXT:    store i32 [[TMP11]], i32* [[T_VAR]], align 4
// CHECK7-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP12:%.*]] = bitcast [2 x %struct.S]* [[S_ARR3]] to %struct.S*
// CHECK7-NEXT:    [[TMP13:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN8]], i64 2
// CHECK7-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN8]], [[TMP13]]
// CHECK7-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE9:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7:       omp.arraycpy.body:
// CHECK7-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP12]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN8]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT:    [[TMP14:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK7-NEXT:    [[TMP15:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP14]], i8* align 8 [[TMP15]], i64 8, i1 false)
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP13]]
// CHECK7-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7:       omp.arraycpy.done9:
// CHECK7-NEXT:    [[TMP16:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK7-NEXT:    [[TMP17:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP16]], i8* align 4 [[TMP17]], i64 8, i1 false)
// CHECK7-NEXT:    [[TMP18:%.*]] = load i32, i32* [[SIVAR]], align 4
// CHECK7-NEXT:    store i32 [[TMP18]], i32* @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i64 2
// CHECK7-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK7:       arraydestroy.body:
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP19]], [[OMP_ARRAYCPY_DONE9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4:[0-9]+]]
// CHECK7-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
// CHECK7-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK7:       arraydestroy.done11:
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR1]]) #[[ATTR4]]
// CHECK7-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK7-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAY_BEGIN12:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN12]], i64 2
// CHECK7-NEXT:    br label [[ARRAYDESTROY_BODY13:%.*]]
// CHECK7:       arraydestroy.body13:
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENTPAST14:%.*]] = phi %struct.S* [ [[TMP20]], [[ARRAYDESTROY_DONE11]] ], [ [[ARRAYDESTROY_ELEMENT15:%.*]], [[ARRAYDESTROY_BODY13]] ]
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENT15]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST14]], i64 -1
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT15]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAYDESTROY_DONE16:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT15]], [[ARRAY_BEGIN12]]
// CHECK7-NEXT:    br i1 [[ARRAYDESTROY_DONE16]], label [[ARRAYDESTROY_DONE17:%.*]], label [[ARRAYDESTROY_BODY13]]
// CHECK7:       arraydestroy.done17:
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]]) #[[ATTR4]]
// CHECK7-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]]) #[[ATTR4]]
// CHECK7-NEXT:    [[TMP21:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK7-NEXT:    ret i32 [[TMP21]]
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ev
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    call void @_ZN1SIdEC2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]])
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ed
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK7-NEXT:    call void @_ZN1SIdEC2Ed(%struct.S* nonnull dereferenceable(8) [[THIS1]], double [[TMP0]])
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdED1Ev
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    call void @_ZN1SIdED2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK7-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK7-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK7-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK7-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK7-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK7-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK7-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK7-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK7-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// CHECK7-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
// CHECK7-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK7-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]])
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]])
// CHECK7-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK7-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK7-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK7-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[VAR]], i32 3)
// CHECK7-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK7-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK7-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK7-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK7-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK7-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK7-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK7-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK7:       arrayctor.loop:
// CHECK7-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK7-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK7-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK7-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK7:       arrayctor.cont:
// CHECK7-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR4]])
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7:       omp.inner.for.cond:
// CHECK7-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
// CHECK7-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !6
// CHECK7-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV5]], [[TMP3]]
// CHECK7-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK7:       omp.inner.for.cond.cleanup:
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK7:       omp.inner.for.body:
// CHECK7-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1
// CHECK7-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR1]], align 128, !llvm.access.group !6
// CHECK7-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i64 0, i64 0
// CHECK7-NEXT:    store i32 [[TMP5]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i64 0, i64 0
// CHECK7-NEXT:    [[TMP6:%.*]] = bitcast %struct.S.0* [[ARRAYIDX6]] to i8*
// CHECK7-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP6]], i8* align 4 [[TMP7]], i64 4, i1 false), !llvm.access.group !6
// CHECK7-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7:       omp.body.continue:
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7:       omp.inner.for.inc:
// CHECK7-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK7-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK7-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK7:       omp.inner.for.end:
// CHECK7-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK7-NEXT:    [[TMP9:%.*]] = load i32, i32* [[T_VAR1]], align 128
// CHECK7-NEXT:    store i32 [[TMP9]], i32* [[T_VAR]], align 128
// CHECK7-NEXT:    [[TMP10:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK7-NEXT:    [[TMP11:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP10]], i8* align 4 [[TMP11]], i64 8, i1 false)
// CHECK7-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP12:%.*]] = bitcast [2 x %struct.S.0]* [[S_ARR3]] to %struct.S.0*
// CHECK7-NEXT:    [[TMP13:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN8]], i64 2
// CHECK7-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN8]], [[TMP13]]
// CHECK7-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE9:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7:       omp.arraycpy.body:
// CHECK7-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN8]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT:    [[TMP14:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK7-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP14]], i8* align 4 [[TMP15]], i64 4, i1 false)
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP13]]
// CHECK7-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7:       omp.arraycpy.done9:
// CHECK7-NEXT:    [[TMP16:%.*]] = bitcast %struct.S.0* [[VAR]] to i8*
// CHECK7-NEXT:    [[TMP17:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK7-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP16]], i8* align 4 [[TMP17]], i64 4, i1 false)
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN10]], i64 2
// CHECK7-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK7:       arraydestroy.body:
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP18]], [[OMP_ARRAYCPY_DONE9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
// CHECK7-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK7:       arraydestroy.done11:
// CHECK7-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAY_BEGIN12:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN12]], i64 2
// CHECK7-NEXT:    br label [[ARRAYDESTROY_BODY13:%.*]]
// CHECK7:       arraydestroy.body13:
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENTPAST14:%.*]] = phi %struct.S.0* [ [[TMP19]], [[ARRAYDESTROY_DONE11]] ], [ [[ARRAYDESTROY_ELEMENT15:%.*]], [[ARRAYDESTROY_BODY13]] ]
// CHECK7-NEXT:    [[ARRAYDESTROY_ELEMENT15]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST14]], i64 -1
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT15]]) #[[ATTR4]]
// CHECK7-NEXT:    [[ARRAYDESTROY_DONE16:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT15]], [[ARRAY_BEGIN12]]
// CHECK7-NEXT:    br i1 [[ARRAYDESTROY_DONE16]], label [[ARRAYDESTROY_DONE17:%.*]], label [[ARRAYDESTROY_BODY13]]
// CHECK7:       arraydestroy.done17:
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK7-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]]) #[[ATTR4]]
// CHECK7-NEXT:    [[TMP20:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK7-NEXT:    ret i32 [[TMP20]]
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ev
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK7-NEXT:    store double 0.000000e+00, double* [[F]], align 8
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdED2Ev
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ed
// CHECK7-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK7-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK7-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK7-NEXT:    store double [[TMP0]], double* [[F]], align 8
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]])
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK7-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK7-NEXT:    store i32 0, i32* [[F]], align 4
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK7-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK7-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK7-NEXT:    ret void
//
//
// CHECK7-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK7-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK7-NEXT:  entry:
// CHECK7-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK7-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK7-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@main
// CHECK8-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK8-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK8-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK8-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 16
// CHECK8-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK8-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK8-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK8-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 8
// CHECK8-NEXT:    [[T_VAR2:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S], align 16
// CHECK8-NEXT:    [[VEC4:%.*]] = alloca [2 x i32], align 4
// CHECK8-NEXT:    [[SIVAR:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]])
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]])
// CHECK8-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK8-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK8-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_BEGIN]], double 1.000000e+00)
// CHECK8-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[ARRAYINIT_ELEMENT]], double 2.000000e+00)
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ed(%struct.S* nonnull dereferenceable(8) [[VAR]], double 3.000000e+00)
// CHECK8-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK8-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK8-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK8-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK8-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[VAR1]])
// CHECK8-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK8-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK8-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK8:       arrayctor.loop:
// CHECK8-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK8-NEXT:    call void @_ZN1SIdEC1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYCTOR_CUR]])
// CHECK8-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK8-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK8-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK8:       arrayctor.cont:
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK8:       omp.inner.for.cond:
// CHECK8-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
// CHECK8-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !2
// CHECK8-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV5]], [[TMP3]]
// CHECK8-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK8:       omp.inner.for.cond.cleanup:
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK8:       omp.inner.for.body:
// CHECK8-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1
// CHECK8-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK8-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR2]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC4]], i64 0, i64 0
// CHECK8-NEXT:    store i32 [[TMP5]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i64 0, i64 0
// CHECK8-NEXT:    [[TMP6:%.*]] = bitcast %struct.S* [[ARRAYIDX6]] to i8*
// CHECK8-NEXT:    [[TMP7:%.*]] = bitcast %struct.S* [[VAR1]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 [[TMP6]], i8* align 8 [[TMP7]], i64 8, i1 false), !llvm.access.group !2
// CHECK8-NEXT:    store i32 33, i32* [[SIVAR]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK8:       omp.body.continue:
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK8:       omp.inner.for.inc:
// CHECK8-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK8-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK8:       omp.inner.for.end:
// CHECK8-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK8-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[VAR]] to i8*
// CHECK8-NEXT:    [[TMP10:%.*]] = bitcast %struct.S* [[VAR1]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP9]], i8* align 8 [[TMP10]], i64 8, i1 false)
// CHECK8-NEXT:    [[TMP11:%.*]] = load i32, i32* [[T_VAR2]], align 4
// CHECK8-NEXT:    store i32 [[TMP11]], i32* [[T_VAR]], align 4
// CHECK8-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP12:%.*]] = bitcast [2 x %struct.S]* [[S_ARR3]] to %struct.S*
// CHECK8-NEXT:    [[TMP13:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN8]], i64 2
// CHECK8-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN8]], [[TMP13]]
// CHECK8-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE9:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK8:       omp.arraycpy.body:
// CHECK8-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP12]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN8]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK8-NEXT:    [[TMP14:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK8-NEXT:    [[TMP15:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP14]], i8* align 8 [[TMP15]], i64 8, i1 false)
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK8-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP13]]
// CHECK8-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_BODY]]
// CHECK8:       omp.arraycpy.done9:
// CHECK8-NEXT:    [[TMP16:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK8-NEXT:    [[TMP17:%.*]] = bitcast [2 x i32]* [[VEC4]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP16]], i8* align 4 [[TMP17]], i64 8, i1 false)
// CHECK8-NEXT:    [[TMP18:%.*]] = load i32, i32* [[SIVAR]], align 4
// CHECK8-NEXT:    store i32 [[TMP18]], i32* @_ZZ4mainE5sivar, align 4
// CHECK8-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR3]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN10]], i64 2
// CHECK8-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK8:       arraydestroy.body:
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP19]], [[OMP_ARRAYCPY_DONE9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4:[0-9]+]]
// CHECK8-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
// CHECK8-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK8:       arraydestroy.done11:
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR1]]) #[[ATTR4]]
// CHECK8-NEXT:    [[CALL:%.*]] = call i32 @_Z5tmainIiET_v()
// CHECK8-NEXT:    store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[VAR]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAY_BEGIN12:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN12]], i64 2
// CHECK8-NEXT:    br label [[ARRAYDESTROY_BODY13:%.*]]
// CHECK8:       arraydestroy.body13:
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENTPAST14:%.*]] = phi %struct.S* [ [[TMP20]], [[ARRAYDESTROY_DONE11]] ], [ [[ARRAYDESTROY_ELEMENT15:%.*]], [[ARRAYDESTROY_BODY13]] ]
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENT15]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST14]], i64 -1
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[ARRAYDESTROY_ELEMENT15]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAYDESTROY_DONE16:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT15]], [[ARRAY_BEGIN12]]
// CHECK8-NEXT:    br i1 [[ARRAYDESTROY_DONE16]], label [[ARRAYDESTROY_DONE17:%.*]], label [[ARRAYDESTROY_BODY13]]
// CHECK8:       arraydestroy.done17:
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TEST]]) #[[ATTR4]]
// CHECK8-NEXT:    call void @_ZN1SIdED1Ev(%struct.S* nonnull dereferenceable(8) [[TTT]]) #[[ATTR4]]
// CHECK8-NEXT:    [[TMP21:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK8-NEXT:    ret i32 [[TMP21]]
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ev
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    call void @_ZN1SIdEC2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]])
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdEC1Ed
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK8-NEXT:    call void @_ZN1SIdEC2Ed(%struct.S* nonnull dereferenceable(8) [[THIS1]], double [[TMP0]])
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdED1Ev
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    call void @_ZN1SIdED2Ev(%struct.S* nonnull dereferenceable(8) [[THIS1]]) #[[ATTR4]]
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK8-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[TTT:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK8-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK8-NEXT:    [[T_VAR:%.*]] = alloca i32, align 128
// CHECK8-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK8-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK8-NEXT:    [[VAR:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK8-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK8-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK8-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 128
// CHECK8-NEXT:    [[VEC2:%.*]] = alloca [2 x i32], align 4
// CHECK8-NEXT:    [[S_ARR3:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK8-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]])
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]])
// CHECK8-NEXT:    store i32 0, i32* [[T_VAR]], align 128
// CHECK8-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK8-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK8-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S.0* nonnull dereferenceable(4) [[VAR]], i32 3)
// CHECK8-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK8-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK8-NEXT:    [[TMP1:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK8-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK8-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK8-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK8-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK8-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK8:       arrayctor.loop:
// CHECK8-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK8-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK8-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK8-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK8:       arrayctor.cont:
// CHECK8-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR4]])
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK8:       omp.inner.for.cond:
// CHECK8-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
// CHECK8-NEXT:    [[TMP3:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !6
// CHECK8-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV5]], [[TMP3]]
// CHECK8-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK8:       omp.inner.for.cond.cleanup:
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK8:       omp.inner.for.body:
// CHECK8-NEXT:    [[TMP4:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1
// CHECK8-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK8-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    [[TMP5:%.*]] = load i32, i32* [[T_VAR1]], align 128, !llvm.access.group !6
// CHECK8-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC2]], i64 0, i64 0
// CHECK8-NEXT:    store i32 [[TMP5]], i32* [[ARRAYIDX]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i64 0, i64 0
// CHECK8-NEXT:    [[TMP6:%.*]] = bitcast %struct.S.0* [[ARRAYIDX6]] to i8*
// CHECK8-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP6]], i8* align 4 [[TMP7]], i64 4, i1 false), !llvm.access.group !6
// CHECK8-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK8:       omp.body.continue:
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK8:       omp.inner.for.inc:
// CHECK8-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK8-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !6
// CHECK8-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK8:       omp.inner.for.end:
// CHECK8-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK8-NEXT:    [[TMP9:%.*]] = load i32, i32* [[T_VAR1]], align 128
// CHECK8-NEXT:    store i32 [[TMP9]], i32* [[T_VAR]], align 128
// CHECK8-NEXT:    [[TMP10:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK8-NEXT:    [[TMP11:%.*]] = bitcast [2 x i32]* [[VEC2]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP10]], i8* align 4 [[TMP11]], i64 8, i1 false)
// CHECK8-NEXT:    [[ARRAY_BEGIN8:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP12:%.*]] = bitcast [2 x %struct.S.0]* [[S_ARR3]] to %struct.S.0*
// CHECK8-NEXT:    [[TMP13:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN8]], i64 2
// CHECK8-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN8]], [[TMP13]]
// CHECK8-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE9:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK8:       omp.arraycpy.body:
// CHECK8-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN8]], [[OMP_INNER_FOR_END]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK8-NEXT:    [[TMP14:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK8-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP14]], i8* align 4 [[TMP15]], i64 4, i1 false)
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK8-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK8-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP13]]
// CHECK8-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_BODY]]
// CHECK8:       omp.arraycpy.done9:
// CHECK8-NEXT:    [[TMP16:%.*]] = bitcast %struct.S.0* [[VAR]] to i8*
// CHECK8-NEXT:    [[TMP17:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK8-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP16]], i8* align 4 [[TMP17]], i64 4, i1 false)
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR4]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAY_BEGIN10:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR3]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN10]], i64 2
// CHECK8-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK8:       arraydestroy.body:
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP18]], [[OMP_ARRAYCPY_DONE9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN10]]
// CHECK8-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE11:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK8:       arraydestroy.done11:
// CHECK8-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAY_BEGIN12:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN12]], i64 2
// CHECK8-NEXT:    br label [[ARRAYDESTROY_BODY13:%.*]]
// CHECK8:       arraydestroy.body13:
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENTPAST14:%.*]] = phi %struct.S.0* [ [[TMP19]], [[ARRAYDESTROY_DONE11]] ], [ [[ARRAYDESTROY_ELEMENT15:%.*]], [[ARRAYDESTROY_BODY13]] ]
// CHECK8-NEXT:    [[ARRAYDESTROY_ELEMENT15]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST14]], i64 -1
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[ARRAYDESTROY_ELEMENT15]]) #[[ATTR4]]
// CHECK8-NEXT:    [[ARRAYDESTROY_DONE16:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT15]], [[ARRAY_BEGIN12]]
// CHECK8-NEXT:    br i1 [[ARRAYDESTROY_DONE16]], label [[ARRAYDESTROY_DONE17:%.*]], label [[ARRAYDESTROY_BODY13]]
// CHECK8:       arraydestroy.done17:
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK8-NEXT:    call void @_ZN1SIiED1Ev(%struct.S.0* nonnull dereferenceable(4) [[TTT]]) #[[ATTR4]]
// CHECK8-NEXT:    [[TMP20:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK8-NEXT:    ret i32 [[TMP20]]
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ev
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK8-NEXT:    store double 0.000000e+00, double* [[F]], align 8
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdED2Ev
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIdEC2Ed
// CHECK8-SAME: (%struct.S* nonnull dereferenceable(8) [[THIS:%.*]], double [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK8-NEXT:    [[A_ADDR:%.*]] = alloca double, align 8
// CHECK8-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    store double [[A]], double* [[A_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP0:%.*]] = load double, double* [[A_ADDR]], align 8
// CHECK8-NEXT:    store double [[TMP0]], double* [[F]], align 8
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]])
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK8-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S.0* nonnull dereferenceable(4) [[THIS1]], i32 [[TMP0]])
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    call void @_ZN1SIiED2Ev(%struct.S.0* nonnull dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK8-NEXT:    store i32 0, i32* [[F]], align 4
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK8-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK8-NEXT:    store i32 [[TMP0]], i32* [[F]], align 4
// CHECK8-NEXT:    ret void
//
//
// CHECK8-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK8-SAME: (%struct.S.0* nonnull dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK8-NEXT:  entry:
// CHECK8-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK8-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK8-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK9-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(%class.anon* nonnull dereferenceable(1) [[REF_TMP]])
// CHECK9-NEXT:    ret i32 0
//
//
// CHECK10-LABEL: define {{[^@]+}}@main
// CHECK10-SAME: () #[[ATTR1:[0-9]+]] {
// CHECK10-NEXT:  entry:
// CHECK10-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK10-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT:    [[TMP0:%.*]] = load i8*, i8** getelementptr inbounds ([[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to %struct.__block_literal_generic*), i32 0, i32 3), align 8
// CHECK10-NEXT:    [[TMP1:%.*]] = bitcast i8* [[TMP0]] to void (i8*)*
// CHECK10-NEXT:    call void [[TMP1]](i8* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor* }* @__block_literal_global to i8*))
// CHECK10-NEXT:    ret i32 0
//
//
// CHECK10-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK10-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK10-NEXT:  entry:
// CHECK10-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK10-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*, align 8
// CHECK10-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK10-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK10-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT:    [[G:%.*]] = alloca double, align 8
// CHECK10-NEXT:    [[SIVAR:%.*]] = alloca i32, align 4
// CHECK10-NEXT:    [[BLOCK2:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, align 8
// CHECK10-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK10-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>*
// CHECK10-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor* }>** [[BLOCK_ADDR]], align 8
// CHECK10-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK10-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK10-NEXT:    [[TMP0:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK10-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10:       omp.inner.for.cond:
// CHECK10-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    [[CONV1:%.*]] = sext i32 [[TMP1]] to i64
// CHECK10-NEXT:    [[TMP2:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV1]], [[TMP2]]
// CHECK10-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10:       omp.inner.for.body:
// CHECK10-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK10-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    store double 1.000000e+00, double* [[G]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    store i32 11, i32* [[SIVAR]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 0
// CHECK10-NEXT:    store i8* bitcast (i8** @_NSConcreteStackBlock to i8*), i8** [[BLOCK_ISA]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 1
// CHECK10-NEXT:    store i32 1073741824, i32* [[BLOCK_FLAGS]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 2
// CHECK10-NEXT:    store i32 0, i32* [[BLOCK_RESERVED]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 3
// CHECK10-NEXT:    store i8* bitcast (void (i8*)* @__main_block_invoke_2 to i8*), i8** [[BLOCK_INVOKE]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 4
// CHECK10-NEXT:    store %struct.__block_descriptor* bitcast ({ i64, i64, i8*, i8* }* @__block_descriptor_tmp.1 to %struct.__block_descriptor*), %struct.__block_descriptor** [[BLOCK_DESCRIPTOR]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 5
// CHECK10-NEXT:    [[TMP4:%.*]] = load volatile double, double* [[G]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    store volatile double [[TMP4]], double* [[BLOCK_CAPTURED]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[BLOCK_CAPTURED3:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]], i32 0, i32 6
// CHECK10-NEXT:    [[TMP5:%.*]] = load i32, i32* [[SIVAR]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    store i32 [[TMP5]], i32* [[BLOCK_CAPTURED3]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[TMP6:%.*]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK2]] to void ()*
// CHECK10-NEXT:    [[BLOCK_LITERAL:%.*]] = bitcast void ()* [[TMP6]] to %struct.__block_literal_generic*
// CHECK10-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], %struct.__block_literal_generic* [[BLOCK_LITERAL]], i32 0, i32 3
// CHECK10-NEXT:    [[TMP8:%.*]] = bitcast %struct.__block_literal_generic* [[BLOCK_LITERAL]] to i8*
// CHECK10-NEXT:    [[TMP9:%.*]] = load i8*, i8** [[TMP7]], align 8, !llvm.access.group !2
// CHECK10-NEXT:    [[TMP10:%.*]] = bitcast i8* [[TMP9]] to void (i8*)*
// CHECK10-NEXT:    call void [[TMP10]](i8* [[TMP8]]), !llvm.access.group !2
// CHECK10-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10:       omp.body.continue:
// CHECK10-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10:       omp.inner.for.inc:
// CHECK10-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK10-NEXT:    store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK10-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK10:       omp.inner.for.end:
// CHECK10-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK10-NEXT:    [[TMP12:%.*]] = load double, double* [[G]], align 8
// CHECK10-NEXT:    store volatile double [[TMP12]], double* @g, align 8
// CHECK10-NEXT:    [[TMP13:%.*]] = load i32, i32* [[SIVAR]], align 4
// CHECK10-NEXT:    store i32 [[TMP13]], i32* @_ZZ4mainE5sivar, align 4
// CHECK10-NEXT:    ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@__main_block_invoke_2
// CHECK10-SAME: (i8* [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK10-NEXT:  entry:
// CHECK10-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca i8*, align 8
// CHECK10-NEXT:    [[BLOCK_ADDR:%.*]] = alloca <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>*, align 8
// CHECK10-NEXT:    store i8* [[DOTBLOCK_DESCRIPTOR]], i8** [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK10-NEXT:    [[BLOCK:%.*]] = bitcast i8* [[DOTBLOCK_DESCRIPTOR]] to <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>*
// CHECK10-NEXT:    store <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>** [[BLOCK_ADDR]], align 8
// CHECK10-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], i32 0, i32 5
// CHECK10-NEXT:    store double 2.000000e+00, double* [[BLOCK_CAPTURE_ADDR]], align 8
// CHECK10-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>, <{ i8*, i32, i32, i8*, %struct.__block_descriptor*, double, i32 }>* [[BLOCK]], i32 0, i32 6
// CHECK10-NEXT:    store i32 22, i32* [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK10-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_Z10array_funciPfP2St
// CHECK11-SAME: (i32 [[N:%.*]], float* [[A:%.*]], %struct.St* [[S:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca float*, align 8
// CHECK11-NEXT:    [[S_ADDR:%.*]] = alloca %struct.St*, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A1:%.*]] = alloca float*, align 8
// CHECK11-NEXT:    [[S2:%.*]] = alloca %struct.St*, align 8
// CHECK11-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
// CHECK11-NEXT:    store float* [[A]], float** [[A_ADDR]], align 8
// CHECK11-NEXT:    store %struct.St* [[S]], %struct.St** [[S_ADDR]], align 8
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK11-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK11-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    [[TMP2:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK11-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK11-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK11-NEXT:    [[CONV3:%.*]] = sext i32 [[TMP3]] to i64
// CHECK11-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !2
// CHECK11-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV3]], [[TMP4]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK11-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK11-NEXT:    store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load float*, float** [[A1]], align 8
// CHECK11-NEXT:    store float* [[TMP7]], float** [[A_ADDR]], align 8
// CHECK11-NEXT:    [[TMP8:%.*]] = load %struct.St*, %struct.St** [[S2]], align 8
// CHECK11-NEXT:    store %struct.St* [[TMP8]], %struct.St** [[S_ADDR]], align 8
// CHECK11-NEXT:    ret void
//
//
// CHECK12-LABEL: define {{[^@]+}}@_Z4loopv
// CHECK12-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK12-NEXT:  entry:
// CHECK12-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK12-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[DOTLINEAR_START:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[DOTLINEAR_START1:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[I2:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[I3:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    [[J4:%.*]] = alloca i32, align 4
// CHECK12-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK12-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK12-NEXT:    [[TMP0:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK12-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK12-NEXT:    store i32 [[CONV]], i32* [[DOTOMP_IV]], align 4
// CHECK12-NEXT:    [[TMP1:%.*]] = load i32, i32* [[I]], align 4
// CHECK12-NEXT:    store i32 [[TMP1]], i32* [[DOTLINEAR_START]], align 4
// CHECK12-NEXT:    [[TMP2:%.*]] = load i32, i32* [[J]], align 4
// CHECK12-NEXT:    store i32 [[TMP2]], i32* [[DOTLINEAR_START1]], align 4
// CHECK12-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK12:       omp.inner.for.cond:
// CHECK12-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[CONV5:%.*]] = sext i32 [[TMP3]] to i64
// CHECK12-NEXT:    [[TMP4:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8, !llvm.access.group !2
// CHECK12-NEXT:    [[CMP:%.*]] = icmp ule i64 [[CONV5]], [[TMP4]]
// CHECK12-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK12:       omp.inner.for.body:
// CHECK12-NEXT:    [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1
// CHECK12-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK12-NEXT:    store i32 [[ADD]], i32* [[I2]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[TMP6:%.*]] = load i32, i32* [[DOTLINEAR_START1]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK12-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP6]], [[MUL6]]
// CHECK12-NEXT:    store i32 [[ADD7]], i32* [[J4]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[TMP8:%.*]] = load i32, i32* [[J4]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK12-NEXT:    store i32 [[INC]], i32* [[J4]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK12:       omp.body.continue:
// CHECK12-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK12:       omp.inner.for.inc:
// CHECK12-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK12-NEXT:    store i32 [[ADD8]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK12-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK12:       omp.inner.for.end:
// CHECK12-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK12-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTLINEAR_START1]], align 4
// CHECK12-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP10]], 10
// CHECK12-NEXT:    store i32 [[ADD9]], i32* [[J]], align 4
// CHECK12-NEXT:    ret void
//