xref: /llvm-project-15.0.7/clang/test/OpenMP/for_reduction_task_codegen.cpp (revision 207b08a9)

// 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-unknown-linux -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2

// RUN: %clang_cc1 -triple x86_64-unknown-linux -verify -fopenmp-simd -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK4
// expected-no-diagnostics
#ifndef HEADER
#define HEADER

int main(int argc, char **argv) {
#pragma omp parallel
#pragma omp for reduction(task, +: argc, argv[0:10][0:argc])
  for (long long i = 0; i < 10; ++i) {
#pragma omp task in_reduction(+: argc, argv[0:10][0:argc])
    ;
  }
}



// Init firstprivate copy of argc

// Init firstprivate copy of argv[0:10][0:argc]

// Register task reduction.









#endif
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK1-NEXT:    store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK1-NEXT:    store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i8***)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[ARGC_ADDR]], i8*** [[ARGV_ADDR]])
// CHECK1-NEXT:    ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8*** nonnull align 8 dereferenceable(8) [[ARGV:%.*]]) #[[ATTR1:[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:    [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8***, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[ARGC1:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca i8**, align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8
// CHECK1-NEXT:    [[DOTTASK_RED_:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
// CHECK1-NEXT:    [[ATOMIC_TEMP:%.*]] = alloca i8, align 1
// CHECK1-NEXT:    [[_TMP28:%.*]] = alloca i8, align 1
// 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* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT:    store i8*** [[ARGV]], i8**** [[ARGV_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i8***, i8**** [[ARGV_ADDR]], align 8
// CHECK1-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK1-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT:    store i64 1, i64* [[DOTOMP_STRIDE]], align 8
// CHECK1-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    store i32 0, i32* [[ARGC1]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0
// CHECK1-NEXT:    [[TMP3:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP3]], i64 0
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP5]]
// CHECK1-NEXT:    [[TMP6:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP6]], i64 9
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 [[LB_ADD_LEN]]
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = sub i64 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    [[TMP11:%.*]] = sdiv exact i64 [[TMP10]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP12:%.*]] = add nuw i64 [[TMP11]], 1
// CHECK1-NEXT:    [[TMP13:%.*]] = mul nuw i64 [[TMP12]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP14:%.*]] = call i8* @llvm.stacksave()
// CHECK1-NEXT:    store i8* [[TMP14]], i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA:%.*]] = alloca i8, i64 [[TMP12]], align 16
// CHECK1-NEXT:    store i64 [[TMP12]], i64* [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP12]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
// CHECK1-NEXT:    [[TMP19:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
// CHECK1-NEXT:    [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]]
// CHECK1-NEXT:    [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP21]]
// CHECK1-NEXT:    store i8** [[_TMP6]], i8*** [[_TMP5]], align 8
// CHECK1-NEXT:    store i8* [[TMP22]], i8** [[_TMP6]], align 8
// CHECK1-NEXT:    [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP24:%.*]] = bitcast i32* [[ARGC1]] to i8*
// CHECK1-NEXT:    store i8* [[TMP24]], i8** [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP26:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK1-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT:    store i64 4, i64* [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP28]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT:    store i8* null, i8** [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT:    [[TMP32:%.*]] = bitcast i32* [[TMP31]] to i8*
// CHECK1-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP32]], i8 0, i64 4, i1 false)
// CHECK1-NEXT:    [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP34:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i8*, i8** [[TMP34]], i64 0
// CHECK1-NEXT:    [[TMP35:%.*]] = load i8*, i8** [[ARRAYIDX8]], align 8
// CHECK1-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, i8* [[TMP35]], i64 0
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP37:%.*]] = sext i32 [[TMP36]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP37]]
// CHECK1-NEXT:    [[TMP38:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds i8*, i8** [[TMP38]], i64 9
// CHECK1-NEXT:    [[TMP39:%.*]] = load i8*, i8** [[ARRAYIDX11]], align 8
// CHECK1-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 [[LB_ADD_LEN10]]
// CHECK1-NEXT:    store i8* [[VLA]], i8** [[TMP33]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 1
// CHECK1-NEXT:    store i8* [[ARRAYIDX9]], i8** [[TMP40]], align 8
// CHECK1-NEXT:    [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX12]] to i64
// CHECK1-NEXT:    [[TMP42:%.*]] = ptrtoint i8* [[ARRAYIDX9]] to i64
// CHECK1-NEXT:    [[TMP43:%.*]] = sub i64 [[TMP41]], [[TMP42]]
// CHECK1-NEXT:    [[TMP44:%.*]] = sdiv exact i64 [[TMP43]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP45:%.*]] = add nuw i64 [[TMP44]], 1
// CHECK1-NEXT:    [[TMP46:%.*]] = mul nuw i64 [[TMP45]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 2
// CHECK1-NEXT:    store i64 [[TMP46]], i64* [[TMP47]], align 8
// CHECK1-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 3
// CHECK1-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP48]], align 8
// CHECK1-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 4
// CHECK1-NEXT:    store i8* null, i8** [[TMP49]], align 8
// CHECK1-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 5
// CHECK1-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP50]], align 8
// CHECK1-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 6
// CHECK1-NEXT:    store i32 1, i32* [[TMP51]], align 8
// CHECK1-NEXT:    [[TMP52:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP53:%.*]] = load i32, i32* [[TMP52]], align 4
// CHECK1-NEXT:    [[TMP54:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8*
// CHECK1-NEXT:    [[TMP55:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP53]], i32 1, i32 2, i8* [[TMP54]])
// CHECK1-NEXT:    store i8* [[TMP55]], i8** [[DOTTASK_RED_]], align 8
// CHECK1-NEXT:    [[TMP56:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP57:%.*]] = load i32, i32* [[TMP56]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP57]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT:    [[TMP58:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i64 [[TMP58]], 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:    [[TMP59:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP59]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT:    [[TMP60:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK1-NEXT:    store i64 [[TMP60]], i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP61:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT:    [[TMP62:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK1-NEXT:    [[CMP13:%.*]] = icmp sle i64 [[TMP61]], [[TMP62]]
// CHECK1-NEXT:    br i1 [[CMP13]], 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:    [[TMP63:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i64 [[TMP63]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i64 0, [[MUL]]
// CHECK1-NEXT:    store i64 [[ADD]], i64* [[I]], align 8
// CHECK1-NEXT:    [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT:    store i8** [[DOTTASK_RED_]], i8*** [[TMP64]], align 8
// CHECK1-NEXT:    [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT:    store i32* [[ARGC1]], i32** [[TMP65]], align 8
// CHECK1-NEXT:    [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP67:%.*]] = load i8**, i8*** [[_TMP5]], align 8
// CHECK1-NEXT:    store i8** [[TMP67]], i8*** [[TMP66]], align 8
// CHECK1-NEXT:    [[TMP68:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP69:%.*]] = load i32, i32* [[TMP68]], align 4
// CHECK1-NEXT:    [[TMP70:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP69]], i32 1, i64 48, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK1-NEXT:    [[TMP71:%.*]] = bitcast i8* [[TMP70]] to %struct.kmp_task_t_with_privates*
// CHECK1-NEXT:    [[TMP72:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP71]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP73:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP72]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP74:%.*]] = load i8*, i8** [[TMP73]], align 8
// CHECK1-NEXT:    [[TMP75:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK1-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP74]], i8* align 8 [[TMP75]], i64 24, i1 false)
// CHECK1-NEXT:    [[TMP76:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP71]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP77:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP76]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP78:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8
// CHECK1-NEXT:    store i8* [[TMP78]], i8** [[TMP77]], align 8
// CHECK1-NEXT:    [[TMP79:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP80:%.*]] = load i32, i32* [[TMP79]], align 4
// CHECK1-NEXT:    [[TMP81:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP80]], i8* [[TMP70]])
// 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:    [[TMP82:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK1-NEXT:    [[ADD14:%.*]] = add nsw i64 [[TMP82]], 1
// CHECK1-NEXT:    store i64 [[ADD14]], i64* [[DOTOMP_IV]], align 8
// 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:    [[TMP83:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP84:%.*]] = load i32, i32* [[TMP83]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP84]])
// CHECK1-NEXT:    [[TMP85:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP86:%.*]] = load i32, i32* [[TMP85]], align 4
// CHECK1-NEXT:    call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP86]], i32 1)
// CHECK1-NEXT:    [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP88:%.*]] = bitcast i32* [[ARGC1]] to i8*
// CHECK1-NEXT:    store i8* [[TMP88]], i8** [[TMP87]], align 8
// CHECK1-NEXT:    [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    store i8* [[VLA]], i8** [[TMP89]], align 8
// CHECK1-NEXT:    [[TMP90:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP91:%.*]] = inttoptr i64 [[TMP12]] to i8*
// CHECK1-NEXT:    store i8* [[TMP91]], i8** [[TMP90]], align 8
// CHECK1-NEXT:    [[TMP92:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP93:%.*]] = load i32, i32* [[TMP92]], align 4
// CHECK1-NEXT:    [[TMP94:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT:    [[TMP95:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP93]], i32 2, i64 24, i8* [[TMP94]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP95]], 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:    [[TMP96:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP97:%.*]] = load i32, i32* [[ARGC1]], align 4
// CHECK1-NEXT:    [[ADD15:%.*]] = add nsw i32 [[TMP96]], [[TMP97]]
// CHECK1-NEXT:    store i32 [[ADD15]], i32* [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP98:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP12]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP98]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP99:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT:    [[CONV:%.*]] = sext i8 [[TMP99]] to i32
// CHECK1-NEXT:    [[TMP100:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT:    [[CONV17:%.*]] = sext i8 [[TMP100]] to i32
// CHECK1-NEXT:    [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]]
// CHECK1-NEXT:    [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8
// CHECK1-NEXT:    store i8 [[CONV19]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP98]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done22:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP93]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP101:%.*]] = load i32, i32* [[ARGC1]], align 4
// CHECK1-NEXT:    [[TMP102:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP101]] monotonic, align 4
// CHECK1-NEXT:    [[TMP103:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP12]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP103]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK1:       omp.arraycpy.body24:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT:    [[TMP104:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT:    [[CONV27:%.*]] = sext i8 [[TMP104]] to i32
// CHECK1-NEXT:    [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1
// CHECK1-NEXT:    br label [[ATOMIC_CONT:%.*]]
// CHECK1:       atomic_cont:
// CHECK1-NEXT:    [[TMP105:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP110:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT:    store i8 [[TMP105]], i8* [[_TMP28]], align 1
// CHECK1-NEXT:    [[TMP106:%.*]] = load i8, i8* [[_TMP28]], align 1
// CHECK1-NEXT:    [[CONV29:%.*]] = sext i8 [[TMP106]] to i32
// CHECK1-NEXT:    [[TMP107:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT:    [[CONV30:%.*]] = sext i8 [[TMP107]] to i32
// CHECK1-NEXT:    [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]]
// CHECK1-NEXT:    [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8
// CHECK1-NEXT:    store i8 [[CONV32]], i8* [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT:    [[TMP108:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT:    [[TMP109:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP105]], i8 [[TMP108]] monotonic monotonic, align 1
// CHECK1-NEXT:    [[TMP110]] = extractvalue { i8, i1 } [[TMP109]], 0
// CHECK1-NEXT:    [[TMP111:%.*]] = extractvalue { i8, i1 } [[TMP109]], 1
// CHECK1-NEXT:    br i1 [[TMP111]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1:       atomic_exit:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP103]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK1:       omp.arraycpy.done36:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP93]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP112:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore(i8* [[TMP112]])
// CHECK1-NEXT:    [[TMP113:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP114:%.*]] = load i32, i32* [[TMP113]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP114]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_init.
// CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[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 i32*
// CHECK1-NEXT:    store i32 0, i32* [[TMP3]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_comb.
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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 i32*
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32*
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[TMP3]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_init..1
// CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// 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:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.red_comb..2
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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 i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[CONV:%.*]] = sext i8 [[TMP6]] to i32
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT:    [[CONV2:%.*]] = sext i8 [[TMP7]] to i32
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT:    [[CONV3:%.*]] = trunc i32 [[ADD]] to i8
// CHECK1-NEXT:    store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done4:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK1-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca i8***, align 8
// CHECK1-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK1-NEXT:    store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8
// CHECK1-NEXT:    store i8** [[TMP3]], i8*** [[TMP4]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK1-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK1-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK1-NEXT:    [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8
// CHECK1-NEXT:    [[TMP_I:%.*]] = alloca i8**, align 8
// CHECK1-NEXT:    [[TMP4_I:%.*]] = alloca i8*, align 8
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK1-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK1-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK1-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK1-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !11
// CHECK1-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, i8***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)*
// CHECK1-NEXT:    call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]]
// CHECK1-NEXT:    [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !11
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !11
// CHECK1-NEXT:    [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8*
// CHECK1-NEXT:    [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]]
// CHECK1-NEXT:    [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32*
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK1-NEXT:    [[TMP29:%.*]] = sext i32 [[TMP28]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]]
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9
// CHECK1-NEXT:    [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]]
// CHECK1-NEXT:    [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64
// CHECK1-NEXT:    [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64
// CHECK1-NEXT:    [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]]
// CHECK1-NEXT:    [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1
// CHECK1-NEXT:    [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK1-NEXT:    [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK1-NEXT:    [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8
// CHECK1-NEXT:    [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8
// CHECK1-NEXT:    [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64
// CHECK1-NEXT:    [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64
// CHECK1-NEXT:    [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]]
// CHECK1-NEXT:    [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]]
// CHECK1-NEXT:    store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !11
// CHECK1-NEXT:    store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !11
// CHECK1-NEXT:    ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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*]* [[TMP5]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT:    store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[CONV:%.*]] = sext i8 [[TMP22]] to i32
// CHECK1-NEXT:    [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT:    [[CONV2:%.*]] = sext i8 [[TMP23]] to i32
// CHECK1-NEXT:    [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT:    [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK1-NEXT:    store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done5:
// CHECK1-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@main
// CHECK2-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK2-NEXT:    store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK2-NEXT:    store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i8***)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[ARGC_ADDR]], i8*** [[ARGV_ADDR]])
// CHECK2-NEXT:    ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8*** nonnull align 8 dereferenceable(8) [[ARGV:%.*]]) #[[ATTR1:[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:    [[ARGC_ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8***, align 8
// CHECK2-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[TMP:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[ARGC1:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[SAVED_STACK:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[_TMP5:%.*]] = alloca i8**, align 8
// CHECK2-NEXT:    [[_TMP6:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8
// CHECK2-NEXT:    [[DOTTASK_RED_:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[I:%.*]] = alloca i64, align 8
// CHECK2-NEXT:    [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8
// CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8
// CHECK2-NEXT:    [[ATOMIC_TEMP:%.*]] = alloca i8, align 1
// CHECK2-NEXT:    [[_TMP28:%.*]] = alloca i8, align 1
// 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* [[ARGC]], i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT:    store i8*** [[ARGV]], i8**** [[ARGV_ADDR]], align 8
// CHECK2-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8
// CHECK2-NEXT:    [[TMP1:%.*]] = load i8***, i8**** [[ARGV_ADDR]], align 8
// CHECK2-NEXT:    store i64 0, i64* [[DOTOMP_LB]], align 8
// CHECK2-NEXT:    store i64 9, i64* [[DOTOMP_UB]], align 8
// CHECK2-NEXT:    store i64 1, i64* [[DOTOMP_STRIDE]], align 8
// CHECK2-NEXT:    store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK2-NEXT:    store i32 0, i32* [[ARGC1]], align 4
// CHECK2-NEXT:    [[TMP2:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP2]], i64 0
// CHECK2-NEXT:    [[TMP3:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8
// CHECK2-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP3]], i64 0
// CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT:    [[TMP5:%.*]] = sext i32 [[TMP4]] to i64
// CHECK2-NEXT:    [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP5]]
// CHECK2-NEXT:    [[TMP6:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP6]], i64 9
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8
// CHECK2-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP7]], i64 [[LB_ADD_LEN]]
// CHECK2-NEXT:    [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64
// CHECK2-NEXT:    [[TMP9:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
// CHECK2-NEXT:    [[TMP10:%.*]] = sub i64 [[TMP8]], [[TMP9]]
// CHECK2-NEXT:    [[TMP11:%.*]] = sdiv exact i64 [[TMP10]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP12:%.*]] = add nuw i64 [[TMP11]], 1
// CHECK2-NEXT:    [[TMP13:%.*]] = mul nuw i64 [[TMP12]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP14:%.*]] = call i8* @llvm.stacksave()
// CHECK2-NEXT:    store i8* [[TMP14]], i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    [[VLA:%.*]] = alloca i8, i64 [[TMP12]], align 16
// CHECK2-NEXT:    store i64 [[TMP12]], i64* [[__VLA_EXPR0]], align 8
// CHECK2-NEXT:    [[TMP15:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP12]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP15]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK2-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
// CHECK2-NEXT:    [[TMP19:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64
// CHECK2-NEXT:    [[TMP20:%.*]] = sub i64 [[TMP18]], [[TMP19]]
// CHECK2-NEXT:    [[TMP21:%.*]] = sdiv exact i64 [[TMP20]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP22:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP21]]
// CHECK2-NEXT:    store i8** [[_TMP6]], i8*** [[_TMP5]], align 8
// CHECK2-NEXT:    store i8* [[TMP22]], i8** [[_TMP6]], align 8
// CHECK2-NEXT:    [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP24:%.*]] = bitcast i32* [[ARGC1]] to i8*
// CHECK2-NEXT:    store i8* [[TMP24]], i8** [[TMP23]], align 8
// CHECK2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP26:%.*]] = bitcast i32* [[TMP0]] to i8*
// CHECK2-NEXT:    store i8* [[TMP26]], i8** [[TMP25]], align 8
// CHECK2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK2-NEXT:    store i64 4, i64* [[TMP27]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK2-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP28]], align 8
// CHECK2-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK2-NEXT:    store i8* null, i8** [[TMP29]], align 8
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK2-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP30]], align 8
// CHECK2-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK2-NEXT:    [[TMP32:%.*]] = bitcast i32* [[TMP31]] to i8*
// CHECK2-NEXT:    call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP32]], i8 0, i64 4, i1 false)
// CHECK2-NEXT:    [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP34:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i8*, i8** [[TMP34]], i64 0
// CHECK2-NEXT:    [[TMP35:%.*]] = load i8*, i8** [[ARRAYIDX8]], align 8
// CHECK2-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, i8* [[TMP35]], i64 0
// CHECK2-NEXT:    [[TMP36:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT:    [[TMP37:%.*]] = sext i32 [[TMP36]] to i64
// CHECK2-NEXT:    [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP37]]
// CHECK2-NEXT:    [[TMP38:%.*]] = load i8**, i8*** [[TMP1]], align 8
// CHECK2-NEXT:    [[ARRAYIDX11:%.*]] = getelementptr inbounds i8*, i8** [[TMP38]], i64 9
// CHECK2-NEXT:    [[TMP39:%.*]] = load i8*, i8** [[ARRAYIDX11]], align 8
// CHECK2-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[TMP39]], i64 [[LB_ADD_LEN10]]
// CHECK2-NEXT:    store i8* [[VLA]], i8** [[TMP33]], align 8
// CHECK2-NEXT:    [[TMP40:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 1
// CHECK2-NEXT:    store i8* [[ARRAYIDX9]], i8** [[TMP40]], align 8
// CHECK2-NEXT:    [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX12]] to i64
// CHECK2-NEXT:    [[TMP42:%.*]] = ptrtoint i8* [[ARRAYIDX9]] to i64
// CHECK2-NEXT:    [[TMP43:%.*]] = sub i64 [[TMP41]], [[TMP42]]
// CHECK2-NEXT:    [[TMP44:%.*]] = sdiv exact i64 [[TMP43]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP45:%.*]] = add nuw i64 [[TMP44]], 1
// CHECK2-NEXT:    [[TMP46:%.*]] = mul nuw i64 [[TMP45]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 2
// CHECK2-NEXT:    store i64 [[TMP46]], i64* [[TMP47]], align 8
// CHECK2-NEXT:    [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 3
// CHECK2-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP48]], align 8
// CHECK2-NEXT:    [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 4
// CHECK2-NEXT:    store i8* null, i8** [[TMP49]], align 8
// CHECK2-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 5
// CHECK2-NEXT:    store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP50]], align 8
// CHECK2-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 6
// CHECK2-NEXT:    store i32 1, i32* [[TMP51]], align 8
// CHECK2-NEXT:    [[TMP52:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP53:%.*]] = load i32, i32* [[TMP52]], align 4
// CHECK2-NEXT:    [[TMP54:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8*
// CHECK2-NEXT:    [[TMP55:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP53]], i32 1, i32 2, i8* [[TMP54]])
// CHECK2-NEXT:    store i8* [[TMP55]], i8** [[DOTTASK_RED_]], align 8
// CHECK2-NEXT:    [[TMP56:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP57:%.*]] = load i32, i32* [[TMP56]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP57]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK2-NEXT:    [[TMP58:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK2-NEXT:    [[CMP:%.*]] = icmp sgt i64 [[TMP58]], 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:    [[TMP59:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK2-NEXT:    br label [[COND_END]]
// CHECK2:       cond.end:
// CHECK2-NEXT:    [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP59]], [[COND_FALSE]] ]
// CHECK2-NEXT:    store i64 [[COND]], i64* [[DOTOMP_UB]], align 8
// CHECK2-NEXT:    [[TMP60:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8
// CHECK2-NEXT:    store i64 [[TMP60]], i64* [[DOTOMP_IV]], align 8
// CHECK2-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2:       omp.inner.for.cond:
// CHECK2-NEXT:    [[TMP61:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK2-NEXT:    [[TMP62:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8
// CHECK2-NEXT:    [[CMP13:%.*]] = icmp sle i64 [[TMP61]], [[TMP62]]
// CHECK2-NEXT:    br i1 [[CMP13]], 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:    [[TMP63:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK2-NEXT:    [[MUL:%.*]] = mul nsw i64 [[TMP63]], 1
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i64 0, [[MUL]]
// CHECK2-NEXT:    store i64 [[ADD]], i64* [[I]], align 8
// CHECK2-NEXT:    [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0
// CHECK2-NEXT:    store i8** [[DOTTASK_RED_]], i8*** [[TMP64]], align 8
// CHECK2-NEXT:    [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1
// CHECK2-NEXT:    store i32* [[ARGC1]], i32** [[TMP65]], align 8
// CHECK2-NEXT:    [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP67:%.*]] = load i8**, i8*** [[_TMP5]], align 8
// CHECK2-NEXT:    store i8** [[TMP67]], i8*** [[TMP66]], align 8
// CHECK2-NEXT:    [[TMP68:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP69:%.*]] = load i32, i32* [[TMP68]], align 4
// CHECK2-NEXT:    [[TMP70:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP69]], i32 1, i64 48, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*))
// CHECK2-NEXT:    [[TMP71:%.*]] = bitcast i8* [[TMP70]] to %struct.kmp_task_t_with_privates*
// CHECK2-NEXT:    [[TMP72:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP71]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP73:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP72]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP74:%.*]] = load i8*, i8** [[TMP73]], align 8
// CHECK2-NEXT:    [[TMP75:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8*
// CHECK2-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP74]], i8* align 8 [[TMP75]], i64 24, i1 false)
// CHECK2-NEXT:    [[TMP76:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP71]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP77:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP76]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP78:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8
// CHECK2-NEXT:    store i8* [[TMP78]], i8** [[TMP77]], align 8
// CHECK2-NEXT:    [[TMP79:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP80:%.*]] = load i32, i32* [[TMP79]], align 4
// CHECK2-NEXT:    [[TMP81:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP80]], i8* [[TMP70]])
// 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:    [[TMP82:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8
// CHECK2-NEXT:    [[ADD14:%.*]] = add nsw i64 [[TMP82]], 1
// CHECK2-NEXT:    store i64 [[ADD14]], i64* [[DOTOMP_IV]], align 8
// 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:    [[TMP83:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP84:%.*]] = load i32, i32* [[TMP83]], align 4
// CHECK2-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP84]])
// CHECK2-NEXT:    [[TMP85:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP86:%.*]] = load i32, i32* [[TMP85]], align 4
// CHECK2-NEXT:    call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP86]], i32 1)
// CHECK2-NEXT:    [[TMP87:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK2-NEXT:    [[TMP88:%.*]] = bitcast i32* [[ARGC1]] to i8*
// CHECK2-NEXT:    store i8* [[TMP88]], i8** [[TMP87]], align 8
// CHECK2-NEXT:    [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK2-NEXT:    store i8* [[VLA]], i8** [[TMP89]], align 8
// CHECK2-NEXT:    [[TMP90:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP91:%.*]] = inttoptr i64 [[TMP12]] to i8*
// CHECK2-NEXT:    store i8* [[TMP91]], i8** [[TMP90]], align 8
// CHECK2-NEXT:    [[TMP92:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP93:%.*]] = load i32, i32* [[TMP92]], align 4
// CHECK2-NEXT:    [[TMP94:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK2-NEXT:    [[TMP95:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP93]], i32 2, i64 24, i8* [[TMP94]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    switch i32 [[TMP95]], 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:    [[TMP96:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT:    [[TMP97:%.*]] = load i32, i32* [[ARGC1]], align 4
// CHECK2-NEXT:    [[ADD15:%.*]] = add nsw i32 [[TMP96]], [[TMP97]]
// CHECK2-NEXT:    store i32 [[ADD15]], i32* [[TMP0]], align 4
// CHECK2-NEXT:    [[TMP98:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP12]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP98]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP99:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK2-NEXT:    [[CONV:%.*]] = sext i8 [[TMP99]] to i32
// CHECK2-NEXT:    [[TMP100:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK2-NEXT:    [[CONV17:%.*]] = sext i8 [[TMP100]] to i32
// CHECK2-NEXT:    [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]]
// CHECK2-NEXT:    [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8
// CHECK2-NEXT:    store i8 [[CONV19]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP98]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done22:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP93]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.case2:
// CHECK2-NEXT:    [[TMP101:%.*]] = load i32, i32* [[ARGC1]], align 4
// CHECK2-NEXT:    [[TMP102:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP101]] monotonic, align 4
// CHECK2-NEXT:    [[TMP103:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP12]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP103]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK2:       omp.arraycpy.body24:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ]
// CHECK2-NEXT:    [[TMP104:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK2-NEXT:    [[CONV27:%.*]] = sext i8 [[TMP104]] to i32
// CHECK2-NEXT:    [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1
// CHECK2-NEXT:    br label [[ATOMIC_CONT:%.*]]
// CHECK2:       atomic_cont:
// CHECK2-NEXT:    [[TMP105:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP110:%.*]], [[ATOMIC_CONT]] ]
// CHECK2-NEXT:    store i8 [[TMP105]], i8* [[_TMP28]], align 1
// CHECK2-NEXT:    [[TMP106:%.*]] = load i8, i8* [[_TMP28]], align 1
// CHECK2-NEXT:    [[CONV29:%.*]] = sext i8 [[TMP106]] to i32
// CHECK2-NEXT:    [[TMP107:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK2-NEXT:    [[CONV30:%.*]] = sext i8 [[TMP107]] to i32
// CHECK2-NEXT:    [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]]
// CHECK2-NEXT:    [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8
// CHECK2-NEXT:    store i8 [[CONV32]], i8* [[ATOMIC_TEMP]], align 1
// CHECK2-NEXT:    [[TMP108:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1
// CHECK2-NEXT:    [[TMP109:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP105]], i8 [[TMP108]] monotonic monotonic, align 1
// CHECK2-NEXT:    [[TMP110]] = extractvalue { i8, i1 } [[TMP109]], 0
// CHECK2-NEXT:    [[TMP111:%.*]] = extractvalue { i8, i1 } [[TMP109]], 1
// CHECK2-NEXT:    br i1 [[TMP111]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK2:       atomic_exit:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP103]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK2:       omp.arraycpy.done36:
// CHECK2-NEXT:    call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP93]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK2:       .omp.reduction.default:
// CHECK2-NEXT:    [[TMP112:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8
// CHECK2-NEXT:    call void @llvm.stackrestore(i8* [[TMP112]])
// CHECK2-NEXT:    [[TMP113:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT:    [[TMP114:%.*]] = load i32, i32* [[TMP113]], align 4
// CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP114]])
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.red_init.
// CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[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 i32*
// CHECK2-NEXT:    store i32 0, i32* [[TMP3]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.red_comb.
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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 i32*
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32*
// CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8
// CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[TMP3]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.red_init..1
// CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// 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:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]]
// CHECK2-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK2:       omp.arrayinit.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK2-NEXT:    store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK2:       omp.arrayinit.done:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.red_comb..2
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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 i64, i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[CONV:%.*]] = sext i8 [[TMP6]] to i32
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK2-NEXT:    [[CONV2:%.*]] = sext i8 [[TMP7]] to i32
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK2-NEXT:    [[CONV3:%.*]] = trunc i32 [[ADD]] to i8
// CHECK2-NEXT:    store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done4:
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map.
// CHECK2-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca i8***, align 8
// CHECK2-NEXT:    store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK2-NEXT:    store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8
// CHECK2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8
// CHECK2-NEXT:    store i8** [[TMP3]], i8*** [[TMP4]], align 8
// CHECK2-NEXT:    ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT:  entry:
// CHECK2-NEXT:    [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8
// CHECK2-NEXT:    [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8
// CHECK2-NEXT:    [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8
// CHECK2-NEXT:    [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8
// CHECK2-NEXT:    [[TMP_I:%.*]] = alloca i8**, align 8
// CHECK2-NEXT:    [[TMP4_I:%.*]] = alloca i8*, align 8
// CHECK2-NEXT:    [[DOTADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT:    [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8
// CHECK2-NEXT:    store i32 [[TMP0]], i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4
// CHECK2-NEXT:    [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8
// CHECK2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0
// CHECK2-NEXT:    [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK2-NEXT:    [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon*
// CHECK2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8*
// CHECK2-NEXT:    [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8*
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META2:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
// CHECK2-NEXT:    call void @llvm.experimental.noalias.scope.decl(metadata [[META9:![0-9]+]])
// CHECK2-NEXT:    store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !11
// CHECK2-NEXT:    store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, i8***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)*
// CHECK2-NEXT:    call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]]
// CHECK2-NEXT:    [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !11
// CHECK2-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8
// CHECK2-NEXT:    [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !11
// CHECK2-NEXT:    [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8*
// CHECK2-NEXT:    [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]]
// CHECK2-NEXT:    [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32*
// CHECK2-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8
// CHECK2-NEXT:    [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8
// CHECK2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1
// CHECK2-NEXT:    [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8
// CHECK2-NEXT:    [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4
// CHECK2-NEXT:    [[TMP29:%.*]] = sext i32 [[TMP28]] to i64
// CHECK2-NEXT:    [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]]
// CHECK2-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8
// CHECK2-NEXT:    [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9
// CHECK2-NEXT:    [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8
// CHECK2-NEXT:    [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]]
// CHECK2-NEXT:    [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64
// CHECK2-NEXT:    [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64
// CHECK2-NEXT:    [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]]
// CHECK2-NEXT:    [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1
// CHECK2-NEXT:    [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8
// CHECK2-NEXT:    [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]]
// CHECK2-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2
// CHECK2-NEXT:    [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8
// CHECK2-NEXT:    [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8
// CHECK2-NEXT:    [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64
// CHECK2-NEXT:    [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64
// CHECK2-NEXT:    [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]]
// CHECK2-NEXT:    [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64)
// CHECK2-NEXT:    [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]]
// CHECK2-NEXT:    store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !11
// CHECK2-NEXT:    store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !11
// CHECK2-NEXT:    ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] {
// 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*]* [[TMP5]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8
// CHECK2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1
// CHECK2-NEXT:    [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8
// CHECK2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2
// CHECK2-NEXT:    [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8
// CHECK2-NEXT:    [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64
// CHECK2-NEXT:    [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK2-NEXT:    [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK2-NEXT:    store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK2-NEXT:    [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK2:       omp.arraycpy.body:
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK2-NEXT:    [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[CONV:%.*]] = sext i8 [[TMP22]] to i32
// CHECK2-NEXT:    [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK2-NEXT:    [[CONV2:%.*]] = sext i8 [[TMP23]] to i32
// CHECK2-NEXT:    [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK2-NEXT:    [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK2-NEXT:    store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK2-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]]
// CHECK2-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK2:       omp.arraycpy.done5:
// CHECK2-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK3-NEXT:    [[I:%.*]] = alloca i64, align 8
// CHECK3-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK3-NEXT:    store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK3-NEXT:    store i64 0, i64* [[I]], align 8
// CHECK3-NEXT:    br label [[FOR_COND:%.*]]
// CHECK3:       for.cond:
// CHECK3-NEXT:    [[TMP0:%.*]] = load i64, i64* [[I]], align 8
// CHECK3-NEXT:    [[CMP:%.*]] = icmp slt i64 [[TMP0]], 10
// CHECK3-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]
// CHECK3:       for.body:
// CHECK3-NEXT:    br label [[FOR_INC:%.*]]
// CHECK3:       for.inc:
// CHECK3-NEXT:    [[TMP1:%.*]] = load i64, i64* [[I]], align 8
// CHECK3-NEXT:    [[INC:%.*]] = add nsw i64 [[TMP1]], 1
// CHECK3-NEXT:    store i64 [[INC]], i64* [[I]], align 8
// CHECK3-NEXT:    br label [[FOR_COND]], !llvm.loop [[LOOP2:![0-9]+]]
// CHECK3:       for.end:
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK3-NEXT:    ret i32 [[TMP2]]
//
//
// CHECK4-LABEL: define {{[^@]+}}@main
// CHECK4-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] !dbg [[DBG6:![0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[ARGV_ADDR:%.*]] = alloca i8**, align 8
// CHECK4-NEXT:    [[I:%.*]] = alloca i64, align 8
// CHECK4-NEXT:    store i32 0, i32* [[RETVAL]], align 4
// CHECK4-NEXT:    store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4
// CHECK4-NEXT:    call void @llvm.dbg.declare(metadata i32* [[ARGC_ADDR]], metadata [[META14:![0-9]+]], metadata !DIExpression()), !dbg [[DBG15:![0-9]+]]
// CHECK4-NEXT:    store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8
// CHECK4-NEXT:    call void @llvm.dbg.declare(metadata i8*** [[ARGV_ADDR]], metadata [[META16:![0-9]+]], metadata !DIExpression()), !dbg [[DBG17:![0-9]+]]
// CHECK4-NEXT:    call void @llvm.dbg.declare(metadata i64* [[I]], metadata [[META18:![0-9]+]], metadata !DIExpression()), !dbg [[DBG23:![0-9]+]]
// CHECK4-NEXT:    store i64 0, i64* [[I]], align 8, !dbg [[DBG23]]
// CHECK4-NEXT:    br label [[FOR_COND:%.*]], !dbg [[DBG24:![0-9]+]]
// CHECK4:       for.cond:
// CHECK4-NEXT:    [[TMP0:%.*]] = load i64, i64* [[I]], align 8, !dbg [[DBG25:![0-9]+]]
// CHECK4-NEXT:    [[CMP:%.*]] = icmp slt i64 [[TMP0]], 10, !dbg [[DBG27:![0-9]+]]
// CHECK4-NEXT:    br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]], !dbg [[DBG28:![0-9]+]]
// CHECK4:       for.body:
// CHECK4-NEXT:    br label [[FOR_INC:%.*]], !dbg [[DBG29:![0-9]+]]
// CHECK4:       for.inc:
// CHECK4-NEXT:    [[TMP1:%.*]] = load i64, i64* [[I]], align 8, !dbg [[DBG31:![0-9]+]]
// CHECK4-NEXT:    [[INC:%.*]] = add nsw i64 [[TMP1]], 1, !dbg [[DBG31]]
// CHECK4-NEXT:    store i64 [[INC]], i64* [[I]], align 8, !dbg [[DBG31]]
// CHECK4-NEXT:    br label [[FOR_COND]], !dbg [[DBG32:![0-9]+]], !llvm.loop [[LOOP33:![0-9]+]]
// CHECK4:       for.end:
// CHECK4-NEXT:    [[TMP2:%.*]] = load i32, i32* [[RETVAL]], align 4, !dbg [[DBG36:![0-9]+]]
// CHECK4-NEXT:    ret i32 [[TMP2]], !dbg [[DBG36]]
//