// 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-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// 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=CHECK4
// expected-no-diagnostics
#ifndef HEADER
#define HEADER

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

volatile double g, g_orig;
volatile double &g1 = g_orig;

struct BaseS {
  int x;
};
struct BaseS1 {
  float y;
};

template <class T>
struct S : public BaseS, public BaseS1 {
  T f;
  S(T a) : f(a + g) {}
  S() : f(g) {}
  S& operator=(const S&);
  ~S() {}
};
void red(BaseS1&, const BaseS1&);
void red_plus(BaseS1&, const BaseS1&);
void init(BaseS1&, const BaseS1&);
void init1(BaseS1&, const BaseS1&);
void init2(BaseS1&, const BaseS1&);
void init_plus(BaseS1&, const BaseS1&);
#pragma omp declare reduction(operator& : BaseS1 : red(omp_out, omp_in)) initializer(init(omp_priv, omp_orig))
#pragma omp declare reduction(+ : BaseS1 : red_plus(omp_out, omp_in)) initializer(init_plus(omp_priv, omp_orig))
#pragma omp declare reduction(&& : S<float>, S<int> : omp_out.f *= omp_in.f) initializer(init1(omp_priv, omp_orig))


#pragma omp declare reduction(operator* : S<int> : omp_out.f = 17 * omp_in.f) initializer(omp_priv = S<int>())
void bazz() {
  S<int> s;
#pragma omp parallel
#pragma omp simd reduction(*: s)
  for (int I = 0; I < 10; ++I)
    ;
}


#pragma omp declare reduction(operator&& : int : omp_out = 111 & omp_in)
template <typename T, int length>
T tmain() {
  T t;
  S<T> test;
  T t_var = T(), t_var1;
  T vec[] = {1, 2};
  S<T> s_arr[] = {1, 2};
  S<T> &var = test;
  S<T> var1;
  S<T> arr[length];
#pragma omp declare reduction(operator& : T : omp_out = 15 + omp_in)
#pragma omp declare reduction(operator+ : T : omp_out = 1513 + omp_in) initializer(omp_priv = 321)
#pragma omp declare reduction(min : T : omp_out = 47 - omp_in) initializer(omp_priv = 432 / omp_orig)
#pragma omp declare reduction(operator&& : S<T> : omp_out.f = 17 * omp_in.f) initializer(init2(omp_priv, omp_orig))
#pragma omp declare reduction(operator&& : T : omp_out = 17 * omp_in)
#pragma omp parallel
#pragma omp for reduction(+ : t_var) reduction(& : var) reduction(&& : var1) reduction(min : t_var1) nowait
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
#pragma omp parallel
#pragma omp for reduction(&& : t_var)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
#pragma omp parallel
#pragma omp for reduction(+ : arr[1:length-2])
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
  return T();
}

extern S<float> **foo();

#pragma omp declare reduction(operator- : float, double : omp_out = 333 + omp_in)
#pragma omp declare reduction(min : float, double : omp_out = 555 + omp_in)
int main() {
#pragma omp declare reduction(operator+ : float, double : omp_out = 222 - omp_in) initializer(omp_priv = -1)
  S<float> test;
  float t_var = 0, t_var1;
  int vec[] = {1, 2};
  S<float> s_arr[] = {1, 2, 3, 4};
  S<float> &var = test;
  S<float> var1, arrs[10][4];
  S<float> **var2 = foo();
  S<float> vvar2[5];
  S<float>(&var3)[4] = s_arr;
#pragma omp declare reduction(operator+ : int : omp_out = 555 * omp_in) initializer(omp_priv = 888)
#pragma omp parallel
#pragma omp for reduction(+ : t_var) reduction(& : var) reduction(&& : var1) reduction(min : t_var1)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
  int arr[10][vec[1]];
#pragma omp parallel for reduction(+ : arr[1][ : vec[1]]) reduction(& : arrs[1 : vec[1]][1 : 2])
  for (int i = 0; i < 10; ++i)
    ++arr[1][i];
#pragma omp parallel
#pragma omp for reduction(+ : arr) reduction(& : arrs)
  for (int i = 0; i < 10; ++i)
    ++arr[1][i];
#pragma omp parallel
#pragma omp for reduction(& : var2[0 : 5][1 : 6])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : vvar2[0 : 5])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var3[1 : 2])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var3) allocate(omp_cgroup_mem_alloc: var3)
  for (int i = 0; i < 10; ++i)
    ;
  return tmain<int, 42>();
}


// Reduction list for runtime.


// For + reduction operation initial value of private variable is -1.

// For & reduction operation initial value of private variable is defined by call of 'init()' function.

// For && reduction operation initial value of private variable is 1.0.

// For min reduction operation initial value of private variable is largest repesentable value.

// Skip checks for internal operations.

// void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;


// void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// t_var_lhs = (float*)lhs[0];
// t_var_rhs = (float*)rhs[0];

// var_lhs = (S<float>*)lhs[1];
// var_rhs = (S<float>*)rhs[1];

// var1_lhs = (S<float>*)lhs[2];
// var1_rhs = (S<float>*)rhs[2];

// t_var1_lhs = (float*)lhs[3];
// t_var1_rhs = (float*)rhs[3];

// t_var_lhs += t_var_rhs;

// var_lhs = var_lhs.operator &(var_rhs);

// var1_lhs = var1_lhs.operator &&(var1_rhs);

// t_var1_lhs = min(t_var1_lhs, t_var1_rhs);


// Reduction list for runtime.


// Check initialization of private copy.


// Check initialization of private copy.

// Skip checks for internal operations.

// void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// break;

// Check destruction of private copy.


// void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// arr_rhs = (int*)rhs[0];
// arr_lhs = (int*)lhs[0];

// arr_size = (size_t)lhs[1];

// arrs_rhs = (S<float>*)rhs[2];
// arrs_lhs = (S<float>*)lhs[2];

// arrs_size = (size_t)lhs[3];

// arr_lhs[:] += arr_rhs[:];

// arrs_lhs = arrs_lhs.operator &(arrs_rhs);


// Reduction list for runtime.


// Check initialization of private copy.

// Check initialization of private copy.


// Skip checks for internal operations.

// void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// break;

// Check destruction of private copy.


// void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// arr_rhs = (int*)rhs[0];
// arr_lhs = (int*)lhs[0];

// arr_size = (size_t)lhs[1];

// arrs_rhs = (S<float>*)rhs[2];
// arrs_lhs = (S<float>*)lhs[2];

// arr_lhs[:] += arr_rhs[:];

// arrs_lhs = arrs_lhs.operator &(arrs_rhs);


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Not interested in this one:

// Reduction list for runtime.


// For + reduction operation initial value of private variable is 0.

// For & reduction operation initial value of private variable is ones in all bits.

// For && reduction operation initial value of private variable is 1.0.

// For min reduction operation initial value of private variable is largest repesentable value.

// Skip checks for internal operations.

// void *RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce_nowait(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce_nowait(<loc>, <gtid>, &<lock>);

// break;

// case 2:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// break;


// void reduce_func(void *lhs[<n>], void *rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// t_var_lhs = (i{{[0-9]+}}*)lhs[0];
// t_var_rhs = (i{{[0-9]+}}*)rhs[0];

// var_lhs = (S<i{{[0-9]+}}>*)lhs[1];
// var_rhs = (S<i{{[0-9]+}}>*)rhs[1];

// var1_lhs = (S<i{{[0-9]+}}>*)lhs[2];
// var1_rhs = (S<i{{[0-9]+}}>*)rhs[2];

// t_var1_lhs = (i{{[0-9]+}}*)lhs[3];
// t_var1_rhs = (i{{[0-9]+}}*)rhs[3];

// t_var_lhs += t_var_rhs;

// var_lhs = var_lhs.operator &(var_rhs);

// var1_lhs = var1_lhs.operator &&(var1_rhs);

// t_var1_lhs = min(t_var1_lhs, t_var1_rhs);


// Reduction list for runtime.


#endif

// CHECK1-LABEL: define {{[^@]+}}@_Z4bazzv
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]])
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.S* [[S]])
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIiEC2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[S:%.*]]) #[[ATTR2:[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:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[S1:%.*]] = alloca [[STRUCT_S:%.*]], 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 %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    call void @.omp_initializer.(%struct.S* [[S1]], %struct.S* [[TMP0]])
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK1-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP1]], 10
// CHECK1-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP2]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !3
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK1-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP3]], 1
// CHECK1-NEXT:    store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK1-NEXT:    call void @.omp_combiner.(%struct.S* [[TMP0]], %struct.S* [[S1]])
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner.
// CHECK1-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer.
// CHECK1-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TMP3]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIiED2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK1-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK1-NEXT:    [[CONV:%.*]] = fptosi double [[TMP4]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], i32* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [4 x %struct.S.0], align 16
// CHECK1-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[ARRS:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK1-NEXT:    [[VAR2:%.*]] = alloca %struct.S.0**, align 8
// CHECK1-NEXT:    [[VVAR2:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK1-NEXT:    [[VAR3:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK1-NEXT:    store float 0.000000e+00, float* [[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 [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT1]], float 3.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT1]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT2]], float 4.000000e+00)
// CHECK1-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// 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 @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[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:    [[CALL:%.*]] = call %struct.S.0** @_Z3foov()
// CHECK1-NEXT:    store %struct.S.0** [[CALL]], %struct.S.0*** [[VAR2]], align 8
// CHECK1-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN3]], i64 5
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP5:%.*]]
// CHECK1:       arrayctor.loop5:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR6:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYCTOR_NEXT7:%.*]], [[ARRAYCTOR_LOOP5]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR6]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT7]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR6]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE8:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT7]], [[ARRAYCTOR_END4]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE8]], label [[ARRAYCTOR_CONT9:%.*]], label [[ARRAYCTOR_LOOP5]]
// CHECK1:       arrayctor.cont9:
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, %struct.S.0*, %struct.S.0*, float*, [2 x i32]*, [4 x %struct.S.0]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float* [[T_VAR]], %struct.S.0* [[TMP1]], %struct.S.0* [[VAR1]], float* [[T_VAR1]], [2 x i32]* [[VEC]], [4 x %struct.S.0]* [[S_ARR]])
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK1-NEXT:    [[TMP4:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT:    store i8* [[TMP4]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = mul nuw i64 10, [[TMP3]]
// CHECK1-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 16
// CHECK1-NEXT:    store i64 [[TMP3]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, [2 x i32]*, [10 x [4 x %struct.S.0]]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 10, i64 [[TMP3]], i32* [[VLA]], [2 x i32]* [[VEC]], [10 x [4 x %struct.S.0]]* [[ARRS]])
// 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*, i64, i64, i32*, [10 x [4 x %struct.S.0]]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 10, i64 [[TMP3]], i32* [[VLA]], [10 x [4 x %struct.S.0]]* [[ARRS]])
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S.0***)* @.omp_outlined..15 to void (i32*, i32*, ...)*), %struct.S.0*** [[VAR2]])
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [5 x %struct.S.0]*)* @.omp_outlined..17 to void (i32*, i32*, ...)*), [5 x %struct.S.0]* [[VVAR2]])
// CHECK1-NEXT:    [[TMP6:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [4 x %struct.S.0]*)* @.omp_outlined..19 to void (i32*, i32*, ...)*), [4 x %struct.S.0]* [[TMP6]])
// CHECK1-NEXT:    [[TMP7:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [4 x %struct.S.0]*)* @.omp_outlined..21 to void (i32*, i32*, ...)*), [4 x %struct.S.0]* [[TMP7]])
// CHECK1-NEXT:    [[CALL10:%.*]] = call i32 @_Z5tmainIiLi42EET_v()
// CHECK1-NEXT:    store i32 [[CALL10]], i32* [[RETVAL]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore(i8* [[TMP8]])
// CHECK1-NEXT:    [[ARRAY_BEGIN11:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN11]], i64 5
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP9]], [[ARRAYCTOR_CONT9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN11]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE12:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done12:
// CHECK1-NEXT:    [[ARRAY_BEGIN13:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN13]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY14:%.*]]
// CHECK1:       arraydestroy.body14:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST15:%.*]] = phi %struct.S.0* [ [[TMP10]], [[ARRAYDESTROY_DONE12]] ], [ [[ARRAYDESTROY_ELEMENT16:%.*]], [[ARRAYDESTROY_BODY14]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT16]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST15]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT16]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE17:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT16]], [[ARRAY_BEGIN13]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE17]], label [[ARRAYDESTROY_DONE18:%.*]], label [[ARRAYDESTROY_BODY14]]
// CHECK1:       arraydestroy.done18:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN19:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN19]], i64 4
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY20:%.*]]
// CHECK1:       arraydestroy.body20:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST21:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ARRAYDESTROY_DONE18]] ], [ [[ARRAYDESTROY_ELEMENT22:%.*]], [[ARRAYDESTROY_BODY20]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT22]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST21]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT22]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE23:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT22]], [[ARRAY_BEGIN19]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE23]], label [[ARRAYDESTROY_DONE24:%.*]], label [[ARRAYDESTROY_BODY20]]
// CHECK1:       arraydestroy.done24:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP12]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIfEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]], float [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(12) [[VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[S_ARR:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[VAR1_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[VAR17:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[T_VAR18:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[VAR1]], %struct.S.0** [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    call void @.omp_initializer..3(float* [[T_VAR3]], float* [[TMP0]])
// CHECK1-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP8]], i64 4
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 4
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP9]], %struct.BaseS1* [[TMP11]])
// CHECK1-NEXT:    store %struct.S.0* [[VAR4]], %struct.S.0** [[_TMP6]], align 8
// CHECK1-NEXT:    call void @.omp_initializer..7(%struct.S.0* [[VAR17]], %struct.S.0* [[TMP2]])
// CHECK1-NEXT:    [[TMP12:%.*]] = load float, float* @.init, align 4
// CHECK1-NEXT:    store float [[TMP12]], float* [[T_VAR18]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP14]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP15]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT:    br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load float, float* [[T_VAR3]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = fptosi float [[TMP21]] to i32
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[CONV]], i32* [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP23:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP6]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP24]] to i64
// CHECK1-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP5]], i64 0, i64 [[IDXPROM10]]
// CHECK1-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S.0* @_ZN1SIfEaSERKS0_(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYIDX11]], %struct.S.0* nonnull align 4 dereferenceable(12) [[TMP23]])
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK1-NEXT:    store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP14]])
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP27:%.*]] = bitcast float* [[T_VAR3]] to i8*
// CHECK1-NEXT:    store i8* [[TMP27]], i8** [[TMP26]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK1-NEXT:    store i8* [[TMP29]], i8** [[TMP28]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP31:%.*]] = bitcast %struct.S.0* [[VAR17]] to i8*
// CHECK1-NEXT:    store i8* [[TMP31]], i8** [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast float* [[T_VAR18]] to i8*
// CHECK1-NEXT:    store i8* [[TMP33]], i8** [[TMP32]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP35:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP14]], i32 4, i64 32, i8* [[TMP34]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP35]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    call void @.omp_combiner..2(float* [[TMP0]], float* [[T_VAR3]])
// CHECK1-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP36]], i64 4
// CHECK1-NEXT:    [[TMP37:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP38:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 4
// CHECK1-NEXT:    [[TMP39:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP37]], %struct.BaseS1* [[TMP39]])
// CHECK1-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP2]], %struct.S.0* [[VAR17]])
// CHECK1-NEXT:    call void @.omp_combiner..8(float* [[TMP3]], float* [[T_VAR18]])
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..2(float* [[TMP0]], float* [[T_VAR3]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP40:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP40]], i64 4
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP42:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR16:%.*]] = getelementptr inbounds i8, i8* [[TMP42]], i64 4
// CHECK1-NEXT:    [[TMP43:%.*]] = bitcast i8* [[ADD_PTR16]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP41]], %struct.BaseS1* [[TMP43]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP2]], %struct.S.0* [[VAR17]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..8(float* [[TMP3]], float* [[T_VAR18]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR17]]) #[[ATTR4]]
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR4]]) #[[ATTR4]]
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP14]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..2
// CHECK1-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK1-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load float, float* [[TMP2]], align 4
// CHECK1-NEXT:    [[SUB:%.*]] = fsub float 2.220000e+02, [[TMP4]]
// CHECK1-NEXT:    store float [[SUB]], float* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..3
// CHECK1-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK1-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK1-NEXT:    store float -1.000000e+00, float* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..4
// CHECK1-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    call void @_Z3redR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..5
// CHECK1-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    call void @_Z4initR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..6
// CHECK1-SAME: (%struct.S.0* noalias [[TMP0:%.*]], %struct.S.0* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP2]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load float, float* [[F]], align 4
// CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP5:%.*]] = load float, float* [[F2]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = fmul float [[TMP5]], [[TMP4]]
// CHECK1-NEXT:    store float [[MUL]], float* [[F2]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..7
// CHECK1-SAME: (%struct.S.0* noalias [[TMP0:%.*]], %struct.S.0* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP6:%.*]] = bitcast %struct.S.0* [[TMP2]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 4
// CHECK1-NEXT:    [[TMP7:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @_Z5init1R6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP5]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP7]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to float*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to float*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i8* [[TMP25]] to float*
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast i8* [[TMP28]] to float*
// CHECK1-NEXT:    call void @.omp_combiner..2(float* [[TMP11]], float* [[TMP8]])
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[TMP17]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK1-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast %struct.S.0* [[TMP14]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP31]], %struct.BaseS1* [[TMP33]])
// CHECK1-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP23]], %struct.S.0* [[TMP20]])
// CHECK1-NEXT:    call void @.omp_combiner..8(float* [[TMP29]], float* [[TMP26]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..8
// CHECK1-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK1-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK1-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load float, float* [[TMP2]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = fadd float 5.550000e+02, [[TMP4]]
// CHECK1-NEXT:    store float [[ADD]], float* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIfED2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [10 x [4 x %struct.S.0]]* nonnull align 4 dereferenceable(480) [[ARRS:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[ARRS_ADDR:%.*]] = alloca [10 x [4 x %struct.S.0]]*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA1]], i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    store i32* [[ARR]], i32** [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store [10 x [4 x %struct.S.0]]* [[ARRS]], [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load [10 x [4 x %struct.S.0]]*, [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[TMP5]]
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 0
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[ARRAYIDX4]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP7]]
// CHECK1-NEXT:    [[TMP8:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[TMP8]]
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX5]], i64 [[LB_ADD_LEN]]
// CHECK1-NEXT:    [[TMP9:%.*]] = ptrtoint i32* [[ARRAYIDX6]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint i32* [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = sub i64 [[TMP9]], [[TMP10]]
// CHECK1-NEXT:    [[TMP12:%.*]] = sdiv exact i64 [[TMP11]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP13:%.*]] = add nuw i64 [[TMP12]], 1
// CHECK1-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP15:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT:    store i8* [[TMP15]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA7:%.*]] = alloca i32, i64 [[TMP13]], align 16
// CHECK1-NEXT:    store i64 [[TMP13]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr i32, i32* [[VLA7]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[VLA7]], [[TMP16]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[ARRAYIDX3]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[VLA7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @.omp_initializer..11(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP16]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP17:%.*]] = ptrtoint i32* [[TMP2]] to i64
// CHECK1-NEXT:    [[TMP18:%.*]] = ptrtoint i32* [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK1-NEXT:    [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr i32, i32* [[VLA7]], i64 [[TMP20]]
// CHECK1-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[TMP4]], i64 0, i64 1
// CHECK1-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[ARRAYIDX9]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAYDECAY]], i64 1
// CHECK1-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX11]], align 4
// CHECK1-NEXT:    [[TMP23:%.*]] = sext i32 [[TMP22]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN12:%.*]] = add nsw i64 0, [[TMP23]]
// CHECK1-NEXT:    [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[TMP4]], i64 0, i64 [[LB_ADD_LEN12]]
// CHECK1-NEXT:    [[ARRAYDECAY14:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[ARRAYIDX13]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDECAY14]], i64 2
// CHECK1-NEXT:    [[TMP24:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX15]] to i64
// CHECK1-NEXT:    [[TMP25:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX10]] to i64
// CHECK1-NEXT:    [[TMP26:%.*]] = sub i64 [[TMP24]], [[TMP25]]
// CHECK1-NEXT:    [[TMP27:%.*]] = sdiv exact i64 [[TMP26]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP28:%.*]] = add nuw i64 [[TMP27]], 1
// CHECK1-NEXT:    [[TMP29:%.*]] = mul nuw i64 [[TMP28]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[VLA16:%.*]] = alloca [[STRUCT_S_0]], i64 [[TMP28]], align 16
// CHECK1-NEXT:    store i64 [[TMP28]], i64* [[__VLA_EXPR1]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[VLA16]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY17]], label [[OMP_ARRAYINIT_DONE25:%.*]], label [[OMP_ARRAYINIT_BODY18:%.*]]
// CHECK1:       omp.arrayinit.body18:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT22:%.*]], [[OMP_ARRAYINIT_BODY18]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYINIT_BODY18]] ]
// CHECK1-NEXT:    [[TMP31:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP31]], i64 4
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK1-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP33]], i64 4
// CHECK1-NEXT:    [[TMP34:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP32]], %struct.BaseS1* [[TMP34]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT22]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE24:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYINIT_DONE25]], label [[OMP_ARRAYINIT_BODY18]]
// CHECK1:       omp.arrayinit.done25:
// CHECK1-NEXT:    [[TMP35:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP4]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP36:%.*]] = ptrtoint %struct.S.0* [[TMP35]] to i64
// CHECK1-NEXT:    [[TMP37:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX10]] to i64
// CHECK1-NEXT:    [[TMP38:%.*]] = sub i64 [[TMP36]], [[TMP37]]
// CHECK1-NEXT:    [[TMP39:%.*]] = sdiv exact i64 [[TMP38]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP39]]
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[TMP40]] to [10 x [4 x %struct.S.0]]*
// CHECK1-NEXT:    [[TMP42:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP43:%.*]] = load i32, i32* [[TMP42]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP43]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP44:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP44]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP45:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP45]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP46]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP47:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP48:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP26:%.*]] = icmp sle i32 [[TMP47]], [[TMP48]]
// CHECK1-NEXT:    br i1 [[CMP26]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP49]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP50:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX27:%.*]] = getelementptr inbounds i32, i32* [[TMP21]], i64 [[TMP50]]
// CHECK1-NEXT:    [[TMP51:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP51]] to i64
// CHECK1-NEXT:    [[ARRAYIDX28:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX27]], i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP52:%.*]] = load i32, i32* [[ARRAYIDX28]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK1-NEXT:    store i32 [[INC]], i32* [[ARRAYIDX28]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP53:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD29:%.*]] = add nsw i32 [[TMP53]], 1
// CHECK1-NEXT:    store i32 [[ADD29]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP54:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP55:%.*]] = load i32, i32* [[TMP54]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP55]])
// CHECK1-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP57:%.*]] = bitcast i32* [[VLA7]] to i8*
// CHECK1-NEXT:    store i8* [[TMP57]], i8** [[TMP56]], align 8
// CHECK1-NEXT:    [[TMP58:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP59:%.*]] = inttoptr i64 [[TMP13]] to i8*
// CHECK1-NEXT:    store i8* [[TMP59]], i8** [[TMP58]], align 8
// CHECK1-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP61:%.*]] = bitcast %struct.S.0* [[VLA16]] to i8*
// CHECK1-NEXT:    store i8* [[TMP61]], i8** [[TMP60]], align 8
// CHECK1-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP63:%.*]] = inttoptr i64 [[TMP28]] to i8*
// CHECK1-NEXT:    store i8* [[TMP63]], i8** [[TMP62]], align 8
// CHECK1-NEXT:    [[TMP64:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP65:%.*]] = load i32, i32* [[TMP64]], align 4
// CHECK1-NEXT:    [[TMP66:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP67:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP65]], i32 2, i64 32, i8* [[TMP66]], void (i8*, i8*)* @.omp.reduction.reduction_func.12, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP67]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP68:%.*]] = getelementptr i32, i32* [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[ARRAYIDX3]], [[TMP68]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE34:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST30:%.*]] = phi i32* [ [[VLA7]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST31:%.*]] = phi i32* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST31]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST30]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST31]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST30]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE33:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP68]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE33]], label [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done34:
// CHECK1-NEXT:    [[TMP69:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX10]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY35:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX10]], [[TMP69]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY35]], label [[OMP_ARRAYCPY_DONE44:%.*]], label [[OMP_ARRAYCPY_BODY36:%.*]]
// CHECK1:       omp.arraycpy.body36:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST37:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT42:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST38:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT41:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK1-NEXT:    [[TMP70:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST38]] to i8*
// CHECK1-NEXT:    [[ADD_PTR39:%.*]] = getelementptr inbounds i8, i8* [[TMP70]], i64 4
// CHECK1-NEXT:    [[TMP71:%.*]] = bitcast i8* [[ADD_PTR39]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP72:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST37]] to i8*
// CHECK1-NEXT:    [[ADD_PTR40:%.*]] = getelementptr inbounds i8, i8* [[TMP72]], i64 4
// CHECK1-NEXT:    [[TMP73:%.*]] = bitcast i8* [[ADD_PTR40]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP71]], %struct.BaseS1* [[TMP73]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT41]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST38]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT42]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST37]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE43:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT41]], [[TMP69]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE43]], label [[OMP_ARRAYCPY_DONE44]], label [[OMP_ARRAYCPY_BODY36]]
// CHECK1:       omp.arraycpy.done44:
// CHECK1-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP65]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP74:%.*]] = getelementptr i32, i32* [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY45:%.*]] = icmp eq i32* [[ARRAYIDX3]], [[TMP74]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY45]], label [[OMP_ARRAYCPY_DONE52:%.*]], label [[OMP_ARRAYCPY_BODY46:%.*]]
// CHECK1:       omp.arraycpy.body46:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST47:%.*]] = phi i32* [ [[VLA7]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT50:%.*]], [[OMP_ARRAYCPY_BODY46]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST48:%.*]] = phi i32* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT49:%.*]], [[OMP_ARRAYCPY_BODY46]] ]
// CHECK1-NEXT:    [[TMP75:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP76:%.*]] = load i32, i32* [[TMP75]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP76]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST48]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST47]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP76]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT49]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST48]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT50]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST47]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE51:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT49]], [[TMP74]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE51]], label [[OMP_ARRAYCPY_DONE52]], label [[OMP_ARRAYCPY_BODY46]]
// CHECK1:       omp.arraycpy.done52:
// CHECK1-NEXT:    [[TMP77:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX10]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY53:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX10]], [[TMP77]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY53]], label [[OMP_ARRAYCPY_DONE62:%.*]], label [[OMP_ARRAYCPY_BODY54:%.*]]
// CHECK1:       omp.arraycpy.body54:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST55:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYCPY_DONE52]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT60:%.*]], [[OMP_ARRAYCPY_BODY54]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST56:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYCPY_DONE52]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT59:%.*]], [[OMP_ARRAYCPY_BODY54]] ]
// CHECK1-NEXT:    [[TMP78:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP79:%.*]] = load i32, i32* [[TMP78]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP80:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST56]] to i8*
// CHECK1-NEXT:    [[ADD_PTR57:%.*]] = getelementptr inbounds i8, i8* [[TMP80]], i64 4
// CHECK1-NEXT:    [[TMP81:%.*]] = bitcast i8* [[ADD_PTR57]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP82:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST55]] to i8*
// CHECK1-NEXT:    [[ADD_PTR58:%.*]] = getelementptr inbounds i8, i8* [[TMP82]], i64 4
// CHECK1-NEXT:    [[TMP83:%.*]] = bitcast i8* [[ADD_PTR58]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP81]], %struct.BaseS1* [[TMP83]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT59]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST56]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT60]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST55]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE61:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT59]], [[TMP77]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE61]], label [[OMP_ARRAYCPY_DONE62]], label [[OMP_ARRAYCPY_BODY54]]
// CHECK1:       omp.arraycpy.done62:
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP84:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP28]]
// CHECK1-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA16]], [[TMP84]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE63:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP84]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[VLA16]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE63]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done63:
// CHECK1-NEXT:    [[TMP85:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore(i8* [[TMP85]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..10
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 555, [[TMP4]]
// CHECK1-NEXT:    store i32 [[MUL]], i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..11
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32 888, i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.12
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = ptrtoint i8* [[TMP22]] to i64
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr i32, i32* [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP20]], i64 [[TMP23]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq %struct.S.0* [[TMP20]], [[TMP25]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK1:       omp.arraycpy.body4:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi %struct.S.0* [ [[TMP20]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP26]], i64 4
// CHECK1-NEXT:    [[TMP27:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP28:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]] to i8*
// CHECK1-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds i8, i8* [[TMP28]], i64 4
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast i8* [[ADD_PTR7]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP27]], %struct.BaseS1* [[TMP29]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT9]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP25]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK1:       omp.arraycpy.done11:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARR:%.*]], [10 x [4 x %struct.S.0]]* nonnull align 4 dereferenceable(480) [[ARRS:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[ARRS_ADDR:%.*]] = alloca [10 x [4 x %struct.S.0]]*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARRS5:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA1]], i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    store i32* [[ARR]], i32** [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store [10 x [4 x %struct.S.0]]* [[ARRS]], [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load [10 x [4 x %struct.S.0]]*, [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = mul nuw i64 [[TMP0]], [[TMP1]]
// CHECK1-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP4]], 4
// CHECK1-NEXT:    [[TMP6:%.*]] = udiv exact i64 [[TMP5]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP7:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT:    store i8* [[TMP7]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA3:%.*]] = alloca i32, i64 [[TMP6]], align 16
// CHECK1-NEXT:    store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr i32, i32* [[VLA3]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[VLA3]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[VLA3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @.omp_initializer..11(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS5]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP3]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY6:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP10]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY6]], label [[OMP_ARRAYINIT_DONE14:%.*]], label [[OMP_ARRAYINIT_BODY7:%.*]]
// CHECK1:       omp.arrayinit.body7:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[TMP9]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYINIT_BODY7]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT12:%.*]], [[OMP_ARRAYINIT_BODY7]] ]
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP11]], i64 4
// CHECK1-NEXT:    [[TMP12:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]] to i8*
// CHECK1-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP12]], %struct.BaseS1* [[TMP14]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT12]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT12]], [[TMP10]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYINIT_DONE14]], label [[OMP_ARRAYINIT_BODY7]]
// CHECK1:       omp.arrayinit.done14:
// CHECK1-NEXT:    [[LHS_BEGIN:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP3]] to %struct.S.0*
// CHECK1-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[ARRS5]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP16]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP17]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP18]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP19]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP15:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[CMP15]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP23:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA3]], i64 [[TMP23]]
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP24]] to i64
// CHECK1-NEXT:    [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX16]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK1-NEXT:    store i32 [[INC]], i32* [[ARRAYIDX16]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD17:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK1-NEXT:    store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP28]])
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast i32* [[VLA3]] to i8*
// CHECK1-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP32:%.*]] = inttoptr i64 [[TMP6]] to i8*
// CHECK1-NEXT:    store i8* [[TMP32]], i8** [[TMP31]], align 8
// CHECK1-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP34:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK1-NEXT:    store i8* [[TMP34]], i8** [[TMP33]], align 8
// CHECK1-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK1-NEXT:    [[TMP37:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP38:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], i32 2, i64 24, i8* [[TMP37]], void (i8*, i8*)* @.omp.reduction.reduction_func.14, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP38]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP39:%.*]] = getelementptr i32, i32* [[TMP2]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP2]], [[TMP39]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST18:%.*]] = phi i32* [ [[VLA3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST19:%.*]] = phi i32* [ [[TMP2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST18]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST18]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP39]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done22:
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE32:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK1:       omp.arraycpy.body24:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[OMP_ARRAYCPY_DONE22]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT30:%.*]], [[OMP_ARRAYCPY_BODY24]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[OMP_ARRAYCPY_DONE22]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT29:%.*]], [[OMP_ARRAYCPY_BODY24]] ]
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] to i8*
// CHECK1-NEXT:    [[ADD_PTR27:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK1-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR27]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP43:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST25]] to i8*
// CHECK1-NEXT:    [[ADD_PTR28:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK1-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR28]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP42]], %struct.BaseS1* [[TMP44]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT29]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT30]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE31:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT29]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_DONE32]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK1:       omp.arraycpy.done32:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP45:%.*]] = getelementptr i32, i32* [[TMP2]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY33:%.*]] = icmp eq i32* [[TMP2]], [[TMP45]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY33]], label [[OMP_ARRAYCPY_DONE40:%.*]], label [[OMP_ARRAYCPY_BODY34:%.*]]
// CHECK1:       omp.arraycpy.body34:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST35:%.*]] = phi i32* [ [[VLA3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT38:%.*]], [[OMP_ARRAYCPY_BODY34]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST36:%.*]] = phi i32* [ [[TMP2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT37:%.*]], [[OMP_ARRAYCPY_BODY34]] ]
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP47:%.*]] = load i32, i32* [[TMP46]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST36]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST35]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT37]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST36]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT38]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST35]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE39:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT37]], [[TMP45]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE39]], label [[OMP_ARRAYCPY_DONE40]], label [[OMP_ARRAYCPY_BODY34]]
// CHECK1:       omp.arraycpy.done40:
// CHECK1-NEXT:    [[TMP48:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY41:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP48]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY41]], label [[OMP_ARRAYCPY_DONE50:%.*]], label [[OMP_ARRAYCPY_BODY42:%.*]]
// CHECK1:       omp.arraycpy.body42:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST43:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[OMP_ARRAYCPY_DONE40]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT48:%.*]], [[OMP_ARRAYCPY_BODY42]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST44:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[OMP_ARRAYCPY_DONE40]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT47:%.*]], [[OMP_ARRAYCPY_BODY42]] ]
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP50:%.*]] = load i32, i32* [[TMP49]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP51:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST44]] to i8*
// CHECK1-NEXT:    [[ADD_PTR45:%.*]] = getelementptr inbounds i8, i8* [[TMP51]], i64 4
// CHECK1-NEXT:    [[TMP52:%.*]] = bitcast i8* [[ADD_PTR45]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP53:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST43]] to i8*
// CHECK1-NEXT:    [[ADD_PTR46:%.*]] = getelementptr inbounds i8, i8* [[TMP53]], i64 4
// CHECK1-NEXT:    [[TMP54:%.*]] = bitcast i8* [[ADD_PTR46]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP52]], %struct.BaseS1* [[TMP54]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT47]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST44]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT48]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST43]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE49:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT47]], [[TMP48]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE49]], label [[OMP_ARRAYCPY_DONE50]], label [[OMP_ARRAYCPY_BODY42]]
// CHECK1:       omp.arraycpy.done50:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN51:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS5]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN51]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN51]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE52:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done52:
// CHECK1-NEXT:    [[TMP56:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore(i8* [[TMP56]])
// CHECK1-NEXT:    [[TMP57:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP58:%.*]] = load i32, i32* [[TMP57]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP58]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.14
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr i32, i32* [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP20]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq %struct.S.0* [[TMP20]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK1:       omp.arraycpy.body4:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi %struct.S.0* [ [[TMP20]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP23]], i64 4
// CHECK1-NEXT:    [[TMP24:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP25:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]] to i8*
// CHECK1-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds i8, i8* [[TMP25]], i64 4
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i8* [[ADD_PTR7]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP24]], %struct.BaseS1* [[TMP26]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT9]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK1:       omp.arraycpy.done11:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S.0*** nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VAR2_ADDR:%.*]] = alloca %struct.S.0***, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca %struct.S.0**, align 8
// CHECK1-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store %struct.S.0*** [[VAR2]], %struct.S.0**** [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load %struct.S.0***, %struct.S.0**** [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds %struct.S.0*, %struct.S.0** [[TMP1]], i64 0
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP2]], i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds %struct.S.0*, %struct.S.0** [[TMP3]], i64 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[ARRAYIDX2]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP4]], i64 6
// CHECK1-NEXT:    [[TMP5:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP6:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP7:%.*]] = sub i64 [[TMP5]], [[TMP6]]
// CHECK1-NEXT:    [[TMP8:%.*]] = sdiv exact i64 [[TMP7]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP9:%.*]] = add nuw i64 [[TMP8]], 1
// CHECK1-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP11:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT:    store i8* [[TMP11]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA:%.*]] = alloca [[STRUCT_S_0]], i64 [[TMP9]], align 16
// CHECK1-NEXT:    store i64 [[TMP9]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[VLA]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR4:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR4]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP17:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP17]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = ptrtoint %struct.S.0* [[TMP18]] to i64
// CHECK1-NEXT:    [[TMP20:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP21:%.*]] = sub i64 [[TMP19]], [[TMP20]]
// CHECK1-NEXT:    [[TMP22:%.*]] = sdiv exact i64 [[TMP21]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP22]]
// CHECK1-NEXT:    store %struct.S.0** [[_TMP7]], %struct.S.0*** [[_TMP6]], align 8
// CHECK1-NEXT:    store %struct.S.0* [[TMP23]], %struct.S.0** [[_TMP7]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP25]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP26]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP27]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP28]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP29]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP31]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK1-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP34]])
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[VLA]] to i8*
// CHECK1-NEXT:    store i8* [[TMP36]], i8** [[TMP35]], align 8
// CHECK1-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP38:%.*]] = inttoptr i64 [[TMP9]] to i8*
// CHECK1-NEXT:    store i8* [[TMP38]], i8** [[TMP37]], align 8
// CHECK1-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = load i32, i32* [[TMP39]], align 4
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP42:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], i32 1, i64 16, i8* [[TMP41]], void (i8*, i8*)* @.omp.reduction.reduction_func.16, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP42]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP43:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX1]], [[TMP43]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE16:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi %struct.S.0* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT14:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP44:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] to i8*
// CHECK1-NEXT:    [[ADD_PTR12:%.*]] = getelementptr inbounds i8, i8* [[TMP44]], i64 4
// CHECK1-NEXT:    [[TMP45:%.*]] = bitcast i8* [[ADD_PTR12]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP46:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]] to i8*
// CHECK1-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP46]], i64 4
// CHECK1-NEXT:    [[TMP47:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP45]], %struct.BaseS1* [[TMP47]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT14]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE15:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT14]], [[TMP43]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done16:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP48:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX1]], [[TMP48]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY17]], label [[OMP_ARRAYCPY_DONE26:%.*]], label [[OMP_ARRAYCPY_BODY18:%.*]]
// CHECK1:       omp.arraycpy.body18:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP50:%.*]] = load i32, i32* [[TMP49]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP51:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK1-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP51]], i64 4
// CHECK1-NEXT:    [[TMP52:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP53:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK1-NEXT:    [[ADD_PTR22:%.*]] = getelementptr inbounds i8, i8* [[TMP53]], i64 4
// CHECK1-NEXT:    [[TMP54:%.*]] = bitcast i8* [[ADD_PTR22]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP52]], %struct.BaseS1* [[TMP54]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT24]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE25:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP48]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_BODY18]]
// CHECK1:       omp.arraycpy.done26:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP9]]
// CHECK1-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA]], [[TMP55]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE27:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[VLA]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE27]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done27:
// CHECK1-NEXT:    [[TMP56:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore(i8* [[TMP56]])
// CHECK1-NEXT:    [[TMP57:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP58:%.*]] = load i32, i32* [[TMP57]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP58]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.16
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP16]], i64 4
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP18:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP18]], i64 4
// CHECK1-NEXT:    [[TMP19:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP17]], %struct.BaseS1* [[TMP19]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..17
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [5 x %struct.S.0]* nonnull align 4 dereferenceable(60) [[VVAR2:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VVAR2_ADDR:%.*]] = alloca [5 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VVAR22:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [5 x %struct.S.0]* [[VVAR2]], [5 x %struct.S.0]** [[VVAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load [5 x %struct.S.0]*, [5 x %struct.S.0]** [[VVAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[TMP0]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[TMP0]], i64 0, i64 4
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP1]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[TMP2:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP4:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR3:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR3]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP3]], %struct.BaseS1* [[TMP5]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP1]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP6:%.*]] = bitcast [5 x %struct.S.0]* [[TMP0]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint %struct.S.0* [[TMP6]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast [5 x %struct.S.0]* [[VVAR22]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[TMP11]], i64 [[TMP10]]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[TMP12]] to [5 x %struct.S.0]*
// CHECK1-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [5 x %struct.S.0]* [[VVAR22]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP15]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP16]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK1-NEXT:    store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP24]])
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK1-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP30:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], i32 1, i64 8, i8* [[TMP29]], void (i8*, i8*)* @.omp.reduction.reduction_func.18, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP30]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE13:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST7:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST8]] to i8*
// CHECK1-NEXT:    [[ADD_PTR9:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR9]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP34:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 4
// CHECK1-NEXT:    [[TMP35:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP33]], %struct.BaseS1* [[TMP35]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST7]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE12:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE12]], label [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done13:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP36:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY14:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP36]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY14]], label [[OMP_ARRAYCPY_DONE23:%.*]], label [[OMP_ARRAYCPY_BODY15:%.*]]
// CHECK1:       omp.arraycpy.body15:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST16:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT21:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST17:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK1-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP38:%.*]] = load i32, i32* [[TMP37]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP38]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP39:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST17]] to i8*
// CHECK1-NEXT:    [[ADD_PTR18:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 4
// CHECK1-NEXT:    [[TMP40:%.*]] = bitcast i8* [[ADD_PTR18]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST16]] to i8*
// CHECK1-NEXT:    [[ADD_PTR19:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK1-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR19]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP40]], %struct.BaseS1* [[TMP42]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP38]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST17]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT21]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST16]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE22:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP36]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_DONE23]], label [[OMP_ARRAYCPY_BODY15]]
// CHECK1:       omp.arraycpy.done23:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN24:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN24]], i64 5
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP43]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN24]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE25:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done25:
// CHECK1-NEXT:    [[TMP44:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP45:%.*]] = load i32, i32* [[TMP44]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP45]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.18
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[VAR3:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR34:%.*]] = alloca [2 x %struct.S.0], align 16
// CHECK1-NEXT:    [[_TMP7:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[VAR3]], [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[TMP0]], [4 x %struct.S.0]** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[TMP]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[TMP1]], [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT6:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP5]], i64 4
// CHECK1-NEXT:    [[TMP6:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP6]], %struct.BaseS1* [[TMP8]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT6]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT6]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP9:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast [4 x %struct.S.0]* [[TMP9]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP11:%.*]] = ptrtoint %struct.S.0* [[TMP10]] to i64
// CHECK1-NEXT:    [[TMP12:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast [2 x %struct.S.0]* [[VAR34]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[TMP15]], i64 [[TMP14]]
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast %struct.S.0* [[TMP16]] to [4 x %struct.S.0]*
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[TMP17]], [4 x %struct.S.0]** [[_TMP7]], align 8
// CHECK1-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [2 x %struct.S.0]* [[VAR34]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP19]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP21]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP22]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP25]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK1-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP28]])
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK1-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP34:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], i32 1, i64 8, i8* [[TMP33]], void (i8*, i8*)* @.omp.reduction.reduction_func.20, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP34]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE16:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT14:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] to i8*
// CHECK1-NEXT:    [[ADD_PTR12:%.*]] = getelementptr inbounds i8, i8* [[TMP36]], i64 4
// CHECK1-NEXT:    [[TMP37:%.*]] = bitcast i8* [[ADD_PTR12]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP38:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]] to i8*
// CHECK1-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 4
// CHECK1-NEXT:    [[TMP39:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP37]], %struct.BaseS1* [[TMP39]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT14]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE15:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT14]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done16:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY17]], label [[OMP_ARRAYCPY_DONE26:%.*]], label [[OMP_ARRAYCPY_BODY18:%.*]]
// CHECK1:       omp.arraycpy.body18:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK1-NEXT:    [[TMP41:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP42:%.*]] = load i32, i32* [[TMP41]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP42]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP43:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK1-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK1-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP45:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK1-NEXT:    [[ADD_PTR22:%.*]] = getelementptr inbounds i8, i8* [[TMP45]], i64 4
// CHECK1-NEXT:    [[TMP46:%.*]] = bitcast i8* [[ADD_PTR22]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP44]], %struct.BaseS1* [[TMP46]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP42]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT24]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE25:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_BODY18]]
// CHECK1:       omp.arraycpy.done26:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN27:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN27]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP47]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN27]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE28:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done28:
// CHECK1-NEXT:    [[TMP48:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32, i32* [[TMP48]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP49]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.20
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..21
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[VAR3:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[VAR3]], [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[TMP0]], [4 x %struct.S.0]** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[TMP]], align 8
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[TMP1]], [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK1-NEXT:    [[DOTVAR3__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP4]], i64 48, i8* inttoptr (i64 6 to i8*))
// CHECK1-NEXT:    [[DOTVAR3__ADDR:%.*]] = bitcast i8* [[DOTVAR3__VOID_ADDR]] to [4 x %struct.S.0]*
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[DOTVAR3__ADDR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast [4 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR3:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 4
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast i8* [[ADD_PTR3]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP8]], %struct.BaseS1* [[TMP10]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    store [4 x %struct.S.0]* [[DOTVAR3__ADDR]], [4 x %struct.S.0]** [[_TMP5]], align 8
// CHECK1-NEXT:    [[LHS_BEGIN:%.*]] = bitcast [4 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK1-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [4 x %struct.S.0]* [[DOTVAR3__ADDR]] to %struct.S.0*
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP19:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK1-NEXT:    store i8* [[TMP19]], i8** [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP21:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, i8* [[TMP20]], void (i8*, i8*)* @.omp.reduction.reduction_func.22, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP21]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]] to i8*
// CHECK1-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP23]], i64 4
// CHECK1-NEXT:    [[TMP24:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP25:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]] to i8*
// CHECK1-NEXT:    [[ADD_PTR11:%.*]] = getelementptr inbounds i8, i8* [[TMP25]], i64 4
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i8* [[ADD_PTR11]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP24]], %struct.BaseS1* [[TMP26]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT12]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT12]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done14:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY15:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY15]], label [[OMP_ARRAYCPY_DONE24:%.*]], label [[OMP_ARRAYCPY_BODY16:%.*]]
// CHECK1:       omp.arraycpy.body16:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST17:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT22:%.*]], [[OMP_ARRAYCPY_BODY16]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST18:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT21:%.*]], [[OMP_ARRAYCPY_BODY16]] ]
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP28:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST18]] to i8*
// CHECK1-NEXT:    [[ADD_PTR19:%.*]] = getelementptr inbounds i8, i8* [[TMP28]], i64 4
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast i8* [[ADD_PTR19]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST17]] to i8*
// CHECK1-NEXT:    [[ADD_PTR20:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK1-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR20]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP29]], %struct.BaseS1* [[TMP31]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT21]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST18]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT22]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST17]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE23:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT21]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE23]], label [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYCPY_BODY16]]
// CHECK1:       omp.arraycpy.done24:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN25:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[DOTVAR3__ADDR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN25]], i64 4
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP32]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN25]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE26:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done26:
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast [4 x %struct.S.0]* [[DOTVAR3__ADDR]] to i8*
// CHECK1-NEXT:    call void @__kmpc_free(i32 [[TMP4]], i8* [[TMP33]], i8* inttoptr (i64 6 to i8*))
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP4]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.22
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v
// CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = 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:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[ARR:%.*]] = alloca [42 x %struct.S], align 16
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[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._Z5tmainIiLi42EET_v.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 @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK1-NEXT:    store %struct.S* [[TEST]], %struct.S** [[VAR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 42
// 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 @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[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:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, %struct.S*, %struct.S*, i32*, [2 x i32]*, [2 x %struct.S]*)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i32* [[T_VAR]], %struct.S* [[TMP1]], %struct.S* [[VAR1]], i32* [[T_VAR1]], [2 x i32]* [[VEC]], [2 x %struct.S]* [[S_ARR]])
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// 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*, i32*, [2 x i32]*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* [[T_VAR]], [2 x i32]* [[VEC]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[TMP2]])
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [42 x %struct.S]*, [2 x i32]*, i32*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), [42 x %struct.S]* [[ARR]], [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[TMP3]])
// CHECK1-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    [[ARRAY_BEGIN1:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN1]], i64 42
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP4]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN1]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN3]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY4:%.*]]
// CHECK1:       arraydestroy.body4:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST5:%.*]] = phi %struct.S* [ [[TMP5]], [[ARRAYDESTROY_DONE2]] ], [ [[ARRAYDESTROY_ELEMENT6:%.*]], [[ARRAYDESTROY_BODY4]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT6]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST5]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT6]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE7:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT6]], [[ARRAY_BEGIN3]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE7]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY4]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR4]]
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP6]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK1-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK1-NEXT:    [[CONV:%.*]] = fptrunc double [[TMP4]] to float
// CHECK1-NEXT:    store float [[CONV]], float* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK1-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = fpext float [[TMP4]] to double
// CHECK1-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK1-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK1-NEXT:    [[CONV2:%.*]] = fptrunc double [[ADD]] to float
// CHECK1-NEXT:    store float [[CONV2]], float* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK1-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]], i32 [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..23
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR1:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[VAR17:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[T_VAR18:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    store i32* [[T_VAR1]], i32** [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP6]], %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    call void @.omp_initializer..25(i32* [[T_VAR3]], i32* [[TMP0]])
// CHECK1-NEXT:    [[TMP7:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP8]], i64 4
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 4
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP9]], %struct.BaseS1* [[TMP11]])
// CHECK1-NEXT:    store %struct.S* [[VAR4]], %struct.S** [[_TMP6]], align 8
// CHECK1-NEXT:    call void @.omp_initializer..27(%struct.S* [[VAR17]], %struct.S* [[TMP2]])
// CHECK1-NEXT:    call void @.omp_initializer..29(i32* [[T_VAR18]], i32* [[TMP3]])
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP14]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP20]], i32* [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP22:%.*]] = load %struct.S*, %struct.S** [[_TMP6]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP23]] to i64
// CHECK1-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP5]], i64 0, i64 [[IDXPROM10]]
// CHECK1-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX11]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP22]])
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK1-NEXT:    store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]])
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i32* [[T_VAR3]] to i8*
// CHECK1-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP28:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK1-NEXT:    store i8* [[TMP28]], i8** [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast %struct.S* [[VAR17]] to i8*
// CHECK1-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast i32* [[T_VAR18]] to i8*
// CHECK1-NEXT:    store i8* [[TMP32]], i8** [[TMP31]], align 8
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP34:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 4, i64 32, i8* [[TMP33]], void (i8*, i8*)* @.omp.reduction.reduction_func.30, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP34]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    call void @.omp_combiner..24(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK1-NEXT:    [[TMP35:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP35]], i64 4
// CHECK1-NEXT:    [[TMP36:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP37:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP37]], i64 4
// CHECK1-NEXT:    [[TMP38:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP36]], %struct.BaseS1* [[TMP38]])
// CHECK1-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP2]], %struct.S* [[VAR17]])
// CHECK1-NEXT:    call void @.omp_combiner..28(i32* [[TMP3]], i32* [[T_VAR18]])
// CHECK1-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..24(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP39:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK1-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 4
// CHECK1-NEXT:    [[TMP40:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK1-NEXT:    [[ADD_PTR16:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK1-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR16]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP40]], %struct.BaseS1* [[TMP42]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP2]], %struct.S* [[VAR17]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..28(i32* [[TMP3]], i32* [[T_VAR18]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR17]]) #[[ATTR4]]
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR4]]) #[[ATTR4]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..24
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 1513, [[TMP4]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..25
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32 321, i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..26
// CHECK1-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK1-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK1-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..27
// CHECK1-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = bitcast %struct.S* [[TMP3]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP6:%.*]] = bitcast %struct.S* [[TMP2]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 4
// CHECK1-NEXT:    [[TMP7:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @_Z5init2R6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP5]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP7]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..28
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[SUB:%.*]] = sub nsw i32 47, [[TMP4]]
// CHECK1-NEXT:    store i32 [[SUB]], i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..29
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[DIV:%.*]] = sdiv i32 432, [[TMP4]]
// CHECK1-NEXT:    store i32 [[DIV]], i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.30
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S*
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S*
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S*
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S*
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i8* [[TMP25]] to i32*
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = bitcast i8* [[TMP28]] to i32*
// CHECK1-NEXT:    call void @.omp_combiner..24(i32* [[TMP11]], i32* [[TMP8]])
// CHECK1-NEXT:    [[TMP30:%.*]] = bitcast %struct.S* [[TMP17]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK1-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast %struct.S* [[TMP14]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK1-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP31]], %struct.BaseS1* [[TMP33]])
// CHECK1-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP23]], %struct.S* [[TMP20]])
// CHECK1-NEXT:    call void @.omp_combiner..28(i32* [[TMP29]], i32* [[TMP26]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..31
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_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:    store %struct.S* [[TMP3]], %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP4]], %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* @.init.32, align 4
// CHECK1-NEXT:    store i32 [[TMP5]], i32* [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP8]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP10]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP13]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, i32* [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP14]], i32* [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM5:%.*]] = sext i32 [[TMP17]] to i64
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 [[IDXPROM5]]
// CHECK1-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX6]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP16]])
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]])
// CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP20:%.*]] = bitcast i32* [[T_VAR3]] to i8*
// CHECK1-NEXT:    store i8* [[TMP20]], i8** [[TMP19]], align 8
// CHECK1-NEXT:    [[TMP21:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP22:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], i32 1, i64 8, i8* [[TMP21]], void (i8*, i8*)* @.omp.reduction.reduction_func.33, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP22]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    call void @.omp_combiner..34(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @.omp_combiner..34(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP7]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.33
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT:    call void @.omp_combiner..34(i32* [[TMP11]], i32* [[TMP8]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..34
// CHECK1-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK1-NEXT:    store i32 [[MUL]], i32* [[TMP3]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..35
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [42 x %struct.S]* nonnull align 4 dereferenceable(504) [[ARR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca [42 x %struct.S]*, 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:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[ARR4:%.*]] = alloca [40 x %struct.S], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store [42 x %struct.S]* [[ARR]], [42 x %struct.S]** [[ARR_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 [42 x %struct.S]*, [42 x %struct.S]** [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP4]], %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK1-NEXT:    store %struct.S* [[TMP5]], %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[TMP0]], i64 0, i64 1
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[TMP0]], i64 0, i64 40
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [40 x %struct.S], [40 x %struct.S]* [[ARR4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT6:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    [[TMP7:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 4
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_initializer..37(%struct.BaseS1* [[TMP8]], %struct.BaseS1* [[TMP10]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT6]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT6]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast [42 x %struct.S]* [[TMP0]] to %struct.S*
// CHECK1-NEXT:    [[TMP12:%.*]] = ptrtoint %struct.S* [[TMP11]] to i64
// CHECK1-NEXT:    [[TMP13:%.*]] = ptrtoint %struct.S* [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP14:%.*]] = sub i64 [[TMP12]], [[TMP13]]
// CHECK1-NEXT:    [[TMP15:%.*]] = sdiv exact i64 [[TMP14]], ptrtoint (%struct.S* getelementptr ([[STRUCT_S]], %struct.S* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast [40 x %struct.S]* [[ARR4]] to %struct.S*
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[TMP16]], i64 [[TMP15]]
// CHECK1-NEXT:    [[TMP18:%.*]] = bitcast %struct.S* [[TMP17]] to [42 x %struct.S]*
// CHECK1-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [40 x %struct.S]* [[ARR4]] to %struct.S*
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP20]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP21]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP22]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP23]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP24]], [[TMP25]]
// CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP26]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP28]] to i64
// CHECK1-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP27]], i32* [[ARRAYIDX8]], align 4
// CHECK1-NEXT:    [[TMP29:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM9:%.*]] = sext i32 [[TMP30]] to i64
// CHECK1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP3]], i64 0, i64 [[IDXPROM9]]
// CHECK1-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX10]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP29]])
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP31]], 1
// CHECK1-NEXT:    store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP33]])
// CHECK1-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP35:%.*]] = bitcast %struct.S* [[RHS_BEGIN]] to i8*
// CHECK1-NEXT:    store i8* [[TMP35]], i8** [[TMP34]], align 8
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4
// CHECK1-NEXT:    [[TMP38:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP39:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], i32 1, i64 8, i8* [[TMP38]], void (i8*, i8*)* @.omp.reduction.reduction_func.38, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP39]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAYIDX]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAYIDX]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST12:%.*]] = phi %struct.S* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST13:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP41:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST13]] to i8*
// CHECK1-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK1-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP43:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST12]] to i8*
// CHECK1-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK1-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP42]], %struct.BaseS1* [[TMP44]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST13]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST12]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done18:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP45:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAYIDX]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq %struct.S* [[ARRAYIDX]], [[TMP45]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE28:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]]
// CHECK1:       omp.arraycpy.body20:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi %struct.S* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT26:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT25:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP47:%.*]] = load i32, i32* [[TMP46]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP48:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST22]] to i8*
// CHECK1-NEXT:    [[ADD_PTR23:%.*]] = getelementptr inbounds i8, i8* [[TMP48]], i64 4
// CHECK1-NEXT:    [[TMP49:%.*]] = bitcast i8* [[ADD_PTR23]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP50:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST21]] to i8*
// CHECK1-NEXT:    [[ADD_PTR24:%.*]] = getelementptr inbounds i8, i8* [[TMP50]], i64 4
// CHECK1-NEXT:    [[TMP51:%.*]] = bitcast i8* [[ADD_PTR24]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP49]], %struct.BaseS1* [[TMP51]])
// CHECK1-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT25]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT26]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE27:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT25]], [[TMP45]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE27]], label [[OMP_ARRAYCPY_DONE28]], label [[OMP_ARRAYCPY_BODY20]]
// CHECK1:       omp.arraycpy.done28:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN29:%.*]] = getelementptr inbounds [40 x %struct.S], [40 x %struct.S]* [[ARR4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN29]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP52]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN29]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE30:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done30:
// CHECK1-NEXT:    [[TMP53:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP54:%.*]] = load i32, i32* [[TMP53]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP54]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_combiner..36
// CHECK1-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    call void @_Z8red_plusR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_initializer..37
// CHECK1-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK1-NEXT:    call void @_Z9init_plusR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.38
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S*
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[TMP11]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[TMP11]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP13:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK1-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK1-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK1-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK1-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK1-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to double
// CHECK1-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK1-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK1-NEXT:    [[CONV2:%.*]] = fptosi double [[ADD]] to i32
// CHECK1-NEXT:    store i32 [[CONV2]], i32* [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z4bazzv
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]])
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.S* [[S]])
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]]) #[[ATTR4:[0-9]+]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIiEC2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[S:%.*]]) #[[ATTR2:[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:    [[S_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[S1:%.*]] = alloca [[STRUCT_S:%.*]], 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 %struct.S* [[S]], %struct.S** [[S_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S*, %struct.S** [[S_ADDR]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    call void @.omp_initializer.(%struct.S* [[S1]], %struct.S* [[TMP0]])
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK2-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP1]], 10
// CHECK2-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP2]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !3
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK2-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP3]], 1
// CHECK2-NEXT:    store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !3
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK2-NEXT:    call void @.omp_combiner.(%struct.S* [[TMP0]], %struct.S* [[S1]])
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S1]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner.
// CHECK2-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK2-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer.
// CHECK2-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TMP3]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIiED2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK2-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK2-NEXT:    [[CONV:%.*]] = fptosi double [[TMP4]] to i32
// CHECK2-NEXT:    store i32 [[CONV]], i32* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[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 {{[^@]+}}@main
// CHECK2-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK2-NEXT:    [[T_VAR:%.*]] = alloca float, align 4
// CHECK2-NEXT:    [[T_VAR1:%.*]] = alloca float, align 4
// CHECK2-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK2-NEXT:    [[S_ARR:%.*]] = alloca [4 x %struct.S.0], align 16
// CHECK2-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK2-NEXT:    [[ARRS:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK2-NEXT:    [[VAR2:%.*]] = alloca %struct.S.0**, align 8
// CHECK2-NEXT:    [[VVAR2:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK2-NEXT:    [[VAR3:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK2-NEXT:    store float 0.000000e+00, float* [[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 [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT1]], float 3.000000e+00)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT1]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT2]], float 4.000000e+00)
// CHECK2-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// 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 @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[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:    [[CALL:%.*]] = call %struct.S.0** @_Z3foov()
// CHECK2-NEXT:    store %struct.S.0** [[CALL]], %struct.S.0*** [[VAR2]], align 8
// CHECK2-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN3]], i64 5
// CHECK2-NEXT:    br label [[ARRAYCTOR_LOOP5:%.*]]
// CHECK2:       arrayctor.loop5:
// CHECK2-NEXT:    [[ARRAYCTOR_CUR6:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYCTOR_NEXT7:%.*]], [[ARRAYCTOR_LOOP5]] ]
// CHECK2-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR6]])
// CHECK2-NEXT:    [[ARRAYCTOR_NEXT7]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR6]], i64 1
// CHECK2-NEXT:    [[ARRAYCTOR_DONE8:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT7]], [[ARRAYCTOR_END4]]
// CHECK2-NEXT:    br i1 [[ARRAYCTOR_DONE8]], label [[ARRAYCTOR_CONT9:%.*]], label [[ARRAYCTOR_LOOP5]]
// CHECK2:       arrayctor.cont9:
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, %struct.S.0*, %struct.S.0*, float*, [2 x i32]*, [4 x %struct.S.0]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), float* [[T_VAR]], %struct.S.0* [[TMP1]], %struct.S.0* [[VAR1]], float* [[T_VAR1]], [2 x i32]* [[VEC]], [4 x %struct.S.0]* [[S_ARR]])
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK2-NEXT:    [[TMP4:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT:    store i8* [[TMP4]], i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = mul nuw i64 10, [[TMP3]]
// CHECK2-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 16
// CHECK2-NEXT:    store i64 [[TMP3]], i64* [[__VLA_EXPR0]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, [2 x i32]*, [10 x [4 x %struct.S.0]]*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 10, i64 [[TMP3]], i32* [[VLA]], [2 x i32]* [[VEC]], [10 x [4 x %struct.S.0]]* [[ARRS]])
// 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*, i64, i64, i32*, [10 x [4 x %struct.S.0]]*)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 10, i64 [[TMP3]], i32* [[VLA]], [10 x [4 x %struct.S.0]]* [[ARRS]])
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.S.0***)* @.omp_outlined..15 to void (i32*, i32*, ...)*), %struct.S.0*** [[VAR2]])
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [5 x %struct.S.0]*)* @.omp_outlined..17 to void (i32*, i32*, ...)*), [5 x %struct.S.0]* [[VVAR2]])
// CHECK2-NEXT:    [[TMP6:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [4 x %struct.S.0]*)* @.omp_outlined..19 to void (i32*, i32*, ...)*), [4 x %struct.S.0]* [[TMP6]])
// CHECK2-NEXT:    [[TMP7:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [4 x %struct.S.0]*)* @.omp_outlined..21 to void (i32*, i32*, ...)*), [4 x %struct.S.0]* [[TMP7]])
// CHECK2-NEXT:    [[CALL10:%.*]] = call i32 @_Z5tmainIiLi42EET_v()
// CHECK2-NEXT:    store i32 [[CALL10]], i32* [[RETVAL]], align 4
// CHECK2-NEXT:    [[TMP8:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    call void @llvm.stackrestore(i8* [[TMP8]])
// CHECK2-NEXT:    [[ARRAY_BEGIN11:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN11]], i64 5
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP9]], [[ARRAYCTOR_CONT9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN11]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE12:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done12:
// CHECK2-NEXT:    [[ARRAY_BEGIN13:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN13]], i64 40
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY14:%.*]]
// CHECK2:       arraydestroy.body14:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST15:%.*]] = phi %struct.S.0* [ [[TMP10]], [[ARRAYDESTROY_DONE12]] ], [ [[ARRAYDESTROY_ELEMENT16:%.*]], [[ARRAYDESTROY_BODY14]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT16]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST15]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT16]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE17:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT16]], [[ARRAY_BEGIN13]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE17]], label [[ARRAYDESTROY_DONE18:%.*]], label [[ARRAYDESTROY_BODY14]]
// CHECK2:       arraydestroy.done18:
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN19:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN19]], i64 4
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY20:%.*]]
// CHECK2:       arraydestroy.body20:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST21:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ARRAYDESTROY_DONE18]] ], [ [[ARRAYDESTROY_ELEMENT22:%.*]], [[ARRAYDESTROY_BODY20]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT22]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST21]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT22]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE23:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT22]], [[ARRAY_BEGIN19]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE23]], label [[ARRAYDESTROY_DONE24:%.*]], label [[ARRAYDESTROY_BODY20]]
// CHECK2:       arraydestroy.done24:
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP12]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIfEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK2-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]], float [[TMP0]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(12) [[VAR:%.*]], %struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1:%.*]], float* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[S_ARR:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[VAR1_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[T_VAR3:%.*]] = alloca float, align 4
// CHECK2-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK2-NEXT:    [[_TMP6:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[VAR17:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK2-NEXT:    [[T_VAR18:%.*]] = alloca float, align 4
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store float* [[T_VAR]], float** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[VAR]], %struct.S.0** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[VAR1]], %struct.S.0** [[VAR1_ADDR]], align 8
// CHECK2-NEXT:    store float* [[T_VAR1]], float** [[T_VAR1_ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load float*, float** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR1_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load float*, float** [[T_VAR1_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP6]], %struct.S.0** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    call void @.omp_initializer..3(float* [[T_VAR3]], float* [[TMP0]])
// CHECK2-NEXT:    [[TMP7:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP8]], i64 4
// CHECK2-NEXT:    [[TMP9:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 4
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP9]], %struct.BaseS1* [[TMP11]])
// CHECK2-NEXT:    store %struct.S.0* [[VAR4]], %struct.S.0** [[_TMP6]], align 8
// CHECK2-NEXT:    call void @.omp_initializer..7(%struct.S.0* [[VAR17]], %struct.S.0* [[TMP2]])
// CHECK2-NEXT:    [[TMP12:%.*]] = load float, float* @.init, align 4
// CHECK2-NEXT:    store float [[TMP12]], float* [[T_VAR18]], align 4
// CHECK2-NEXT:    [[TMP13:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[TMP13]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP14]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP15]], 1
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP17]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
// CHECK2-NEXT:    br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP21:%.*]] = load float, float* [[T_VAR3]], align 4
// CHECK2-NEXT:    [[CONV:%.*]] = fptosi float [[TMP21]] to i32
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT:    store i32 [[CONV]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT:    [[TMP23:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP6]], align 8
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP24]] to i64
// CHECK2-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP5]], i64 0, i64 [[IDXPROM10]]
// CHECK2-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S.0* @_ZN1SIfEaSERKS0_(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYIDX11]], %struct.S.0* nonnull align 4 dereferenceable(12) [[TMP23]])
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK2-NEXT:    store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP14]])
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP27:%.*]] = bitcast float* [[T_VAR3]] to i8*
// CHECK2-NEXT:    store i8* [[TMP27]], i8** [[TMP26]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK2-NEXT:    store i8* [[TMP29]], i8** [[TMP28]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP31:%.*]] = bitcast %struct.S.0* [[VAR17]] to i8*
// CHECK2-NEXT:    store i8* [[TMP31]], i8** [[TMP30]], align 8
// CHECK2-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast float* [[T_VAR18]] to i8*
// CHECK2-NEXT:    store i8* [[TMP33]], i8** [[TMP32]], align 8
// CHECK2-NEXT:    [[TMP34:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP35:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP14]], i32 4, i64 32, i8* [[TMP34]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP35]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    call void @.omp_combiner..2(float* [[TMP0]], float* [[T_VAR3]])
// CHECK2-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP36]], i64 4
// CHECK2-NEXT:    [[TMP37:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP38:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 4
// CHECK2-NEXT:    [[TMP39:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP37]], %struct.BaseS1* [[TMP39]])
// CHECK2-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP2]], %struct.S.0* [[VAR17]])
// CHECK2-NEXT:    call void @.omp_combiner..8(float* [[TMP3]], float* [[T_VAR18]])
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..2(float* [[TMP0]], float* [[T_VAR3]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP40:%.*]] = bitcast %struct.S.0* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP40]], i64 4
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP42:%.*]] = bitcast %struct.S.0* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR16:%.*]] = getelementptr inbounds i8, i8* [[TMP42]], i64 4
// CHECK2-NEXT:    [[TMP43:%.*]] = bitcast i8* [[ADD_PTR16]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP41]], %struct.BaseS1* [[TMP43]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP2]], %struct.S.0* [[VAR17]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..8(float* [[TMP3]], float* [[T_VAR18]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP14]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR17]]) #[[ATTR4]]
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR4]]) #[[ATTR4]]
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP14]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..2
// CHECK2-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK2-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load float, float* [[TMP2]], align 4
// CHECK2-NEXT:    [[SUB:%.*]] = fsub float 2.220000e+02, [[TMP4]]
// CHECK2-NEXT:    store float [[SUB]], float* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..3
// CHECK2-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK2-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK2-NEXT:    store float -1.000000e+00, float* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..4
// CHECK2-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    call void @_Z3redR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..5
// CHECK2-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    call void @_Z4initR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..6
// CHECK2-SAME: (%struct.S.0* noalias [[TMP0:%.*]], %struct.S.0* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP2]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load float, float* [[F]], align 4
// CHECK2-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP5:%.*]] = load float, float* [[F2]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = fmul float [[TMP5]], [[TMP4]]
// CHECK2-NEXT:    store float [[MUL]], float* [[F2]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..7
// CHECK2-SAME: (%struct.S.0* noalias [[TMP0:%.*]], %struct.S.0* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP0]], %struct.S.0** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S.0*, %struct.S.0** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = bitcast %struct.S.0* [[TMP3]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP6:%.*]] = bitcast %struct.S.0* [[TMP2]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 4
// CHECK2-NEXT:    [[TMP7:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @_Z5init1R6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP5]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP7]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to float*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to float*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK2-NEXT:    [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i8* [[TMP25]] to float*
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast i8* [[TMP28]] to float*
// CHECK2-NEXT:    call void @.omp_combiner..2(float* [[TMP11]], float* [[TMP8]])
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[TMP17]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK2-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast %struct.S.0* [[TMP14]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP31]], %struct.BaseS1* [[TMP33]])
// CHECK2-NEXT:    call void @.omp_combiner..6(%struct.S.0* [[TMP23]], %struct.S.0* [[TMP20]])
// CHECK2-NEXT:    call void @.omp_combiner..8(float* [[TMP29]], float* [[TMP26]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..8
// CHECK2-SAME: (float* noalias [[TMP0:%.*]], float* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca float*, align 8
// CHECK2-NEXT:    store float* [[TMP0]], float** [[DOTADDR]], align 8
// CHECK2-NEXT:    store float* [[TMP1]], float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load float*, float** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load float*, float** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load float, float* [[TMP2]], align 4
// CHECK2-NEXT:    [[ADD:%.*]] = fadd float 5.550000e+02, [[TMP4]]
// CHECK2-NEXT:    store float [[ADD]], float* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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 @_ZN1SIfED2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [10 x [4 x %struct.S.0]]* nonnull align 4 dereferenceable(480) [[ARRS:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[ARR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[ARRS_ADDR:%.*]] = alloca [10 x [4 x %struct.S.0]]*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT:    store i64 [[VLA1]], i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT:    store i32* [[ARR]], i32** [[ARR_ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store [10 x [4 x %struct.S.0]]* [[ARRS]], [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load [10 x [4 x %struct.S.0]]*, [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP5:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[TMP5]]
// CHECK2-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 0
// CHECK2-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[ARRAYIDX4]], align 4
// CHECK2-NEXT:    [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK2-NEXT:    [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP7]]
// CHECK2-NEXT:    [[TMP8:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK2-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[TMP8]]
// CHECK2-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX5]], i64 [[LB_ADD_LEN]]
// CHECK2-NEXT:    [[TMP9:%.*]] = ptrtoint i32* [[ARRAYIDX6]] to i64
// CHECK2-NEXT:    [[TMP10:%.*]] = ptrtoint i32* [[ARRAYIDX3]] to i64
// CHECK2-NEXT:    [[TMP11:%.*]] = sub i64 [[TMP9]], [[TMP10]]
// CHECK2-NEXT:    [[TMP12:%.*]] = sdiv exact i64 [[TMP11]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP13:%.*]] = add nuw i64 [[TMP12]], 1
// CHECK2-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP15:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT:    store i8* [[TMP15]], i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    [[VLA7:%.*]] = alloca i32, i64 [[TMP13]], align 16
// CHECK2-NEXT:    store i64 [[TMP13]], i64* [[__VLA_EXPR0]], align 8
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr i32, i32* [[VLA7]], i64 [[TMP13]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[VLA7]], [[TMP16]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[ARRAYIDX3]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[VLA7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    call void @.omp_initializer..11(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP16]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP17:%.*]] = ptrtoint i32* [[TMP2]] to i64
// CHECK2-NEXT:    [[TMP18:%.*]] = ptrtoint i32* [[ARRAYIDX3]] to i64
// CHECK2-NEXT:    [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK2-NEXT:    [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr i32, i32* [[VLA7]], i64 [[TMP20]]
// CHECK2-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[TMP4]], i64 0, i64 1
// CHECK2-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[ARRAYIDX9]], i64 0, i64 0
// CHECK2-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAYDECAY]], i64 1
// CHECK2-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[ARRAYIDX11]], align 4
// CHECK2-NEXT:    [[TMP23:%.*]] = sext i32 [[TMP22]] to i64
// CHECK2-NEXT:    [[LB_ADD_LEN12:%.*]] = add nsw i64 0, [[TMP23]]
// CHECK2-NEXT:    [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[TMP4]], i64 0, i64 [[LB_ADD_LEN12]]
// CHECK2-NEXT:    [[ARRAYDECAY14:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[ARRAYIDX13]], i64 0, i64 0
// CHECK2-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDECAY14]], i64 2
// CHECK2-NEXT:    [[TMP24:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX15]] to i64
// CHECK2-NEXT:    [[TMP25:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX10]] to i64
// CHECK2-NEXT:    [[TMP26:%.*]] = sub i64 [[TMP24]], [[TMP25]]
// CHECK2-NEXT:    [[TMP27:%.*]] = sdiv exact i64 [[TMP26]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP28:%.*]] = add nuw i64 [[TMP27]], 1
// CHECK2-NEXT:    [[TMP29:%.*]] = mul nuw i64 [[TMP28]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[VLA16:%.*]] = alloca [[STRUCT_S_0]], i64 [[TMP28]], align 16
// CHECK2-NEXT:    store i64 [[TMP28]], i64* [[__VLA_EXPR1]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP28]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[VLA16]], [[TMP30]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY17]], label [[OMP_ARRAYINIT_DONE25:%.*]], label [[OMP_ARRAYINIT_BODY18:%.*]]
// CHECK2:       omp.arrayinit.body18:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT22:%.*]], [[OMP_ARRAYINIT_BODY18]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYINIT_BODY18]] ]
// CHECK2-NEXT:    [[TMP31:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP31]], i64 4
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK2-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP33]], i64 4
// CHECK2-NEXT:    [[TMP34:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP32]], %struct.BaseS1* [[TMP34]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT22]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE24:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP30]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYINIT_DONE25]], label [[OMP_ARRAYINIT_BODY18]]
// CHECK2:       omp.arrayinit.done25:
// CHECK2-NEXT:    [[TMP35:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP4]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP36:%.*]] = ptrtoint %struct.S.0* [[TMP35]] to i64
// CHECK2-NEXT:    [[TMP37:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX10]] to i64
// CHECK2-NEXT:    [[TMP38:%.*]] = sub i64 [[TMP36]], [[TMP37]]
// CHECK2-NEXT:    [[TMP39:%.*]] = sdiv exact i64 [[TMP38]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP39]]
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[TMP40]] to [10 x [4 x %struct.S.0]]*
// CHECK2-NEXT:    [[TMP42:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP43:%.*]] = load i32, i32* [[TMP42]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP43]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP44:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP44]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP45:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP45]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP46:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP46]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP47:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP48:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP26:%.*]] = icmp sle i32 [[TMP47]], [[TMP48]]
// CHECK2-NEXT:    br i1 [[CMP26]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP49]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP50:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK2-NEXT:    [[ARRAYIDX27:%.*]] = getelementptr inbounds i32, i32* [[TMP21]], i64 [[TMP50]]
// CHECK2-NEXT:    [[TMP51:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP51]] to i64
// CHECK2-NEXT:    [[ARRAYIDX28:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX27]], i64 [[IDXPROM]]
// CHECK2-NEXT:    [[TMP52:%.*]] = load i32, i32* [[ARRAYIDX28]], align 4
// CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP52]], 1
// CHECK2-NEXT:    store i32 [[INC]], i32* [[ARRAYIDX28]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP53:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD29:%.*]] = add nsw i32 [[TMP53]], 1
// CHECK2-NEXT:    store i32 [[ADD29]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP54:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP55:%.*]] = load i32, i32* [[TMP54]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP55]])
// CHECK2-NEXT:    [[TMP56:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP57:%.*]] = bitcast i32* [[VLA7]] to i8*
// CHECK2-NEXT:    store i8* [[TMP57]], i8** [[TMP56]], align 8
// CHECK2-NEXT:    [[TMP58:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP59:%.*]] = inttoptr i64 [[TMP13]] to i8*
// CHECK2-NEXT:    store i8* [[TMP59]], i8** [[TMP58]], align 8
// CHECK2-NEXT:    [[TMP60:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP61:%.*]] = bitcast %struct.S.0* [[VLA16]] to i8*
// CHECK2-NEXT:    store i8* [[TMP61]], i8** [[TMP60]], align 8
// CHECK2-NEXT:    [[TMP62:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP63:%.*]] = inttoptr i64 [[TMP28]] to i8*
// CHECK2-NEXT:    store i8* [[TMP63]], i8** [[TMP62]], align 8
// CHECK2-NEXT:    [[TMP64:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP65:%.*]] = load i32, i32* [[TMP64]], align 4
// CHECK2-NEXT:    [[TMP66:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP67:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP65]], i32 2, i64 32, i8* [[TMP66]], void (i8*, i8*)* @.omp.reduction.reduction_func.12, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP67]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP68:%.*]] = getelementptr i32, i32* [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[ARRAYIDX3]], [[TMP68]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE34:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST30:%.*]] = phi i32* [ [[VLA7]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST31:%.*]] = phi i32* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST31]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST30]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST31]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST30]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE33:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP68]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE33]], label [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done34:
// CHECK2-NEXT:    [[TMP69:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX10]], i64 [[TMP28]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY35:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX10]], [[TMP69]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY35]], label [[OMP_ARRAYCPY_DONE44:%.*]], label [[OMP_ARRAYCPY_BODY36:%.*]]
// CHECK2:       omp.arraycpy.body36:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST37:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT42:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST38:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT41:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK2-NEXT:    [[TMP70:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST38]] to i8*
// CHECK2-NEXT:    [[ADD_PTR39:%.*]] = getelementptr inbounds i8, i8* [[TMP70]], i64 4
// CHECK2-NEXT:    [[TMP71:%.*]] = bitcast i8* [[ADD_PTR39]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP72:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST37]] to i8*
// CHECK2-NEXT:    [[ADD_PTR40:%.*]] = getelementptr inbounds i8, i8* [[TMP72]], i64 4
// CHECK2-NEXT:    [[TMP73:%.*]] = bitcast i8* [[ADD_PTR40]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP71]], %struct.BaseS1* [[TMP73]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT41]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST38]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT42]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST37]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE43:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT41]], [[TMP69]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE43]], label [[OMP_ARRAYCPY_DONE44]], label [[OMP_ARRAYCPY_BODY36]]
// CHECK2:       omp.arraycpy.done44:
// CHECK2-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP65]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP74:%.*]] = getelementptr i32, i32* [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY45:%.*]] = icmp eq i32* [[ARRAYIDX3]], [[TMP74]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY45]], label [[OMP_ARRAYCPY_DONE52:%.*]], label [[OMP_ARRAYCPY_BODY46:%.*]]
// CHECK2:       omp.arraycpy.body46:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST47:%.*]] = phi i32* [ [[VLA7]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT50:%.*]], [[OMP_ARRAYCPY_BODY46]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST48:%.*]] = phi i32* [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT49:%.*]], [[OMP_ARRAYCPY_BODY46]] ]
// CHECK2-NEXT:    [[TMP75:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP76:%.*]] = load i32, i32* [[TMP75]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP76]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST48]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST47]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP76]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT49]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST48]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT50]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST47]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE51:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT49]], [[TMP74]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE51]], label [[OMP_ARRAYCPY_DONE52]], label [[OMP_ARRAYCPY_BODY46]]
// CHECK2:       omp.arraycpy.done52:
// CHECK2-NEXT:    [[TMP77:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX10]], i64 [[TMP28]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY53:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX10]], [[TMP77]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY53]], label [[OMP_ARRAYCPY_DONE62:%.*]], label [[OMP_ARRAYCPY_BODY54:%.*]]
// CHECK2:       omp.arraycpy.body54:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST55:%.*]] = phi %struct.S.0* [ [[VLA16]], [[OMP_ARRAYCPY_DONE52]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT60:%.*]], [[OMP_ARRAYCPY_BODY54]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST56:%.*]] = phi %struct.S.0* [ [[ARRAYIDX10]], [[OMP_ARRAYCPY_DONE52]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT59:%.*]], [[OMP_ARRAYCPY_BODY54]] ]
// CHECK2-NEXT:    [[TMP78:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP79:%.*]] = load i32, i32* [[TMP78]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP80:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST56]] to i8*
// CHECK2-NEXT:    [[ADD_PTR57:%.*]] = getelementptr inbounds i8, i8* [[TMP80]], i64 4
// CHECK2-NEXT:    [[TMP81:%.*]] = bitcast i8* [[ADD_PTR57]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP82:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST55]] to i8*
// CHECK2-NEXT:    [[ADD_PTR58:%.*]] = getelementptr inbounds i8, i8* [[TMP82]], i64 4
// CHECK2-NEXT:    [[TMP83:%.*]] = bitcast i8* [[ADD_PTR58]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP81]], %struct.BaseS1* [[TMP83]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT59]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST56]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT60]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST55]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE61:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT59]], [[TMP77]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE61]], label [[OMP_ARRAYCPY_DONE62]], label [[OMP_ARRAYCPY_BODY54]]
// CHECK2:       omp.arraycpy.done62:
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[TMP84:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[VLA16]], i64 [[TMP28]]
// CHECK2-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA16]], [[TMP84]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE63:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP84]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[VLA16]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE63]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done63:
// CHECK2-NEXT:    [[TMP85:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    call void @llvm.stackrestore(i8* [[TMP85]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..10
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 555, [[TMP4]]
// CHECK2-NEXT:    store i32 [[MUL]], i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..11
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32 888, i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.12
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK2-NEXT:    [[TMP23:%.*]] = ptrtoint i8* [[TMP22]] to i64
// CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr i32, i32* [[TMP11]], i64 [[TMP14]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP24]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP24]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done2:
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP20]], i64 [[TMP23]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq %struct.S.0* [[TMP20]], [[TMP25]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK2:       omp.arraycpy.body4:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi %struct.S.0* [ [[TMP20]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP26]], i64 4
// CHECK2-NEXT:    [[TMP27:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP28:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]] to i8*
// CHECK2-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds i8, i8* [[TMP28]], i64 4
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast i8* [[ADD_PTR7]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP27]], %struct.BaseS1* [[TMP29]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT9]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP25]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK2:       omp.arraycpy.done11:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARR:%.*]], [10 x [4 x %struct.S.0]]* nonnull align 4 dereferenceable(480) [[ARRS:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[ARR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[ARRS_ADDR:%.*]] = alloca [10 x [4 x %struct.S.0]]*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[ARRS5:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store i64 [[VLA]], i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT:    store i64 [[VLA1]], i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT:    store i32* [[ARR]], i32** [[ARR_ADDR]], align 8
// CHECK2-NEXT:    store [10 x [4 x %struct.S.0]]* [[ARRS]], [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR2]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load [10 x [4 x %struct.S.0]]*, [10 x [4 x %struct.S.0]]** [[ARRS_ADDR]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP4:%.*]] = mul nuw i64 [[TMP0]], [[TMP1]]
// CHECK2-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP4]], 4
// CHECK2-NEXT:    [[TMP6:%.*]] = udiv exact i64 [[TMP5]], ptrtoint (i32* getelementptr (i32, i32* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP7:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT:    store i8* [[TMP7]], i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    [[VLA3:%.*]] = alloca i32, i64 [[TMP6]], align 16
// CHECK2-NEXT:    store i64 [[TMP6]], i64* [[__VLA_EXPR0]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = getelementptr i32, i32* [[VLA3]], i64 [[TMP6]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i32* [[VLA3]], [[TMP8]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[VLA3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    call void @.omp_initializer..11(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP8]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS5]], i32 0, i32 0, i32 0
// CHECK2-NEXT:    [[TMP9:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP3]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP10:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY6:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP10]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY6]], label [[OMP_ARRAYINIT_DONE14:%.*]], label [[OMP_ARRAYINIT_BODY7:%.*]]
// CHECK2:       omp.arrayinit.body7:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[TMP9]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYINIT_BODY7]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT12:%.*]], [[OMP_ARRAYINIT_BODY7]] ]
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP11]], i64 4
// CHECK2-NEXT:    [[TMP12:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]] to i8*
// CHECK2-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP12]], %struct.BaseS1* [[TMP14]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT12]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT12]], [[TMP10]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYINIT_DONE14]], label [[OMP_ARRAYINIT_BODY7]]
// CHECK2:       omp.arrayinit.done14:
// CHECK2-NEXT:    [[LHS_BEGIN:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[TMP3]] to %struct.S.0*
// CHECK2-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [10 x [4 x %struct.S.0]]* [[ARRS5]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP16]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP17]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP18]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP19]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP15:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]]
// CHECK2-NEXT:    br i1 [[CMP15]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP22]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP23:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA3]], i64 [[TMP23]]
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP24]] to i64
// CHECK2-NEXT:    [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX]], i64 [[IDXPROM]]
// CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[ARRAYIDX16]], align 4
// CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP25]], 1
// CHECK2-NEXT:    store i32 [[INC]], i32* [[ARRAYIDX16]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD17:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK2-NEXT:    store i32 [[ADD17]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP28]])
// CHECK2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast i32* [[VLA3]] to i8*
// CHECK2-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK2-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP32:%.*]] = inttoptr i64 [[TMP6]] to i8*
// CHECK2-NEXT:    store i8* [[TMP32]], i8** [[TMP31]], align 8
// CHECK2-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP34:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT:    store i8* [[TMP34]], i8** [[TMP33]], align 8
// CHECK2-NEXT:    [[TMP35:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP36:%.*]] = load i32, i32* [[TMP35]], align 4
// CHECK2-NEXT:    [[TMP37:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP38:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], i32 2, i64 24, i8* [[TMP37]], void (i8*, i8*)* @.omp.reduction.reduction_func.14, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP38]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP39:%.*]] = getelementptr i32, i32* [[TMP2]], i64 [[TMP6]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP2]], [[TMP39]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST18:%.*]] = phi i32* [ [[VLA3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST19:%.*]] = phi i32* [ [[TMP2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST18]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST18]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP39]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done22:
// CHECK2-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE32:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK2:       omp.arraycpy.body24:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[OMP_ARRAYCPY_DONE22]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT30:%.*]], [[OMP_ARRAYCPY_BODY24]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[OMP_ARRAYCPY_DONE22]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT29:%.*]], [[OMP_ARRAYCPY_BODY24]] ]
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] to i8*
// CHECK2-NEXT:    [[ADD_PTR27:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK2-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR27]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP43:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST25]] to i8*
// CHECK2-NEXT:    [[ADD_PTR28:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK2-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR28]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP42]], %struct.BaseS1* [[TMP44]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT29]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT30]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE31:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT29]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_DONE32]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK2:       omp.arraycpy.done32:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP45:%.*]] = getelementptr i32, i32* [[TMP2]], i64 [[TMP6]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY33:%.*]] = icmp eq i32* [[TMP2]], [[TMP45]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY33]], label [[OMP_ARRAYCPY_DONE40:%.*]], label [[OMP_ARRAYCPY_BODY34:%.*]]
// CHECK2:       omp.arraycpy.body34:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST35:%.*]] = phi i32* [ [[VLA3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT38:%.*]], [[OMP_ARRAYCPY_BODY34]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST36:%.*]] = phi i32* [ [[TMP2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT37:%.*]], [[OMP_ARRAYCPY_BODY34]] ]
// CHECK2-NEXT:    [[TMP46:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP47:%.*]] = load i32, i32* [[TMP46]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST36]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST35]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT37]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST36]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT38]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST35]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE39:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT37]], [[TMP45]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE39]], label [[OMP_ARRAYCPY_DONE40]], label [[OMP_ARRAYCPY_BODY34]]
// CHECK2:       omp.arraycpy.done40:
// CHECK2-NEXT:    [[TMP48:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY41:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP48]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY41]], label [[OMP_ARRAYCPY_DONE50:%.*]], label [[OMP_ARRAYCPY_BODY42:%.*]]
// CHECK2:       omp.arraycpy.body42:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST43:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[OMP_ARRAYCPY_DONE40]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT48:%.*]], [[OMP_ARRAYCPY_BODY42]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST44:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[OMP_ARRAYCPY_DONE40]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT47:%.*]], [[OMP_ARRAYCPY_BODY42]] ]
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP50:%.*]] = load i32, i32* [[TMP49]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP51:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST44]] to i8*
// CHECK2-NEXT:    [[ADD_PTR45:%.*]] = getelementptr inbounds i8, i8* [[TMP51]], i64 4
// CHECK2-NEXT:    [[TMP52:%.*]] = bitcast i8* [[ADD_PTR45]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP53:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST43]] to i8*
// CHECK2-NEXT:    [[ADD_PTR46:%.*]] = getelementptr inbounds i8, i8* [[TMP53]], i64 4
// CHECK2-NEXT:    [[TMP54:%.*]] = bitcast i8* [[ADD_PTR46]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP52]], %struct.BaseS1* [[TMP54]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT47]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST44]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT48]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST43]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE49:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT47]], [[TMP48]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE49]], label [[OMP_ARRAYCPY_DONE50]], label [[OMP_ARRAYCPY_BODY42]]
// CHECK2:       omp.arraycpy.done50:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP36]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[ARRAY_BEGIN51:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS5]], i32 0, i32 0, i32 0
// CHECK2-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN51]], i64 40
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN51]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE52:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done52:
// CHECK2-NEXT:    [[TMP56:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    call void @llvm.stackrestore(i8* [[TMP56]])
// CHECK2-NEXT:    [[TMP57:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP58:%.*]] = load i32, i32* [[TMP57]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP58]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.14
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr i32, i32* [[TMP11]], i64 [[TMP14]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i32* [[TMP11]], [[TMP21]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i32* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i32* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    call void @.omp_combiner..10(i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, i32* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i32* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done2:
// CHECK2-NEXT:    [[TMP22:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP20]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq %struct.S.0* [[TMP20]], [[TMP22]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK2:       omp.arraycpy.body4:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi %struct.S.0* [ [[TMP20]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK2-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP23]], i64 4
// CHECK2-NEXT:    [[TMP24:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP25:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]] to i8*
// CHECK2-NEXT:    [[ADD_PTR7:%.*]] = getelementptr inbounds i8, i8* [[TMP25]], i64 4
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i8* [[ADD_PTR7]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP24]], %struct.BaseS1* [[TMP26]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT9]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP22]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK2:       omp.arraycpy.done11:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.S.0*** nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VAR2_ADDR:%.*]] = alloca %struct.S.0***, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[_TMP6:%.*]] = alloca %struct.S.0**, align 8
// CHECK2-NEXT:    [[_TMP7:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store %struct.S.0*** [[VAR2]], %struct.S.0**** [[VAR2_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load %struct.S.0***, %struct.S.0**** [[VAR2_ADDR]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP1:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds %struct.S.0*, %struct.S.0** [[TMP1]], i64 0
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[ARRAYIDX]], align 8
// CHECK2-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP2]], i64 1
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK2-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds %struct.S.0*, %struct.S.0** [[TMP3]], i64 4
// CHECK2-NEXT:    [[TMP4:%.*]] = load %struct.S.0*, %struct.S.0** [[ARRAYIDX2]], align 8
// CHECK2-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[TMP4]], i64 6
// CHECK2-NEXT:    [[TMP5:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX3]] to i64
// CHECK2-NEXT:    [[TMP6:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX1]] to i64
// CHECK2-NEXT:    [[TMP7:%.*]] = sub i64 [[TMP5]], [[TMP6]]
// CHECK2-NEXT:    [[TMP8:%.*]] = sdiv exact i64 [[TMP7]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP9:%.*]] = add nuw i64 [[TMP8]], 1
// CHECK2-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP11:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT:    store i8* [[TMP11]], i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    [[VLA:%.*]] = alloca [[STRUCT_S_0]], i64 [[TMP9]], align 16
// CHECK2-NEXT:    store i64 [[TMP9]], i64* [[__VLA_EXPR0]], align 8
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP9]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[VLA]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT5:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR4:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR4]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT5]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT5]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP17:%.*]] = load %struct.S.0**, %struct.S.0*** [[TMP0]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP17]], align 8
// CHECK2-NEXT:    [[TMP19:%.*]] = ptrtoint %struct.S.0* [[TMP18]] to i64
// CHECK2-NEXT:    [[TMP20:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX1]] to i64
// CHECK2-NEXT:    [[TMP21:%.*]] = sub i64 [[TMP19]], [[TMP20]]
// CHECK2-NEXT:    [[TMP22:%.*]] = sdiv exact i64 [[TMP21]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP23:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP22]]
// CHECK2-NEXT:    store %struct.S.0** [[_TMP7]], %struct.S.0*** [[_TMP6]], align 8
// CHECK2-NEXT:    store %struct.S.0* [[TMP23]], %struct.S.0** [[_TMP7]], align 8
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP25]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP26]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP27]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP28]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP29:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP30:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP29]], [[TMP30]]
// CHECK2-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP31]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP32:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP32]], 1
// CHECK2-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP33:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP34:%.*]] = load i32, i32* [[TMP33]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP34]])
// CHECK2-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[VLA]] to i8*
// CHECK2-NEXT:    store i8* [[TMP36]], i8** [[TMP35]], align 8
// CHECK2-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP38:%.*]] = inttoptr i64 [[TMP9]] to i8*
// CHECK2-NEXT:    store i8* [[TMP38]], i8** [[TMP37]], align 8
// CHECK2-NEXT:    [[TMP39:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP40:%.*]] = load i32, i32* [[TMP39]], align 4
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast [2 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP42:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], i32 1, i64 16, i8* [[TMP41]], void (i8*, i8*)* @.omp.reduction.reduction_func.16, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP42]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP43:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX1]], [[TMP43]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE16:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi %struct.S.0* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT14:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP44:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] to i8*
// CHECK2-NEXT:    [[ADD_PTR12:%.*]] = getelementptr inbounds i8, i8* [[TMP44]], i64 4
// CHECK2-NEXT:    [[TMP45:%.*]] = bitcast i8* [[ADD_PTR12]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP46:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]] to i8*
// CHECK2-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP46]], i64 4
// CHECK2-NEXT:    [[TMP47:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP45]], %struct.BaseS1* [[TMP47]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT14]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE15:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT14]], [[TMP43]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done16:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP48:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX1]], [[TMP48]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY17]], label [[OMP_ARRAYCPY_DONE26:%.*]], label [[OMP_ARRAYCPY_BODY18:%.*]]
// CHECK2:       omp.arraycpy.body18:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP50:%.*]] = load i32, i32* [[TMP49]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP51:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK2-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP51]], i64 4
// CHECK2-NEXT:    [[TMP52:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP53:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK2-NEXT:    [[ADD_PTR22:%.*]] = getelementptr inbounds i8, i8* [[TMP53]], i64 4
// CHECK2-NEXT:    [[TMP54:%.*]] = bitcast i8* [[ADD_PTR22]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP52]], %struct.BaseS1* [[TMP54]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP50]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT24]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE25:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP48]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_BODY18]]
// CHECK2:       omp.arraycpy.done26:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP40]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[VLA]], i64 [[TMP9]]
// CHECK2-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[VLA]], [[TMP55]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE27:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP55]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[VLA]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE27]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done27:
// CHECK2-NEXT:    [[TMP56:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    call void @llvm.stackrestore(i8* [[TMP56]])
// CHECK2-NEXT:    [[TMP57:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP58:%.*]] = load i32, i32* [[TMP57]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP58]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.16
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [2 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [2 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = ptrtoint i8* [[TMP13]] to i64
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 [[TMP14]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP15]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP16]], i64 4
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP18:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP18]], i64 4
// CHECK2-NEXT:    [[TMP19:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP17]], %struct.BaseS1* [[TMP19]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done3:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..17
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [5 x %struct.S.0]* nonnull align 4 dereferenceable(60) [[VVAR2:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VVAR2_ADDR:%.*]] = alloca [5 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[VVAR22:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [5 x %struct.S.0]* [[VVAR2]], [5 x %struct.S.0]** [[VVAR2_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load [5 x %struct.S.0]*, [5 x %struct.S.0]** [[VVAR2_ADDR]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[TMP0]], i64 0, i64 0
// CHECK2-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[TMP0]], i64 0, i64 4
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR22]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP1:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 5
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP1]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[TMP2:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 4
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP4:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR3:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR3]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP3]], %struct.BaseS1* [[TMP5]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP1]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP6:%.*]] = bitcast [5 x %struct.S.0]* [[TMP0]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP7:%.*]] = ptrtoint %struct.S.0* [[TMP6]] to i64
// CHECK2-NEXT:    [[TMP8:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX]] to i64
// CHECK2-NEXT:    [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]]
// CHECK2-NEXT:    [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast [5 x %struct.S.0]* [[VVAR22]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[TMP11]], i64 [[TMP10]]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[TMP12]] to [5 x %struct.S.0]*
// CHECK2-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [5 x %struct.S.0]* [[VVAR22]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP14:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[TMP14]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP15]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP16]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP18]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP19]], [[TMP20]]
// CHECK2-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP21]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK2-NEXT:    store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP23:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[TMP23]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP24]])
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP30:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], i32 1, i64 8, i8* [[TMP29]], void (i8*, i8*)* @.omp.reduction.reduction_func.18, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP30]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP31:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 5
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP31]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE13:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST7:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST8]] to i8*
// CHECK2-NEXT:    [[ADD_PTR9:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR9]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP34:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 4
// CHECK2-NEXT:    [[TMP35:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP33]], %struct.BaseS1* [[TMP35]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST8]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST7]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE12:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP31]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE12]], label [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done13:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP36:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 5
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY14:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP36]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY14]], label [[OMP_ARRAYCPY_DONE23:%.*]], label [[OMP_ARRAYCPY_BODY15:%.*]]
// CHECK2:       omp.arraycpy.body15:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST16:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT21:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST17:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY15]] ]
// CHECK2-NEXT:    [[TMP37:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP38:%.*]] = load i32, i32* [[TMP37]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP38]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP39:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST17]] to i8*
// CHECK2-NEXT:    [[ADD_PTR18:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 4
// CHECK2-NEXT:    [[TMP40:%.*]] = bitcast i8* [[ADD_PTR18]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST16]] to i8*
// CHECK2-NEXT:    [[ADD_PTR19:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK2-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR19]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP40]], %struct.BaseS1* [[TMP42]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP38]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST17]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT21]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST16]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE22:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP36]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_DONE23]], label [[OMP_ARRAYCPY_BODY15]]
// CHECK2:       omp.arraycpy.done23:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP28]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[ARRAY_BEGIN24:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR22]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN24]], i64 5
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP43]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN24]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE25:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done25:
// CHECK2-NEXT:    [[TMP44:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP45:%.*]] = load i32, i32* [[TMP44]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP45]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.18
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 5
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done3:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..19
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[VAR3:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VAR3_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[VAR34:%.*]] = alloca [2 x %struct.S.0], align 16
// CHECK2-NEXT:    [[_TMP7:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[VAR3]], [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[TMP0]], [4 x %struct.S.0]** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[TMP]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[TMP1]], [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP2:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP2]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP3:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[VAR34]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP4]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT6:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP5]], i64 4
// CHECK2-NEXT:    [[TMP6:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP6]], %struct.BaseS1* [[TMP8]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT6]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT6]], [[TMP4]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP9:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast [4 x %struct.S.0]* [[TMP9]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP11:%.*]] = ptrtoint %struct.S.0* [[TMP10]] to i64
// CHECK2-NEXT:    [[TMP12:%.*]] = ptrtoint %struct.S.0* [[ARRAYIDX]] to i64
// CHECK2-NEXT:    [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (%struct.S.0* getelementptr ([[STRUCT_S_0]], %struct.S.0* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast [2 x %struct.S.0]* [[VAR34]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[TMP15]], i64 [[TMP14]]
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast %struct.S.0* [[TMP16]] to [4 x %struct.S.0]*
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[TMP17]], [4 x %struct.S.0]** [[_TMP7]], align 8
// CHECK2-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [2 x %struct.S.0]* [[VAR34]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP19]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP21]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP22]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]]
// CHECK2-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP25]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK2-NEXT:    store i32 [[ADD9]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP28]])
// CHECK2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK2-NEXT:    [[TMP31:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP32:%.*]] = load i32, i32* [[TMP31]], align 4
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP34:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], i32 1, i64 8, i8* [[TMP33]], void (i8*, i8*)* @.omp.reduction.reduction_func.20, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP34]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP35:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 2
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP35]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE16:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST10:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT14:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP36:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]] to i8*
// CHECK2-NEXT:    [[ADD_PTR12:%.*]] = getelementptr inbounds i8, i8* [[TMP36]], i64 4
// CHECK2-NEXT:    [[TMP37:%.*]] = bitcast i8* [[ADD_PTR12]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP38:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]] to i8*
// CHECK2-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 4
// CHECK2-NEXT:    [[TMP39:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP37]], %struct.BaseS1* [[TMP39]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT14]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST10]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE15:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT14]], [[TMP35]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done16:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[ARRAYIDX]], i64 2
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY17:%.*]] = icmp eq %struct.S.0* [[ARRAYIDX]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY17]], label [[OMP_ARRAYCPY_DONE26:%.*]], label [[OMP_ARRAYCPY_BODY18:%.*]]
// CHECK2:       omp.arraycpy.body18:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST19:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST20:%.*]] = phi %struct.S.0* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY18]] ]
// CHECK2-NEXT:    [[TMP41:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP42:%.*]] = load i32, i32* [[TMP41]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP42]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP43:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]] to i8*
// CHECK2-NEXT:    [[ADD_PTR21:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK2-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR21]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP45:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]] to i8*
// CHECK2-NEXT:    [[ADD_PTR22:%.*]] = getelementptr inbounds i8, i8* [[TMP45]], i64 4
// CHECK2-NEXT:    [[TMP46:%.*]] = bitcast i8* [[ADD_PTR22]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP44]], %struct.BaseS1* [[TMP46]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP42]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST20]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT24]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST19]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE25:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_BODY18]]
// CHECK2:       omp.arraycpy.done26:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP32]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[ARRAY_BEGIN27:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[VAR34]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN27]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP47]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN27]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE28:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done28:
// CHECK2-NEXT:    [[TMP48:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP49:%.*]] = load i32, i32* [[TMP48]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP49]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.20
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 2
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done3:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..21
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [4 x %struct.S.0]* nonnull align 4 dereferenceable(48) [[VAR3:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VAR3_ADDR:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP5:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[VAR3]], [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3_ADDR]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[TMP0]], [4 x %struct.S.0]** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[TMP]], align 8
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[TMP1]], [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 9, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP2:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4
// CHECK2-NEXT:    [[DOTVAR3__VOID_ADDR:%.*]] = call i8* @__kmpc_alloc(i32 [[TMP4]], i64 48, i8* inttoptr (i64 6 to i8*))
// CHECK2-NEXT:    [[DOTVAR3__ADDR:%.*]] = bitcast i8* [[DOTVAR3__VOID_ADDR]] to [4 x %struct.S.0]*
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[DOTVAR3__ADDR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast [4 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[ARRAY_BEGIN]], i64 4
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[TMP7:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP9:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR3:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 4
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast i8* [[ADD_PTR3]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP8]], %struct.BaseS1* [[TMP10]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP6]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    store [4 x %struct.S.0]* [[DOTVAR3__ADDR]], [4 x %struct.S.0]** [[_TMP5]], align 8
// CHECK2-NEXT:    [[LHS_BEGIN:%.*]] = bitcast [4 x %struct.S.0]* [[TMP2]] to %struct.S.0*
// CHECK2-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [4 x %struct.S.0]* [[DOTVAR3__ADDR]] to %struct.S.0*
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP11]], 9
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]]
// CHECK2-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK2-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]])
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP19:%.*]] = bitcast %struct.S.0* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT:    store i8* [[TMP19]], i8** [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP21:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, i8* [[TMP20]], void (i8*, i8*)* @.omp.reduction.reduction_func.22, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP21]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP22:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 4
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP22]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST8:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP23:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]] to i8*
// CHECK2-NEXT:    [[ADD_PTR10:%.*]] = getelementptr inbounds i8, i8* [[TMP23]], i64 4
// CHECK2-NEXT:    [[TMP24:%.*]] = bitcast i8* [[ADD_PTR10]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP25:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]] to i8*
// CHECK2-NEXT:    [[ADD_PTR11:%.*]] = getelementptr inbounds i8, i8* [[TMP25]], i64 4
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i8* [[ADD_PTR11]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP24]], %struct.BaseS1* [[TMP26]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT12]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST8]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT12]], [[TMP22]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done14:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[LHS_BEGIN]], i64 4
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY15:%.*]] = icmp eq %struct.S.0* [[LHS_BEGIN]], [[TMP27]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY15]], label [[OMP_ARRAYCPY_DONE24:%.*]], label [[OMP_ARRAYCPY_BODY16:%.*]]
// CHECK2:       omp.arraycpy.body16:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST17:%.*]] = phi %struct.S.0* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT22:%.*]], [[OMP_ARRAYCPY_BODY16]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST18:%.*]] = phi %struct.S.0* [ [[LHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT21:%.*]], [[OMP_ARRAYCPY_BODY16]] ]
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP28:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST18]] to i8*
// CHECK2-NEXT:    [[ADD_PTR19:%.*]] = getelementptr inbounds i8, i8* [[TMP28]], i64 4
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast i8* [[ADD_PTR19]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST17]] to i8*
// CHECK2-NEXT:    [[ADD_PTR20:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK2-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR20]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP29]], %struct.BaseS1* [[TMP31]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT21]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST18]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT22]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST17]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE23:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT21]], [[TMP27]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE23]], label [[OMP_ARRAYCPY_DONE24]], label [[OMP_ARRAYCPY_BODY16]]
// CHECK2:       omp.arraycpy.done24:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP4]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[ARRAY_BEGIN25:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[DOTVAR3__ADDR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN25]], i64 4
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP32]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN25]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE26:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done26:
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast [4 x %struct.S.0]* [[DOTVAR3__ADDR]] to i8*
// CHECK2-NEXT:    call void @__kmpc_free(i32 [[TMP4]], i8* [[TMP33]], i8* inttoptr (i64 6 to i8*))
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP4]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.22
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S.0*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S_0:%.*]], %struct.S.0* [[TMP11]], i64 4
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S.0* [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], %struct.S.0* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S.0* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done3:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v
// CHECK2-SAME: () #[[ATTR0]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[T:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK2-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[T_VAR1:%.*]] = 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:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK2-NEXT:    [[ARR:%.*]] = alloca [42 x %struct.S], align 16
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[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._Z5tmainIiLi42EET_v.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 @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK2-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK2-NEXT:    store %struct.S* [[TEST]], %struct.S** [[VAR]], align 8
// CHECK2-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 42
// 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 @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[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:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 6, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, %struct.S*, %struct.S*, i32*, [2 x i32]*, [2 x %struct.S]*)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i32* [[T_VAR]], %struct.S* [[TMP1]], %struct.S* [[VAR1]], i32* [[T_VAR1]], [2 x i32]* [[VEC]], [2 x %struct.S]* [[S_ARR]])
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// 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*, i32*, [2 x i32]*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* [[T_VAR]], [2 x i32]* [[VEC]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[TMP2]])
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [42 x %struct.S]*, [2 x i32]*, i32*, [2 x %struct.S]*, %struct.S*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), [42 x %struct.S]* [[ARR]], [2 x i32]* [[VEC]], i32* [[T_VAR]], [2 x %struct.S]* [[S_ARR]], %struct.S* [[TMP3]])
// CHECK2-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    [[ARRAY_BEGIN1:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN1]], i64 42
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP4]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN1]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done2:
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN3]], i64 2
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY4:%.*]]
// CHECK2:       arraydestroy.body4:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST5:%.*]] = phi %struct.S* [ [[TMP5]], [[ARRAYDESTROY_DONE2]] ], [ [[ARRAYDESTROY_ELEMENT6:%.*]], [[ARRAYDESTROY_BODY4]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT6]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST5]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT6]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE7:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT6]], [[ARRAY_BEGIN3]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE7]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY4]]
// CHECK2:       arraydestroy.done8:
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR4]]
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT:    ret i32 [[TMP6]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK2-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK2-NEXT:    [[CONV:%.*]] = fptrunc double [[TMP4]] to float
// CHECK2-NEXT:    store float [[CONV]], float* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK2-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK2-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[CONV:%.*]] = fpext float [[TMP4]] to double
// CHECK2-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK2-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK2-NEXT:    [[CONV2:%.*]] = fptrunc double [[ADD]] to float
// CHECK2-NEXT:    store float [[CONV2]], float* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK2-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[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
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]], i32 [[TMP0]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..23
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR1:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[VAR1_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK2-NEXT:    [[_TMP6:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[VAR17:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK2-NEXT:    [[T_VAR18:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[VAR1]], %struct.S** [[VAR1_ADDR]], align 8
// CHECK2-NEXT:    store i32* [[T_VAR1]], i32** [[T_VAR1_ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[VAR1_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[T_VAR1_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP6:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP6]], %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    call void @.omp_initializer..25(i32* [[T_VAR3]], i32* [[TMP0]])
// CHECK2-NEXT:    [[TMP7:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP8]], i64 4
// CHECK2-NEXT:    [[TMP9:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 4
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..5(%struct.BaseS1* [[TMP9]], %struct.BaseS1* [[TMP11]])
// CHECK2-NEXT:    store %struct.S* [[VAR4]], %struct.S** [[_TMP6]], align 8
// CHECK2-NEXT:    call void @.omp_initializer..27(%struct.S* [[VAR17]], %struct.S* [[TMP2]])
// CHECK2-NEXT:    call void @.omp_initializer..29(i32* [[T_VAR18]], i32* [[TMP3]])
// CHECK2-NEXT:    [[TMP12:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[TMP12]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP14]], 1
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP16:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP16]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP9:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
// CHECK2-NEXT:    br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[T_VAR3]], align 4
// CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT:    store i32 [[TMP20]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT:    [[TMP22:%.*]] = load %struct.S*, %struct.S** [[_TMP6]], align 8
// CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP23]] to i64
// CHECK2-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP5]], i64 0, i64 [[IDXPROM10]]
// CHECK2-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX11]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP22]])
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK2-NEXT:    store i32 [[ADD12]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP13]])
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i32* [[T_VAR3]] to i8*
// CHECK2-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP28:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK2-NEXT:    store i8* [[TMP28]], i8** [[TMP27]], align 8
// CHECK2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast %struct.S* [[VAR17]] to i8*
// CHECK2-NEXT:    store i8* [[TMP30]], i8** [[TMP29]], align 8
// CHECK2-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast i32* [[T_VAR18]] to i8*
// CHECK2-NEXT:    store i8* [[TMP32]], i8** [[TMP31]], align 8
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast [4 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP34:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], i32 4, i64 32, i8* [[TMP33]], void (i8*, i8*)* @.omp.reduction.reduction_func.30, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP34]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    call void @.omp_combiner..24(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK2-NEXT:    [[TMP35:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR13:%.*]] = getelementptr inbounds i8, i8* [[TMP35]], i64 4
// CHECK2-NEXT:    [[TMP36:%.*]] = bitcast i8* [[ADD_PTR13]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP37:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP37]], i64 4
// CHECK2-NEXT:    [[TMP38:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP36]], %struct.BaseS1* [[TMP38]])
// CHECK2-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP2]], %struct.S* [[VAR17]])
// CHECK2-NEXT:    call void @.omp_combiner..28(i32* [[TMP3]], i32* [[T_VAR18]])
// CHECK2-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..24(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP39:%.*]] = bitcast %struct.S* [[TMP7]] to i8*
// CHECK2-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 4
// CHECK2-NEXT:    [[TMP40:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast %struct.S* [[VAR4]] to i8*
// CHECK2-NEXT:    [[ADD_PTR16:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK2-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR16]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP40]], %struct.BaseS1* [[TMP42]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP2]], %struct.S* [[VAR17]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..28(i32* [[TMP3]], i32* [[T_VAR18]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP13]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR17]]) #[[ATTR4]]
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR4]]) #[[ATTR4]]
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..24
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 1513, [[TMP4]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..25
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32 321, i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..26
// CHECK2-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK2-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK2-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..27
// CHECK2-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = bitcast %struct.S* [[TMP3]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP4]], i64 4
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP6:%.*]] = bitcast %struct.S* [[TMP2]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 4
// CHECK2-NEXT:    [[TMP7:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @_Z5init2R6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP5]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP7]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..28
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[SUB:%.*]] = sub nsw i32 47, [[TMP4]]
// CHECK2-NEXT:    store i32 [[SUB]], i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..29
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[DIV:%.*]] = sdiv i32 432, [[TMP4]]
// CHECK2-NEXT:    store i32 [[DIV]], i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.30
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [4 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[TMP13]] to %struct.S*
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = bitcast i8* [[TMP16]] to %struct.S*
// CHECK2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i8* [[TMP19]] to %struct.S*
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP22:%.*]] = load i8*, i8** [[TMP21]], align 8
// CHECK2-NEXT:    [[TMP23:%.*]] = bitcast i8* [[TMP22]] to %struct.S*
// CHECK2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP5]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i8* [[TMP25]] to i32*
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[TMP3]], i64 0, i64 3
// CHECK2-NEXT:    [[TMP28:%.*]] = load i8*, i8** [[TMP27]], align 8
// CHECK2-NEXT:    [[TMP29:%.*]] = bitcast i8* [[TMP28]] to i32*
// CHECK2-NEXT:    call void @.omp_combiner..24(i32* [[TMP11]], i32* [[TMP8]])
// CHECK2-NEXT:    [[TMP30:%.*]] = bitcast %struct.S* [[TMP17]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP30]], i64 4
// CHECK2-NEXT:    [[TMP31:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast %struct.S* [[TMP14]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 4
// CHECK2-NEXT:    [[TMP33:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..4(%struct.BaseS1* [[TMP31]], %struct.BaseS1* [[TMP33]])
// CHECK2-NEXT:    call void @.omp_combiner..26(%struct.S* [[TMP23]], %struct.S* [[TMP20]])
// CHECK2-NEXT:    call void @.omp_combiner..28(i32* [[TMP29]], i32* [[TMP26]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..31
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[VEC_ADDR:%.*]] = alloca [2 x i32]*, align 8
// CHECK2-NEXT:    [[S_ARR_ADDR:%.*]] = alloca [2 x %struct.S]*, align 8
// CHECK2-NEXT:    [[VAR_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    store [2 x i32]* [[VEC]], [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    store [2 x %struct.S]* [[S_ARR]], [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[VAR]], %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_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:    store %struct.S* [[TMP3]], %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP4]], %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* @.init.32, align 4
// CHECK2-NEXT:    store i32 [[TMP5]], i32* [[T_VAR3]], align 4
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP8]], 1
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP10:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP10]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP13]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP14:%.*]] = load i32, i32* [[T_VAR3]], align 4
// CHECK2-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT:    store i32 [[TMP14]], i32* [[ARRAYIDX]], align 4
// CHECK2-NEXT:    [[TMP16:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM5:%.*]] = sext i32 [[TMP17]] to i64
// CHECK2-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP2]], i64 0, i64 [[IDXPROM5]]
// CHECK2-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX6]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP16]])
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1
// CHECK2-NEXT:    store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP7]])
// CHECK2-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP20:%.*]] = bitcast i32* [[T_VAR3]] to i8*
// CHECK2-NEXT:    store i8* [[TMP20]], i8** [[TMP19]], align 8
// CHECK2-NEXT:    [[TMP21:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP22:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], i32 1, i64 8, i8* [[TMP21]], void (i8*, i8*)* @.omp.reduction.reduction_func.33, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP22]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    call void @.omp_combiner..34(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @.omp_combiner..34(i32* [[TMP0]], i32* [[T_VAR3]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP7]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP7]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.33
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK2-NEXT:    call void @.omp_combiner..34(i32* [[TMP11]], i32* [[TMP8]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..34
// CHECK2-SAME: (i32* noalias [[TMP0:%.*]], i32* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    store i32* [[TMP0]], i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i32* [[TMP1]], i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i32*, i32** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK2-NEXT:    store i32 [[MUL]], i32* [[TMP3]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..35
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], [42 x %struct.S]* nonnull align 4 dereferenceable(504) [[ARR:%.*]], [2 x i32]* nonnull align 4 dereferenceable(8) [[VEC:%.*]], i32* nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], [2 x %struct.S]* nonnull align 4 dereferenceable(24) [[S_ARR:%.*]], %struct.S* nonnull align 4 dereferenceable(12) [[VAR:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[ARR_ADDR:%.*]] = alloca [42 x %struct.S]*, 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:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[ARR4:%.*]] = alloca [40 x %struct.S], align 16
// CHECK2-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT:    store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT:    store [42 x %struct.S]* [[ARR]], [42 x %struct.S]** [[ARR_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 [42 x %struct.S]*, [42 x %struct.S]** [[ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load [2 x i32]*, [2 x i32]** [[VEC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load [2 x %struct.S]*, [2 x %struct.S]** [[S_ARR_ADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load %struct.S*, %struct.S** [[VAR_ADDR]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP4]], %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK2-NEXT:    store %struct.S* [[TMP5]], %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[TMP0]], i64 0, i64 1
// CHECK2-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[TMP0]], i64 0, i64 40
// CHECK2-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [40 x %struct.S], [40 x %struct.S]* [[ARR4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[ARRAY_BEGIN]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAY_BEGIN]], [[TMP6]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT6:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    [[TMP7:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 4
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP9:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR5:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 4
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast i8* [[ADD_PTR5]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_initializer..37(%struct.BaseS1* [[TMP8]], %struct.BaseS1* [[TMP10]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT6]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT6]], [[TMP6]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast [42 x %struct.S]* [[TMP0]] to %struct.S*
// CHECK2-NEXT:    [[TMP12:%.*]] = ptrtoint %struct.S* [[TMP11]] to i64
// CHECK2-NEXT:    [[TMP13:%.*]] = ptrtoint %struct.S* [[ARRAYIDX]] to i64
// CHECK2-NEXT:    [[TMP14:%.*]] = sub i64 [[TMP12]], [[TMP13]]
// CHECK2-NEXT:    [[TMP15:%.*]] = sdiv exact i64 [[TMP14]], ptrtoint (%struct.S* getelementptr ([[STRUCT_S]], %struct.S* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast [40 x %struct.S]* [[ARR4]] to %struct.S*
// CHECK2-NEXT:    [[TMP17:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[TMP16]], i64 [[TMP15]]
// CHECK2-NEXT:    [[TMP18:%.*]] = bitcast %struct.S* [[TMP17]] to [42 x %struct.S]*
// CHECK2-NEXT:    [[RHS_BEGIN:%.*]] = bitcast [40 x %struct.S]* [[ARR4]] to %struct.S*
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP20]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT:    [[TMP21:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP21]], 1
// CHECK2-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK2:       cond.true:
// CHECK2-NEXT:    br label [[COND_END:%.*]]
// CHECK2:       cond.false:
// CHECK2-NEXT:    [[TMP22:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP22]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[TMP23:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT:    store i32 [[TMP23]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP24:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[TMP25:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP24]], [[TMP25]]
// CHECK2-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK2:       omp.inner.for.cond.cleanup:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK2:       omp.inner.for.body:
// CHECK2-NEXT:    [[TMP26:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP26]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP28]] to i64
// CHECK2-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK2-NEXT:    store i32 [[TMP27]], i32* [[ARRAYIDX8]], align 4
// CHECK2-NEXT:    [[TMP29:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = load i32, i32* [[I]], align 4
// CHECK2-NEXT:    [[IDXPROM9:%.*]] = sext i32 [[TMP30]] to i64
// CHECK2-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[TMP3]], i64 0, i64 [[IDXPROM9]]
// CHECK2-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX10]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP29]])
// CHECK2-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK2:       omp.body.continue:
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2:       omp.inner.for.inc:
// CHECK2-NEXT:    [[TMP31:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP31]], 1
// CHECK2-NEXT:    store i32 [[ADD11]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK2:       omp.inner.for.end:
// CHECK2-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK2:       omp.loop.exit:
// CHECK2-NEXT:    [[TMP32:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP33:%.*]] = load i32, i32* [[TMP32]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP33]])
// CHECK2-NEXT:    [[TMP34:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP35:%.*]] = bitcast %struct.S* [[RHS_BEGIN]] to i8*
// CHECK2-NEXT:    store i8* [[TMP35]], i8** [[TMP34]], align 8
// CHECK2-NEXT:    [[TMP36:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP37:%.*]] = load i32, i32* [[TMP36]], align 4
// CHECK2-NEXT:    [[TMP38:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP39:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], i32 1, i64 8, i8* [[TMP38]], void (i8*, i8*)* @.omp.reduction.reduction_func.38, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP39]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK2-NEXT:    ]
// CHECK2:       .omp.reduction.case1:
// CHECK2-NEXT:    [[TMP40:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAYIDX]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[ARRAYIDX]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST12:%.*]] = phi %struct.S* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST13:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP41:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST13]] to i8*
// CHECK2-NEXT:    [[ADD_PTR14:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 4
// CHECK2-NEXT:    [[TMP42:%.*]] = bitcast i8* [[ADD_PTR14]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP43:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST12]] to i8*
// CHECK2-NEXT:    [[ADD_PTR15:%.*]] = getelementptr inbounds i8, i8* [[TMP43]], i64 4
// CHECK2-NEXT:    [[TMP44:%.*]] = bitcast i8* [[ADD_PTR15]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP42]], %struct.BaseS1* [[TMP44]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST13]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST12]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP40]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done18:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP45:%.*]] = getelementptr [[STRUCT_S]], %struct.S* [[ARRAYIDX]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq %struct.S* [[ARRAYIDX]], [[TMP45]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE28:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]]
// CHECK2:       omp.arraycpy.body20:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi %struct.S* [ [[RHS_BEGIN]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT26:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi %struct.S* [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT25:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK2-NEXT:    [[TMP46:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP47:%.*]] = load i32, i32* [[TMP46]], align 4
// CHECK2-NEXT:    call void @__kmpc_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[TMP48:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST22]] to i8*
// CHECK2-NEXT:    [[ADD_PTR23:%.*]] = getelementptr inbounds i8, i8* [[TMP48]], i64 4
// CHECK2-NEXT:    [[TMP49:%.*]] = bitcast i8* [[ADD_PTR23]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP50:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST21]] to i8*
// CHECK2-NEXT:    [[ADD_PTR24:%.*]] = getelementptr inbounds i8, i8* [[TMP50]], i64 4
// CHECK2-NEXT:    [[TMP51:%.*]] = bitcast i8* [[ADD_PTR24]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP49]], %struct.BaseS1* [[TMP51]])
// CHECK2-NEXT:    call void @__kmpc_end_critical(%struct.ident_t* @[[GLOB1]], i32 [[TMP47]], [8 x i32]* @.gomp_critical_user_.atomic_reduction.var)
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT25]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT26]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE27:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT25]], [[TMP45]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE27]], label [[OMP_ARRAYCPY_DONE28]], label [[OMP_ARRAYCPY_BODY20]]
// CHECK2:       omp.arraycpy.done28:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP37]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[ARRAY_BEGIN29:%.*]] = getelementptr inbounds [40 x %struct.S], [40 x %struct.S]* [[ARR4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN29]], i64 40
// CHECK2-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK2:       arraydestroy.body:
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP52]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK2-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK2-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK2-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN29]]
// CHECK2-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE30:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK2:       arraydestroy.done30:
// CHECK2-NEXT:    [[TMP53:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP54:%.*]] = load i32, i32* [[TMP53]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4]], i32 [[TMP54]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_combiner..36
// CHECK2-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    call void @_Z8red_plusR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_initializer..37
// CHECK2-SAME: (%struct.BaseS1* noalias [[TMP0:%.*]], %struct.BaseS1* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.BaseS1*, align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP0]], %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    store %struct.BaseS1* [[TMP1]], %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.BaseS1*, %struct.BaseS1** [[DOTADDR]], align 8
// CHECK2-NEXT:    call void @_Z9init_plusR6BaseS1RKS_(%struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP3]], %struct.BaseS1* nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.38
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR8]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.S*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast i8* [[TMP10]] to %struct.S*
// CHECK2-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S:%.*]], %struct.S* [[TMP11]], i64 40
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq %struct.S* [[TMP11]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP8]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP13:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i8, i8* [[TMP13]], i64 4
// CHECK2-NEXT:    [[TMP14:%.*]] = bitcast i8* [[ADD_PTR]] to %struct.BaseS1*
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]] to i8*
// CHECK2-NEXT:    [[ADD_PTR2:%.*]] = getelementptr inbounds i8, i8* [[TMP15]], i64 4
// CHECK2-NEXT:    [[TMP16:%.*]] = bitcast i8* [[ADD_PTR2]] to %struct.BaseS1*
// CHECK2-NEXT:    call void @.omp_combiner..36(%struct.BaseS1* [[TMP14]], %struct.BaseS1* [[TMP16]])
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], %struct.S* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq %struct.S* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done3:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK2-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK2-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK2-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK2-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to double
// CHECK2-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK2-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK2-NEXT:    [[CONV2:%.*]] = fptosi double [[ADD]] to i32
// CHECK2-NEXT:    store i32 [[CONV2]], i32* [[F]], align 4
// CHECK2-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z4bazzv
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[S1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]])
// CHECK3-NEXT:    store i32 0, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    call void @.omp_initializer.(%struct.S* [[S1]], %struct.S* [[S]])
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK3-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP0]], 10
// CHECK3-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK3:       omp.inner.for.cond.cleanup:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP1]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK3-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3:       omp.body.continue:
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK3-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK3-NEXT:    store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK3:       omp.inner.for.end:
// CHECK3-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK3-NEXT:    call void @.omp_combiner.(%struct.S* [[S]], %struct.S* [[S1]])
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S1]]) #[[ATTR7:[0-9]+]]
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]]) #[[ATTR7]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_combiner.
// CHECK3-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK3-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK3-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_initializer.
// CHECK3-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK3-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TMP3]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIiED2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR7]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK3-NEXT:    [[T_VAR:%.*]] = alloca float, align 4
// CHECK3-NEXT:    [[T_VAR1:%.*]] = alloca float, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [4 x %struct.S.0], align 16
// CHECK3-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK3-NEXT:    [[ARRS:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK3-NEXT:    [[VAR2:%.*]] = alloca %struct.S.0**, align 8
// CHECK3-NEXT:    [[VVAR2:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK3-NEXT:    [[VAR3:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    [[_TMP10:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK3-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK3-NEXT:    [[I15:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I26:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I37:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[I44:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP51:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK3-NEXT:    [[_TMP52:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK3-NEXT:    [[I53:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP60:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK3-NEXT:    [[_TMP61:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK3-NEXT:    [[I62:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK3-NEXT:    store float 0.000000e+00, float* [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK3-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)
// CHECK3-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK3-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK3-NEXT:    [[ARRAYINIT_ELEMENT1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT]], i64 1
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT1]], float 3.000000e+00)
// CHECK3-NEXT:    [[ARRAYINIT_ELEMENT2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT1]], i64 1
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT2]], float 4.000000e+00)
// CHECK3-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    [[CALL:%.*]] = call %struct.S.0** @_Z3foov()
// CHECK3-NEXT:    store %struct.S.0** [[CALL]], %struct.S.0*** [[VAR2]], align 8
// CHECK3-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN3]], i64 5
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP5:%.*]]
// CHECK3:       arrayctor.loop5:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR6:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYCTOR_NEXT7:%.*]], [[ARRAYCTOR_LOOP5]] ]
// CHECK3-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR6]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT7]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR6]], i64 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE8:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT7]], [[ARRAYCTOR_END4]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE8]], label [[ARRAYCTOR_CONT9:%.*]], label [[ARRAYCTOR_LOOP5]]
// CHECK3:       arrayctor.cont9:
// CHECK3-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK3-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
// CHECK3-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
// CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
// CHECK3-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[_TMP10]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I]], align 4
// CHECK3-NEXT:    br label [[FOR_COND:%.*]]
// CHECK3:       for.cond:
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP3]], 2
// CHECK3-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK3:       for.body:
// CHECK3-NEXT:    [[TMP4:%.*]] = load float, float* [[T_VAR]], align 4
// CHECK3-NEXT:    [[CONV:%.*]] = fptosi float [[TMP4]] to i32
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK3-NEXT:    store i32 [[CONV]], i32* [[ARRAYIDX]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP10]], align 8
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP7]] to i64
// CHECK3-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 [[IDXPROM11]]
// CHECK3-NEXT:    [[CALL13:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S.0* @_ZN1SIfEaSERKS0_(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYIDX12]], %struct.S.0* nonnull align 4 dereferenceable(12) [[TMP6]])
// CHECK3-NEXT:    br label [[FOR_INC:%.*]]
// CHECK3:       for.inc:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK3-NEXT:    store i32 [[INC]], i32* [[I]], align 4
// CHECK3-NEXT:    br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK3:       for.end:
// CHECK3-NEXT:    [[ARRAYIDX14:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 1
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX14]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK3-NEXT:    [[TMP11:%.*]] = call i8* @llvm.stacksave()
// CHECK3-NEXT:    store i8* [[TMP11]], i8** [[SAVED_STACK]], align 8
// CHECK3-NEXT:    [[TMP12:%.*]] = mul nuw i64 10, [[TMP10]]
// CHECK3-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP12]], align 16
// CHECK3-NEXT:    store i64 [[TMP10]], i64* [[__VLA_EXPR0]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I15]], align 4
// CHECK3-NEXT:    br label [[FOR_COND16:%.*]]
// CHECK3:       for.cond16:
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[I15]], align 4
// CHECK3-NEXT:    [[CMP17:%.*]] = icmp slt i32 [[TMP13]], 10
// CHECK3-NEXT:    br i1 [[CMP17]], label [[FOR_BODY18:%.*]], label [[FOR_END25:%.*]]
// CHECK3:       for.body18:
// CHECK3-NEXT:    [[TMP14:%.*]] = mul nsw i64 1, [[TMP10]]
// CHECK3-NEXT:    [[ARRAYIDX19:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[TMP14]]
// CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I15]], align 4
// CHECK3-NEXT:    [[IDXPROM20:%.*]] = sext i32 [[TMP15]] to i64
// CHECK3-NEXT:    [[ARRAYIDX21:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX19]], i64 [[IDXPROM20]]
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[ARRAYIDX21]], align 4
// CHECK3-NEXT:    [[INC22:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK3-NEXT:    store i32 [[INC22]], i32* [[ARRAYIDX21]], align 4
// CHECK3-NEXT:    br label [[FOR_INC23:%.*]]
// CHECK3:       for.inc23:
// CHECK3-NEXT:    [[TMP17:%.*]] = load i32, i32* [[I15]], align 4
// CHECK3-NEXT:    [[INC24:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK3-NEXT:    store i32 [[INC24]], i32* [[I15]], align 4
// CHECK3-NEXT:    br label [[FOR_COND16]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK3:       for.end25:
// CHECK3-NEXT:    store i32 0, i32* [[I26]], align 4
// CHECK3-NEXT:    br label [[FOR_COND27:%.*]]
// CHECK3:       for.cond27:
// CHECK3-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I26]], align 4
// CHECK3-NEXT:    [[CMP28:%.*]] = icmp slt i32 [[TMP18]], 10
// CHECK3-NEXT:    br i1 [[CMP28]], label [[FOR_BODY29:%.*]], label [[FOR_END36:%.*]]
// CHECK3:       for.body29:
// CHECK3-NEXT:    [[TMP19:%.*]] = mul nsw i64 1, [[TMP10]]
// CHECK3-NEXT:    [[ARRAYIDX30:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[TMP19]]
// CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[I26]], align 4
// CHECK3-NEXT:    [[IDXPROM31:%.*]] = sext i32 [[TMP20]] to i64
// CHECK3-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX30]], i64 [[IDXPROM31]]
// CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[ARRAYIDX32]], align 4
// CHECK3-NEXT:    [[INC33:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK3-NEXT:    store i32 [[INC33]], i32* [[ARRAYIDX32]], align 4
// CHECK3-NEXT:    br label [[FOR_INC34:%.*]]
// CHECK3:       for.inc34:
// CHECK3-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I26]], align 4
// CHECK3-NEXT:    [[INC35:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK3-NEXT:    store i32 [[INC35]], i32* [[I26]], align 4
// CHECK3-NEXT:    br label [[FOR_COND27]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK3:       for.end36:
// CHECK3-NEXT:    store i32 0, i32* [[I37]], align 4
// CHECK3-NEXT:    br label [[FOR_COND38:%.*]]
// CHECK3:       for.cond38:
// CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I37]], align 4
// CHECK3-NEXT:    [[CMP39:%.*]] = icmp slt i32 [[TMP23]], 10
// CHECK3-NEXT:    br i1 [[CMP39]], label [[FOR_BODY40:%.*]], label [[FOR_END43:%.*]]
// CHECK3:       for.body40:
// CHECK3-NEXT:    br label [[FOR_INC41:%.*]]
// CHECK3:       for.inc41:
// CHECK3-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I37]], align 4
// CHECK3-NEXT:    [[INC42:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK3-NEXT:    store i32 [[INC42]], i32* [[I37]], align 4
// CHECK3-NEXT:    br label [[FOR_COND38]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK3:       for.end43:
// CHECK3-NEXT:    store i32 0, i32* [[I44]], align 4
// CHECK3-NEXT:    br label [[FOR_COND45:%.*]]
// CHECK3:       for.cond45:
// CHECK3-NEXT:    [[TMP25:%.*]] = load i32, i32* [[I44]], align 4
// CHECK3-NEXT:    [[CMP46:%.*]] = icmp slt i32 [[TMP25]], 10
// CHECK3-NEXT:    br i1 [[CMP46]], label [[FOR_BODY47:%.*]], label [[FOR_END50:%.*]]
// CHECK3:       for.body47:
// CHECK3-NEXT:    br label [[FOR_INC48:%.*]]
// CHECK3:       for.inc48:
// CHECK3-NEXT:    [[TMP26:%.*]] = load i32, i32* [[I44]], align 4
// CHECK3-NEXT:    [[INC49:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK3-NEXT:    store i32 [[INC49]], i32* [[I44]], align 4
// CHECK3-NEXT:    br label [[FOR_COND45]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK3:       for.end50:
// CHECK3-NEXT:    [[TMP27:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK3-NEXT:    store [4 x %struct.S.0]* [[TMP27]], [4 x %struct.S.0]** [[_TMP51]], align 8
// CHECK3-NEXT:    [[TMP28:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP51]], align 8
// CHECK3-NEXT:    store [4 x %struct.S.0]* [[TMP28]], [4 x %struct.S.0]** [[_TMP52]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I53]], align 4
// CHECK3-NEXT:    br label [[FOR_COND54:%.*]]
// CHECK3:       for.cond54:
// CHECK3-NEXT:    [[TMP29:%.*]] = load i32, i32* [[I53]], align 4
// CHECK3-NEXT:    [[CMP55:%.*]] = icmp slt i32 [[TMP29]], 10
// CHECK3-NEXT:    br i1 [[CMP55]], label [[FOR_BODY56:%.*]], label [[FOR_END59:%.*]]
// CHECK3:       for.body56:
// CHECK3-NEXT:    br label [[FOR_INC57:%.*]]
// CHECK3:       for.inc57:
// CHECK3-NEXT:    [[TMP30:%.*]] = load i32, i32* [[I53]], align 4
// CHECK3-NEXT:    [[INC58:%.*]] = add nsw i32 [[TMP30]], 1
// CHECK3-NEXT:    store i32 [[INC58]], i32* [[I53]], align 4
// CHECK3-NEXT:    br label [[FOR_COND54]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK3:       for.end59:
// CHECK3-NEXT:    [[TMP31:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK3-NEXT:    store [4 x %struct.S.0]* [[TMP31]], [4 x %struct.S.0]** [[_TMP60]], align 8
// CHECK3-NEXT:    [[TMP32:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP60]], align 8
// CHECK3-NEXT:    store [4 x %struct.S.0]* [[TMP32]], [4 x %struct.S.0]** [[_TMP61]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I62]], align 4
// CHECK3-NEXT:    br label [[FOR_COND63:%.*]]
// CHECK3:       for.cond63:
// CHECK3-NEXT:    [[TMP33:%.*]] = load i32, i32* [[I62]], align 4
// CHECK3-NEXT:    [[CMP64:%.*]] = icmp slt i32 [[TMP33]], 10
// CHECK3-NEXT:    br i1 [[CMP64]], label [[FOR_BODY65:%.*]], label [[FOR_END68:%.*]]
// CHECK3:       for.body65:
// CHECK3-NEXT:    br label [[FOR_INC66:%.*]]
// CHECK3:       for.inc66:
// CHECK3-NEXT:    [[TMP34:%.*]] = load i32, i32* [[I62]], align 4
// CHECK3-NEXT:    [[INC67:%.*]] = add nsw i32 [[TMP34]], 1
// CHECK3-NEXT:    store i32 [[INC67]], i32* [[I62]], align 4
// CHECK3-NEXT:    br label [[FOR_COND63]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK3:       for.end68:
// CHECK3-NEXT:    [[CALL69:%.*]] = call i32 @_Z5tmainIiLi42EET_v()
// CHECK3-NEXT:    store i32 [[CALL69]], i32* [[RETVAL]], align 4
// CHECK3-NEXT:    [[TMP35:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK3-NEXT:    call void @llvm.stackrestore(i8* [[TMP35]])
// CHECK3-NEXT:    [[ARRAY_BEGIN70:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN70]], i64 5
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP36]], [[FOR_END68]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN70]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE71:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done71:
// CHECK3-NEXT:    [[ARRAY_BEGIN72:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK3-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN72]], i64 40
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY73:%.*]]
// CHECK3:       arraydestroy.body73:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST74:%.*]] = phi %struct.S.0* [ [[TMP37]], [[ARRAYDESTROY_DONE71]] ], [ [[ARRAYDESTROY_ELEMENT75:%.*]], [[ARRAYDESTROY_BODY73]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT75]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST74]], i64 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT75]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE76:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT75]], [[ARRAY_BEGIN72]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE76]], label [[ARRAYDESTROY_DONE77:%.*]], label [[ARRAYDESTROY_BODY73]]
// CHECK3:       arraydestroy.done77:
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAY_BEGIN78:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP38:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN78]], i64 4
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY79:%.*]]
// CHECK3:       arraydestroy.body79:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST80:%.*]] = phi %struct.S.0* [ [[TMP38]], [[ARRAYDESTROY_DONE77]] ], [ [[ARRAYDESTROY_ELEMENT81:%.*]], [[ARRAYDESTROY_BODY79]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT81]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST80]], i64 -1
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT81]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE82:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT81]], [[ARRAY_BEGIN78]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE82]], label [[ARRAYDESTROY_DONE83:%.*]], label [[ARRAYDESTROY_BODY79]]
// CHECK3:       arraydestroy.done83:
// CHECK3-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR7]]
// CHECK3-NEXT:    [[TMP39:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    ret i32 [[TMP39]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]], float [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v
// CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[T:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK3-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 16
// CHECK3-NEXT:    [[VAR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK3-NEXT:    [[ARR:%.*]] = alloca [42 x %struct.S], align 16
// CHECK3-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP4:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[_TMP5:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[I6:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP18:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[_TMP19:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[I20:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK3-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK3-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiLi42EET_v.vec to i8*), i64 8, i1 false)
// CHECK3-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK3-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK3-NEXT:    store %struct.S* [[TEST]], %struct.S** [[VAR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK3-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 42
// CHECK3-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK3:       arrayctor.loop:
// CHECK3-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK3-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR]])
// CHECK3-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK3-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK3-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK3:       arrayctor.cont:
// CHECK3-NEXT:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[TMP]], align 8
// CHECK3-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP2]], %struct.S** [[_TMP1]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I]], align 4
// CHECK3-NEXT:    br label [[FOR_COND:%.*]]
// CHECK3:       for.cond:
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP3]], 2
// CHECK3-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK3:       for.body:
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK3-NEXT:    store i32 [[TMP4]], i32* [[ARRAYIDX]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[IDXPROM2:%.*]] = sext i32 [[TMP7]] to i64
// CHECK3-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM2]]
// CHECK3-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX3]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP6]])
// CHECK3-NEXT:    br label [[FOR_INC:%.*]]
// CHECK3:       for.inc:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK3-NEXT:    store i32 [[INC]], i32* [[I]], align 4
// CHECK3-NEXT:    br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK3:       for.end:
// CHECK3-NEXT:    [[TMP9:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP9]], %struct.S** [[_TMP4]], align 8
// CHECK3-NEXT:    [[TMP10:%.*]] = load %struct.S*, %struct.S** [[_TMP4]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP10]], %struct.S** [[_TMP5]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I6]], align 4
// CHECK3-NEXT:    br label [[FOR_COND7:%.*]]
// CHECK3:       for.cond7:
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, i32* [[I6]], align 4
// CHECK3-NEXT:    [[CMP8:%.*]] = icmp slt i32 [[TMP11]], 2
// CHECK3-NEXT:    br i1 [[CMP8]], label [[FOR_BODY9:%.*]], label [[FOR_END17:%.*]]
// CHECK3:       for.body9:
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, i32* [[I6]], align 4
// CHECK3-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP13]] to i64
// CHECK3-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM10]]
// CHECK3-NEXT:    store i32 [[TMP12]], i32* [[ARRAYIDX11]], align 4
// CHECK3-NEXT:    [[TMP14:%.*]] = load %struct.S*, %struct.S** [[_TMP5]], align 8
// CHECK3-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I6]], align 4
// CHECK3-NEXT:    [[IDXPROM12:%.*]] = sext i32 [[TMP15]] to i64
// CHECK3-NEXT:    [[ARRAYIDX13:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM12]]
// CHECK3-NEXT:    [[CALL14:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX13]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP14]])
// CHECK3-NEXT:    br label [[FOR_INC15:%.*]]
// CHECK3:       for.inc15:
// CHECK3-NEXT:    [[TMP16:%.*]] = load i32, i32* [[I6]], align 4
// CHECK3-NEXT:    [[INC16:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK3-NEXT:    store i32 [[INC16]], i32* [[I6]], align 4
// CHECK3-NEXT:    br label [[FOR_COND7]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK3:       for.end17:
// CHECK3-NEXT:    [[TMP17:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP17]], %struct.S** [[_TMP18]], align 8
// CHECK3-NEXT:    [[TMP18:%.*]] = load %struct.S*, %struct.S** [[_TMP18]], align 8
// CHECK3-NEXT:    store %struct.S* [[TMP18]], %struct.S** [[_TMP19]], align 8
// CHECK3-NEXT:    store i32 0, i32* [[I20]], align 4
// CHECK3-NEXT:    br label [[FOR_COND21:%.*]]
// CHECK3:       for.cond21:
// CHECK3-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I20]], align 4
// CHECK3-NEXT:    [[CMP22:%.*]] = icmp slt i32 [[TMP19]], 2
// CHECK3-NEXT:    br i1 [[CMP22]], label [[FOR_BODY23:%.*]], label [[FOR_END31:%.*]]
// CHECK3:       for.body23:
// CHECK3-NEXT:    [[TMP20:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK3-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I20]], align 4
// CHECK3-NEXT:    [[IDXPROM24:%.*]] = sext i32 [[TMP21]] to i64
// CHECK3-NEXT:    [[ARRAYIDX25:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM24]]
// CHECK3-NEXT:    store i32 [[TMP20]], i32* [[ARRAYIDX25]], align 4
// CHECK3-NEXT:    [[TMP22:%.*]] = load %struct.S*, %struct.S** [[_TMP19]], align 8
// CHECK3-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I20]], align 4
// CHECK3-NEXT:    [[IDXPROM26:%.*]] = sext i32 [[TMP23]] to i64
// CHECK3-NEXT:    [[ARRAYIDX27:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM26]]
// CHECK3-NEXT:    [[CALL28:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX27]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP22]])
// CHECK3-NEXT:    br label [[FOR_INC29:%.*]]
// CHECK3:       for.inc29:
// CHECK3-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I20]], align 4
// CHECK3-NEXT:    [[INC30:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK3-NEXT:    store i32 [[INC30]], i32* [[I20]], align 4
// CHECK3-NEXT:    br label [[FOR_COND21]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK3:       for.end31:
// CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    [[ARRAY_BEGIN32:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN32]], i64 42
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK3:       arraydestroy.body:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP25]], [[FOR_END31]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN32]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE33:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK3:       arraydestroy.done33:
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAY_BEGIN34:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN34]], i64 2
// CHECK3-NEXT:    br label [[ARRAYDESTROY_BODY35:%.*]]
// CHECK3:       arraydestroy.body35:
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENTPAST36:%.*]] = phi %struct.S* [ [[TMP26]], [[ARRAYDESTROY_DONE33]] ], [ [[ARRAYDESTROY_ELEMENT37:%.*]], [[ARRAYDESTROY_BODY35]] ]
// CHECK3-NEXT:    [[ARRAYDESTROY_ELEMENT37]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST36]], i64 -1
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT37]]) #[[ATTR7]]
// CHECK3-NEXT:    [[ARRAYDESTROY_DONE38:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT37]], [[ARRAY_BEGIN34]]
// CHECK3-NEXT:    br i1 [[ARRAYDESTROY_DONE38]], label [[ARRAYDESTROY_DONE39:%.*]], label [[ARRAYDESTROY_BODY35]]
// CHECK3:       arraydestroy.done39:
// CHECK3-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR7]]
// CHECK3-NEXT:    [[TMP27:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    ret i32 [[TMP27]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    call void @_ZN1SIfED2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR7]]
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK3-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK3-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK3-NEXT:    [[CONV:%.*]] = fptosi double [[TMP4]] to i32
// CHECK3-NEXT:    store i32 [[CONV]], i32* [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK3-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK3-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK3-NEXT:    [[CONV:%.*]] = fptrunc double [[TMP4]] to float
// CHECK3-NEXT:    store float [[CONV]], float* [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK3-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK3-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK3-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK3-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP4:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[CONV:%.*]] = fpext float [[TMP4]] to double
// CHECK3-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK3-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK3-NEXT:    [[CONV2:%.*]] = fptrunc double [[ADD]] to float
// CHECK3-NEXT:    store float [[CONV2]], float* [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]], i32 [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK3-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK3-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK3-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK3-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK3-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK3-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK3-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to double
// CHECK3-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK3-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK3-NEXT:    [[CONV2:%.*]] = fptosi double [[ADD]] to i32
// CHECK3-NEXT:    store i32 [[CONV2]], i32* [[F]], align 4
// CHECK3-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_Z4bazzv
// CHECK4-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK4-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[S1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]])
// CHECK4-NEXT:    store i32 0, i32* [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    call void @.omp_initializer.(%struct.S* [[S1]], %struct.S* [[S]])
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK4:       omp.inner.for.cond:
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK4-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP0]], 10
// CHECK4-NEXT:    br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK4:       omp.inner.for.cond.cleanup:
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK4:       omp.inner.for.body:
// CHECK4-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP1]], 1
// CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK4-NEXT:    store i32 [[ADD]], i32* [[I]], align 4, !llvm.access.group !2
// CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK4:       omp.body.continue:
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK4:       omp.inner.for.inc:
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK4-NEXT:    [[ADD2:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK4-NEXT:    store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4, !llvm.access.group !2
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK4:       omp.inner.for.end:
// CHECK4-NEXT:    store i32 10, i32* [[I]], align 4
// CHECK4-NEXT:    call void @.omp_combiner.(%struct.S* [[S]], %struct.S* [[S1]])
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S1]]) #[[ATTR7:[0-9]+]]
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[S]]) #[[ATTR7]]
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIiEC2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_combiner.
// CHECK4-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[TMP2]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[F]], align 4
// CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 17, [[TMP4]]
// CHECK4-NEXT:    [[F2:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[TMP3]], i32 0, i32 2
// CHECK4-NEXT:    store i32 [[MUL]], i32* [[F2]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@.omp_initializer.
// CHECK4-SAME: (%struct.S* noalias [[TMP0:%.*]], %struct.S* noalias [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[TMP0]], %struct.S** [[DOTADDR]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load %struct.S*, %struct.S** [[DOTADDR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TMP3]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIiED2Ev(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR7]]
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@main
// CHECK4-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK4-NEXT:    [[T_VAR:%.*]] = alloca float, align 4
// CHECK4-NEXT:    [[T_VAR1:%.*]] = alloca float, align 4
// CHECK4-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK4-NEXT:    [[S_ARR:%.*]] = alloca [4 x %struct.S.0], align 16
// CHECK4-NEXT:    [[VAR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK4-NEXT:    [[ARRS:%.*]] = alloca [10 x [4 x %struct.S.0]], align 16
// CHECK4-NEXT:    [[VAR2:%.*]] = alloca %struct.S.0**, align 8
// CHECK4-NEXT:    [[VVAR2:%.*]] = alloca [5 x %struct.S.0], align 16
// CHECK4-NEXT:    [[VAR3:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    [[_TMP10:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK4-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK4-NEXT:    [[I15:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[I26:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[I37:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[I44:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[_TMP51:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK4-NEXT:    [[_TMP52:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK4-NEXT:    [[I53:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[_TMP60:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK4-NEXT:    [[_TMP61:%.*]] = alloca [4 x %struct.S.0]*, align 8
// CHECK4-NEXT:    [[I62:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK4-NEXT:    store float 0.000000e+00, float* [[T_VAR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK4-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)
// CHECK4-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], float 1.000000e+00)
// CHECK4-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], float 2.000000e+00)
// CHECK4-NEXT:    [[ARRAYINIT_ELEMENT1:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT]], i64 1
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT1]], float 3.000000e+00)
// CHECK4-NEXT:    [[ARRAYINIT_ELEMENT2:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_ELEMENT1]], i64 1
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT2]], float 4.000000e+00)
// CHECK4-NEXT:    store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK4-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 40
// CHECK4-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK4:       arrayctor.loop:
// CHECK4-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR]])
// CHECK4-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK4-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK4-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK4:       arrayctor.cont:
// CHECK4-NEXT:    [[CALL:%.*]] = call %struct.S.0** @_Z3foov()
// CHECK4-NEXT:    store %struct.S.0** [[CALL]], %struct.S.0*** [[VAR2]], align 8
// CHECK4-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK4-NEXT:    [[ARRAYCTOR_END4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN3]], i64 5
// CHECK4-NEXT:    br label [[ARRAYCTOR_LOOP5:%.*]]
// CHECK4:       arrayctor.loop5:
// CHECK4-NEXT:    [[ARRAYCTOR_CUR6:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYCTOR_NEXT7:%.*]], [[ARRAYCTOR_LOOP5]] ]
// CHECK4-NEXT:    call void @_ZN1SIfEC1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR6]])
// CHECK4-NEXT:    [[ARRAYCTOR_NEXT7]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR6]], i64 1
// CHECK4-NEXT:    [[ARRAYCTOR_DONE8:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT7]], [[ARRAYCTOR_END4]]
// CHECK4-NEXT:    br i1 [[ARRAYCTOR_DONE8]], label [[ARRAYCTOR_CONT9:%.*]], label [[ARRAYCTOR_LOOP5]]
// CHECK4:       arrayctor.cont9:
// CHECK4-NEXT:    store [4 x %struct.S.0]* [[S_ARR]], [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK4-NEXT:    [[TMP1:%.*]] = load %struct.S.0*, %struct.S.0** [[VAR]], align 8
// CHECK4-NEXT:    store %struct.S.0* [[TMP1]], %struct.S.0** [[TMP]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S.0*, %struct.S.0** [[TMP]], align 8
// CHECK4-NEXT:    store %struct.S.0* [[TMP2]], %struct.S.0** [[_TMP10]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I]], align 4
// CHECK4-NEXT:    br label [[FOR_COND:%.*]]
// CHECK4:       for.cond:
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP3]], 2
// CHECK4-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK4:       for.body:
// CHECK4-NEXT:    [[TMP4:%.*]] = load float, float* [[T_VAR]], align 4
// CHECK4-NEXT:    [[CONV:%.*]] = fptosi float [[TMP4]] to i32
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64
// CHECK4-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK4-NEXT:    store i32 [[CONV]], i32* [[ARRAYIDX]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP10]], align 8
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP7]] to i64
// CHECK4-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i64 0, i64 [[IDXPROM11]]
// CHECK4-NEXT:    [[CALL13:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S.0* @_ZN1SIfEaSERKS0_(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYIDX12]], %struct.S.0* nonnull align 4 dereferenceable(12) [[TMP6]])
// CHECK4-NEXT:    br label [[FOR_INC:%.*]]
// CHECK4:       for.inc:
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK4-NEXT:    store i32 [[INC]], i32* [[I]], align 4
// CHECK4-NEXT:    br label [[FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK4:       for.end:
// CHECK4-NEXT:    [[ARRAYIDX14:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 1
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX14]], align 4
// CHECK4-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK4-NEXT:    [[TMP11:%.*]] = call i8* @llvm.stacksave()
// CHECK4-NEXT:    store i8* [[TMP11]], i8** [[SAVED_STACK]], align 8
// CHECK4-NEXT:    [[TMP12:%.*]] = mul nuw i64 10, [[TMP10]]
// CHECK4-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP12]], align 16
// CHECK4-NEXT:    store i64 [[TMP10]], i64* [[__VLA_EXPR0]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I15]], align 4
// CHECK4-NEXT:    br label [[FOR_COND16:%.*]]
// CHECK4:       for.cond16:
// CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[I15]], align 4
// CHECK4-NEXT:    [[CMP17:%.*]] = icmp slt i32 [[TMP13]], 10
// CHECK4-NEXT:    br i1 [[CMP17]], label [[FOR_BODY18:%.*]], label [[FOR_END25:%.*]]
// CHECK4:       for.body18:
// CHECK4-NEXT:    [[TMP14:%.*]] = mul nsw i64 1, [[TMP10]]
// CHECK4-NEXT:    [[ARRAYIDX19:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[TMP14]]
// CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I15]], align 4
// CHECK4-NEXT:    [[IDXPROM20:%.*]] = sext i32 [[TMP15]] to i64
// CHECK4-NEXT:    [[ARRAYIDX21:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX19]], i64 [[IDXPROM20]]
// CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[ARRAYIDX21]], align 4
// CHECK4-NEXT:    [[INC22:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK4-NEXT:    store i32 [[INC22]], i32* [[ARRAYIDX21]], align 4
// CHECK4-NEXT:    br label [[FOR_INC23:%.*]]
// CHECK4:       for.inc23:
// CHECK4-NEXT:    [[TMP17:%.*]] = load i32, i32* [[I15]], align 4
// CHECK4-NEXT:    [[INC24:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK4-NEXT:    store i32 [[INC24]], i32* [[I15]], align 4
// CHECK4-NEXT:    br label [[FOR_COND16]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK4:       for.end25:
// CHECK4-NEXT:    store i32 0, i32* [[I26]], align 4
// CHECK4-NEXT:    br label [[FOR_COND27:%.*]]
// CHECK4:       for.cond27:
// CHECK4-NEXT:    [[TMP18:%.*]] = load i32, i32* [[I26]], align 4
// CHECK4-NEXT:    [[CMP28:%.*]] = icmp slt i32 [[TMP18]], 10
// CHECK4-NEXT:    br i1 [[CMP28]], label [[FOR_BODY29:%.*]], label [[FOR_END36:%.*]]
// CHECK4:       for.body29:
// CHECK4-NEXT:    [[TMP19:%.*]] = mul nsw i64 1, [[TMP10]]
// CHECK4-NEXT:    [[ARRAYIDX30:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 [[TMP19]]
// CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[I26]], align 4
// CHECK4-NEXT:    [[IDXPROM31:%.*]] = sext i32 [[TMP20]] to i64
// CHECK4-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, i32* [[ARRAYIDX30]], i64 [[IDXPROM31]]
// CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[ARRAYIDX32]], align 4
// CHECK4-NEXT:    [[INC33:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK4-NEXT:    store i32 [[INC33]], i32* [[ARRAYIDX32]], align 4
// CHECK4-NEXT:    br label [[FOR_INC34:%.*]]
// CHECK4:       for.inc34:
// CHECK4-NEXT:    [[TMP22:%.*]] = load i32, i32* [[I26]], align 4
// CHECK4-NEXT:    [[INC35:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK4-NEXT:    store i32 [[INC35]], i32* [[I26]], align 4
// CHECK4-NEXT:    br label [[FOR_COND27]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK4:       for.end36:
// CHECK4-NEXT:    store i32 0, i32* [[I37]], align 4
// CHECK4-NEXT:    br label [[FOR_COND38:%.*]]
// CHECK4:       for.cond38:
// CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I37]], align 4
// CHECK4-NEXT:    [[CMP39:%.*]] = icmp slt i32 [[TMP23]], 10
// CHECK4-NEXT:    br i1 [[CMP39]], label [[FOR_BODY40:%.*]], label [[FOR_END43:%.*]]
// CHECK4:       for.body40:
// CHECK4-NEXT:    br label [[FOR_INC41:%.*]]
// CHECK4:       for.inc41:
// CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I37]], align 4
// CHECK4-NEXT:    [[INC42:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK4-NEXT:    store i32 [[INC42]], i32* [[I37]], align 4
// CHECK4-NEXT:    br label [[FOR_COND38]], !llvm.loop [[LOOP10:![0-9]+]]
// CHECK4:       for.end43:
// CHECK4-NEXT:    store i32 0, i32* [[I44]], align 4
// CHECK4-NEXT:    br label [[FOR_COND45:%.*]]
// CHECK4:       for.cond45:
// CHECK4-NEXT:    [[TMP25:%.*]] = load i32, i32* [[I44]], align 4
// CHECK4-NEXT:    [[CMP46:%.*]] = icmp slt i32 [[TMP25]], 10
// CHECK4-NEXT:    br i1 [[CMP46]], label [[FOR_BODY47:%.*]], label [[FOR_END50:%.*]]
// CHECK4:       for.body47:
// CHECK4-NEXT:    br label [[FOR_INC48:%.*]]
// CHECK4:       for.inc48:
// CHECK4-NEXT:    [[TMP26:%.*]] = load i32, i32* [[I44]], align 4
// CHECK4-NEXT:    [[INC49:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK4-NEXT:    store i32 [[INC49]], i32* [[I44]], align 4
// CHECK4-NEXT:    br label [[FOR_COND45]], !llvm.loop [[LOOP11:![0-9]+]]
// CHECK4:       for.end50:
// CHECK4-NEXT:    [[TMP27:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK4-NEXT:    store [4 x %struct.S.0]* [[TMP27]], [4 x %struct.S.0]** [[_TMP51]], align 8
// CHECK4-NEXT:    [[TMP28:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP51]], align 8
// CHECK4-NEXT:    store [4 x %struct.S.0]* [[TMP28]], [4 x %struct.S.0]** [[_TMP52]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I53]], align 4
// CHECK4-NEXT:    br label [[FOR_COND54:%.*]]
// CHECK4:       for.cond54:
// CHECK4-NEXT:    [[TMP29:%.*]] = load i32, i32* [[I53]], align 4
// CHECK4-NEXT:    [[CMP55:%.*]] = icmp slt i32 [[TMP29]], 10
// CHECK4-NEXT:    br i1 [[CMP55]], label [[FOR_BODY56:%.*]], label [[FOR_END59:%.*]]
// CHECK4:       for.body56:
// CHECK4-NEXT:    br label [[FOR_INC57:%.*]]
// CHECK4:       for.inc57:
// CHECK4-NEXT:    [[TMP30:%.*]] = load i32, i32* [[I53]], align 4
// CHECK4-NEXT:    [[INC58:%.*]] = add nsw i32 [[TMP30]], 1
// CHECK4-NEXT:    store i32 [[INC58]], i32* [[I53]], align 4
// CHECK4-NEXT:    br label [[FOR_COND54]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK4:       for.end59:
// CHECK4-NEXT:    [[TMP31:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[VAR3]], align 8
// CHECK4-NEXT:    store [4 x %struct.S.0]* [[TMP31]], [4 x %struct.S.0]** [[_TMP60]], align 8
// CHECK4-NEXT:    [[TMP32:%.*]] = load [4 x %struct.S.0]*, [4 x %struct.S.0]** [[_TMP60]], align 8
// CHECK4-NEXT:    store [4 x %struct.S.0]* [[TMP32]], [4 x %struct.S.0]** [[_TMP61]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I62]], align 4
// CHECK4-NEXT:    br label [[FOR_COND63:%.*]]
// CHECK4:       for.cond63:
// CHECK4-NEXT:    [[TMP33:%.*]] = load i32, i32* [[I62]], align 4
// CHECK4-NEXT:    [[CMP64:%.*]] = icmp slt i32 [[TMP33]], 10
// CHECK4-NEXT:    br i1 [[CMP64]], label [[FOR_BODY65:%.*]], label [[FOR_END68:%.*]]
// CHECK4:       for.body65:
// CHECK4-NEXT:    br label [[FOR_INC66:%.*]]
// CHECK4:       for.inc66:
// CHECK4-NEXT:    [[TMP34:%.*]] = load i32, i32* [[I62]], align 4
// CHECK4-NEXT:    [[INC67:%.*]] = add nsw i32 [[TMP34]], 1
// CHECK4-NEXT:    store i32 [[INC67]], i32* [[I62]], align 4
// CHECK4-NEXT:    br label [[FOR_COND63]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK4:       for.end68:
// CHECK4-NEXT:    [[CALL69:%.*]] = call i32 @_Z5tmainIiLi42EET_v()
// CHECK4-NEXT:    store i32 [[CALL69]], i32* [[RETVAL]], align 4
// CHECK4-NEXT:    [[TMP35:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK4-NEXT:    call void @llvm.stackrestore(i8* [[TMP35]])
// CHECK4-NEXT:    [[ARRAY_BEGIN70:%.*]] = getelementptr inbounds [5 x %struct.S.0], [5 x %struct.S.0]* [[VVAR2]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN70]], i64 5
// CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK4:       arraydestroy.body:
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP36]], [[FOR_END68]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN70]]
// CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE71:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK4:       arraydestroy.done71:
// CHECK4-NEXT:    [[ARRAY_BEGIN72:%.*]] = getelementptr inbounds [10 x [4 x %struct.S.0]], [10 x [4 x %struct.S.0]]* [[ARRS]], i32 0, i32 0, i32 0
// CHECK4-NEXT:    [[TMP37:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN72]], i64 40
// CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY73:%.*]]
// CHECK4:       arraydestroy.body73:
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST74:%.*]] = phi %struct.S.0* [ [[TMP37]], [[ARRAYDESTROY_DONE71]] ], [ [[ARRAYDESTROY_ELEMENT75:%.*]], [[ARRAYDESTROY_BODY73]] ]
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT75]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST74]], i64 -1
// CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT75]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAYDESTROY_DONE76:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT75]], [[ARRAY_BEGIN72]]
// CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE76]], label [[ARRAYDESTROY_DONE77:%.*]], label [[ARRAYDESTROY_BODY73]]
// CHECK4:       arraydestroy.done77:
// CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAY_BEGIN78:%.*]] = getelementptr inbounds [4 x %struct.S.0], [4 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP38:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN78]], i64 4
// CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY79:%.*]]
// CHECK4:       arraydestroy.body79:
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST80:%.*]] = phi %struct.S.0* [ [[TMP38]], [[ARRAYDESTROY_DONE77]] ], [ [[ARRAYDESTROY_ELEMENT81:%.*]], [[ARRAYDESTROY_BODY79]] ]
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT81]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST80]], i64 -1
// CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT81]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAYDESTROY_DONE82:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT81]], [[ARRAY_BEGIN78]]
// CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE82]], label [[ARRAYDESTROY_DONE83:%.*]], label [[ARRAYDESTROY_BODY79]]
// CHECK4:       arraydestroy.done83:
// CHECK4-NEXT:    call void @_ZN1SIfED1Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR7]]
// CHECK4-NEXT:    [[TMP39:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    ret i32 [[TMP39]]
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIfEC2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK4-NEXT:    call void @_ZN1SIfEC2Ef(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]], float [[TMP0]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v
// CHECK4-SAME: () #[[ATTR0]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[T:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK4-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK4-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S], align 16
// CHECK4-NEXT:    [[VAR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK4-NEXT:    [[ARR:%.*]] = alloca [42 x %struct.S], align 16
// CHECK4-NEXT:    [[TMP:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[_TMP4:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[_TMP5:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[I6:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[_TMP18:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[_TMP19:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[I20:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]])
// CHECK4-NEXT:    store i32 0, i32* [[T_VAR]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK4-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiLi42EET_v.vec to i8*), i64 8, i1 false)
// CHECK4-NEXT:    [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_BEGIN]], i32 1)
// CHECK4-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ei(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYINIT_ELEMENT]], i32 2)
// CHECK4-NEXT:    store %struct.S* [[TEST]], %struct.S** [[VAR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]])
// CHECK4-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK4-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 42
// CHECK4-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK4:       arrayctor.loop:
// CHECK4-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK4-NEXT:    call void @_ZN1SIiEC1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYCTOR_CUR]])
// CHECK4-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK4-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK4-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK4:       arrayctor.cont:
// CHECK4-NEXT:    [[TMP1:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP1]], %struct.S** [[TMP]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load %struct.S*, %struct.S** [[TMP]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP2]], %struct.S** [[_TMP1]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I]], align 4
// CHECK4-NEXT:    br label [[FOR_COND:%.*]]
// CHECK4:       for.cond:
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP3]], 2
// CHECK4-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK4:       for.body:
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64
// CHECK4-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK4-NEXT:    store i32 [[TMP4]], i32* [[ARRAYIDX]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load %struct.S*, %struct.S** [[_TMP1]], align 8
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[IDXPROM2:%.*]] = sext i32 [[TMP7]] to i64
// CHECK4-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM2]]
// CHECK4-NEXT:    [[CALL:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX3]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP6]])
// CHECK4-NEXT:    br label [[FOR_INC:%.*]]
// CHECK4:       for.inc:
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, i32* [[I]], align 4
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK4-NEXT:    store i32 [[INC]], i32* [[I]], align 4
// CHECK4-NEXT:    br label [[FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]]
// CHECK4:       for.end:
// CHECK4-NEXT:    [[TMP9:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP9]], %struct.S** [[_TMP4]], align 8
// CHECK4-NEXT:    [[TMP10:%.*]] = load %struct.S*, %struct.S** [[_TMP4]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP10]], %struct.S** [[_TMP5]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I6]], align 4
// CHECK4-NEXT:    br label [[FOR_COND7:%.*]]
// CHECK4:       for.cond7:
// CHECK4-NEXT:    [[TMP11:%.*]] = load i32, i32* [[I6]], align 4
// CHECK4-NEXT:    [[CMP8:%.*]] = icmp slt i32 [[TMP11]], 2
// CHECK4-NEXT:    br i1 [[CMP8]], label [[FOR_BODY9:%.*]], label [[FOR_END17:%.*]]
// CHECK4:       for.body9:
// CHECK4-NEXT:    [[TMP12:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK4-NEXT:    [[TMP13:%.*]] = load i32, i32* [[I6]], align 4
// CHECK4-NEXT:    [[IDXPROM10:%.*]] = sext i32 [[TMP13]] to i64
// CHECK4-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM10]]
// CHECK4-NEXT:    store i32 [[TMP12]], i32* [[ARRAYIDX11]], align 4
// CHECK4-NEXT:    [[TMP14:%.*]] = load %struct.S*, %struct.S** [[_TMP5]], align 8
// CHECK4-NEXT:    [[TMP15:%.*]] = load i32, i32* [[I6]], align 4
// CHECK4-NEXT:    [[IDXPROM12:%.*]] = sext i32 [[TMP15]] to i64
// CHECK4-NEXT:    [[ARRAYIDX13:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM12]]
// CHECK4-NEXT:    [[CALL14:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX13]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP14]])
// CHECK4-NEXT:    br label [[FOR_INC15:%.*]]
// CHECK4:       for.inc15:
// CHECK4-NEXT:    [[TMP16:%.*]] = load i32, i32* [[I6]], align 4
// CHECK4-NEXT:    [[INC16:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK4-NEXT:    store i32 [[INC16]], i32* [[I6]], align 4
// CHECK4-NEXT:    br label [[FOR_COND7]], !llvm.loop [[LOOP15:![0-9]+]]
// CHECK4:       for.end17:
// CHECK4-NEXT:    [[TMP17:%.*]] = load %struct.S*, %struct.S** [[VAR]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP17]], %struct.S** [[_TMP18]], align 8
// CHECK4-NEXT:    [[TMP18:%.*]] = load %struct.S*, %struct.S** [[_TMP18]], align 8
// CHECK4-NEXT:    store %struct.S* [[TMP18]], %struct.S** [[_TMP19]], align 8
// CHECK4-NEXT:    store i32 0, i32* [[I20]], align 4
// CHECK4-NEXT:    br label [[FOR_COND21:%.*]]
// CHECK4:       for.cond21:
// CHECK4-NEXT:    [[TMP19:%.*]] = load i32, i32* [[I20]], align 4
// CHECK4-NEXT:    [[CMP22:%.*]] = icmp slt i32 [[TMP19]], 2
// CHECK4-NEXT:    br i1 [[CMP22]], label [[FOR_BODY23:%.*]], label [[FOR_END31:%.*]]
// CHECK4:       for.body23:
// CHECK4-NEXT:    [[TMP20:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK4-NEXT:    [[TMP21:%.*]] = load i32, i32* [[I20]], align 4
// CHECK4-NEXT:    [[IDXPROM24:%.*]] = sext i32 [[TMP21]] to i64
// CHECK4-NEXT:    [[ARRAYIDX25:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM24]]
// CHECK4-NEXT:    store i32 [[TMP20]], i32* [[ARRAYIDX25]], align 4
// CHECK4-NEXT:    [[TMP22:%.*]] = load %struct.S*, %struct.S** [[_TMP19]], align 8
// CHECK4-NEXT:    [[TMP23:%.*]] = load i32, i32* [[I20]], align 4
// CHECK4-NEXT:    [[IDXPROM26:%.*]] = sext i32 [[TMP23]] to i64
// CHECK4-NEXT:    [[ARRAYIDX27:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM26]]
// CHECK4-NEXT:    [[CALL28:%.*]] = call nonnull align 4 dereferenceable(12) %struct.S* @_ZN1SIiEaSERKS0_(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYIDX27]], %struct.S* nonnull align 4 dereferenceable(12) [[TMP22]])
// CHECK4-NEXT:    br label [[FOR_INC29:%.*]]
// CHECK4:       for.inc29:
// CHECK4-NEXT:    [[TMP24:%.*]] = load i32, i32* [[I20]], align 4
// CHECK4-NEXT:    [[INC30:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK4-NEXT:    store i32 [[INC30]], i32* [[I20]], align 4
// CHECK4-NEXT:    br label [[FOR_COND21]], !llvm.loop [[LOOP16:![0-9]+]]
// CHECK4:       for.end31:
// CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    [[ARRAY_BEGIN32:%.*]] = getelementptr inbounds [42 x %struct.S], [42 x %struct.S]* [[ARR]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN32]], i64 42
// CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK4:       arraydestroy.body:
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP25]], [[FOR_END31]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN32]]
// CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE33:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK4:       arraydestroy.done33:
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[VAR1]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAY_BEGIN34:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN34]], i64 2
// CHECK4-NEXT:    br label [[ARRAYDESTROY_BODY35:%.*]]
// CHECK4:       arraydestroy.body35:
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENTPAST36:%.*]] = phi %struct.S* [ [[TMP26]], [[ARRAYDESTROY_DONE33]] ], [ [[ARRAYDESTROY_ELEMENT37:%.*]], [[ARRAYDESTROY_BODY35]] ]
// CHECK4-NEXT:    [[ARRAYDESTROY_ELEMENT37]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST36]], i64 -1
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[ARRAYDESTROY_ELEMENT37]]) #[[ATTR7]]
// CHECK4-NEXT:    [[ARRAYDESTROY_DONE38:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT37]], [[ARRAY_BEGIN34]]
// CHECK4-NEXT:    br i1 [[ARRAYDESTROY_DONE38]], label [[ARRAYDESTROY_DONE39:%.*]], label [[ARRAYDESTROY_BODY35]]
// CHECK4:       arraydestroy.done39:
// CHECK4-NEXT:    call void @_ZN1SIiED1Ev(%struct.S* nonnull align 4 dereferenceable(12) [[TEST]]) #[[ATTR7]]
// CHECK4-NEXT:    [[TMP27:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    ret i32 [[TMP27]]
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    call void @_ZN1SIfED2Ev(%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS1]]) #[[ATTR7]]
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK4-NEXT:    [[CONV:%.*]] = fptosi double [[TMP4]] to i32
// CHECK4-NEXT:    store i32 [[CONV]], i32* [[F]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP4:%.*]] = load volatile double, double* @g, align 8
// CHECK4-NEXT:    [[CONV:%.*]] = fptrunc double [[TMP4]] to float
// CHECK4-NEXT:    store float [[CONV]], float* [[F]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK4-SAME: (%struct.S.0* nonnull align 4 dereferenceable(12) [[THIS:%.*]], float [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK4-NEXT:    store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    store float [[A]], float* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast %struct.S.0* [[THIS1]] to %struct.BaseS*
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast %struct.S.0* [[THIS1]] to i8*
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP4:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[CONV:%.*]] = fpext float [[TMP4]] to double
// CHECK4-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK4-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK4-NEXT:    [[CONV2:%.*]] = fptrunc double [[ADD]] to float
// CHECK4-NEXT:    store float [[CONV2]], float* [[F]], align 4
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    call void @_ZN1SIiEC2Ei(%struct.S* nonnull align 4 dereferenceable(12) [[THIS1]], i32 [[TMP0]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK4-SAME: (%struct.S* nonnull align 4 dereferenceable(12) [[THIS:%.*]], i32 [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK4-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = bitcast %struct.S* [[THIS1]] to %struct.BaseS*
// CHECK4-NEXT:    [[TMP1:%.*]] = bitcast %struct.S* [[THIS1]] to i8*
// CHECK4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i8, i8* [[TMP1]], i64 4
// CHECK4-NEXT:    [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.BaseS1*
// CHECK4-NEXT:    [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 2
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK4-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP4]] to double
// CHECK4-NEXT:    [[TMP5:%.*]] = load volatile double, double* @g, align 8
// CHECK4-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP5]]
// CHECK4-NEXT:    [[CONV2:%.*]] = fptosi double [[ADD]] to i32
// CHECK4-NEXT:    store i32 [[CONV2]], i32* [[F]], align 4
// CHECK4-NEXT:    ret void
//