Google Git
Sign in
llvm / llvm-project / clang / 5ddec09286a9f8daf4e16015a6bf2f6023e4c284 / . / test / OpenMP / distribute_parallel_for_num_threads_codegen.cpp
blob: b1a573ef88561eee96d640024a9469db6f7c1d02 [file] [log] [blame]
// 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 -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2
// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK6
// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK10
// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-version=45 -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -verify -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK13
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK14
// RUN: %clang_cc1 -verify -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -std=c++11 -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -fopenmp-targets=powerpc64le-ibm-linux-gnu -x c++ -triple aarch64-unknown-unknown -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
typedef __INTPTR_TYPE__ intptr_t;
void foo();
struct S {
intptr_t a, b, c;
S(intptr_t a) : a(a) {}
operator char() { extern void mayThrow(); mayThrow(); return a; }
~S() {}
};
template <typename T, int C>
int tmain() {
#pragma omp target
#pragma omp teams
#pragma omp distribute parallel for num_threads(C)
for (int i = 0; i < 100; i++)
foo();
#pragma omp target
#pragma omp teams
#pragma omp distribute parallel for num_threads(T(23))
for (int i = 0; i < 100; i++)
foo();
return 0;
}
int main() {
S s(0);
char a = s;
#pragma omp target
#pragma omp teams
#pragma omp distribute parallel for num_threads(2)
for (int i = 0; i < 100; i++) {
foo();
}
#pragma omp target
#pragma omp teams
#pragma omp distribute parallel for num_threads(a)
for (int i = 0; i < 100; i++) {
foo();
}
return a + tmain<char, 5>() + tmain<S, 1>();
}
// tmain 5
// tmain 1
#endif
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK1-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK1-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK1-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK1: invoke.cont:
// CHECK1-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: lpad:
// CHECK1-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: cleanup
// CHECK1-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK1-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK1-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK1-NEXT: br label [[EH_RESUME:%.*]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK1-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK1: omp_offload.failed2:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK1: omp_offload.cont3:
// CHECK1-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK1-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK1-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK1-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK1: invoke.cont5:
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK1-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK1-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK1: invoke.cont7:
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK1-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK1-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK1-NEXT: ret i32 [[TMP17]]
// CHECK1: eh.resume:
// CHECK1-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK1-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK1-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK1-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK1-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK1-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK1-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK1-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat 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 @_Z8mayThrowv()
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK1-NEXT: ret i8 [[CONV]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK1-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK1-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK1-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK1-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK1-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK1: omp_offload.failed2:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK1: omp_offload.cont3:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK1-SAME: () #[[ATTR7]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK1-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK1: omp_offload.failed2:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK1: omp_offload.cont3:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK1-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat 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 @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK1-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK1-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat 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 {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK1-SAME: () #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK1-SAME: () #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK1-SAME: () #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK1-SAME: () #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK1-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// CHECK1-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK1-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK1: invoke.cont2:
// CHECK1-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK1-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK1-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK1: lpad:
// CHECK1-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK1-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK1-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK1-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK1-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
// CHECK1: terminate.handler:
// CHECK1-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: invoke void @_Z3foov()
// CHECK1-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK1: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void
// CHECK1: terminate.lpad:
// CHECK1-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK1-NEXT: catch i8* null
// CHECK1-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK1-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK1-NEXT: unreachable
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@main
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK2-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK2-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK2-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK2-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK2: invoke.cont:
// CHECK2-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: lpad:
// CHECK2-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: cleanup
// CHECK2-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK2-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK2-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK2-NEXT: br label [[EH_RESUME:%.*]]
// CHECK2: omp_offload.cont:
// CHECK2-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK2-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK2-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK2-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK2-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK2: omp_offload.failed2:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK2: omp_offload.cont3:
// CHECK2-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK2-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK2-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK2-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK2: invoke.cont5:
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK2-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK2-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK2: invoke.cont7:
// CHECK2-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK2-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK2-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK2-NEXT: ret i32 [[TMP17]]
// CHECK2: eh.resume:
// CHECK2-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK2-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK2-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK2-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK2-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK2-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK2-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK2-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat 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 @_Z8mayThrowv()
// CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK2-NEXT: ret i8 [[CONV]]
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK2-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK2-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK2-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK2-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK2-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK2-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK2-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK2-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP2]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK2-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont:
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK2: omp_offload.failed2:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK2: omp_offload.cont3:
// CHECK2-NEXT: ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK2-SAME: () #[[ATTR7]] comdat {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK2-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK2: omp_offload.failed:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK2: omp_offload.cont:
// CHECK2-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK2: omp_offload.failed2:
// CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK2-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK2: omp_offload.cont3:
// CHECK2-NEXT: ret i32 0
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK2-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat 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 @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK2-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK2-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK2-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK2-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat 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 {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK2-SAME: () #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK2-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK2-SAME: () #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK2-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK2-SAME: () #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK2-NEXT: store i32 [[ADD]], 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK2-SAME: () #[[ATTR3]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK2-NEXT: ret void
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK2-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK2-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_COMB_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: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// 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: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK2-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// CHECK2-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK2-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK2: invoke.cont2:
// CHECK2-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK2-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK2-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK2-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK2: omp.inner.for.inc:
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK2: lpad:
// CHECK2-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK2-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK2-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK2-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK2-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// 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* @[[GLOB1]], i32 [[TMP1]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
// CHECK2: terminate.handler:
// CHECK2-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, 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: [[I:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK2-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK2-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK2-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK2-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[CONV1]], 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 i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// 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: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: br label [[COND_END]]
// CHECK2: cond.end:
// CHECK2-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK2-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK2-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK2: omp.inner.for.cond:
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK2-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK2-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK2: omp.inner.for.body:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK2-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK2-NEXT: invoke void @_Z3foov()
// CHECK2-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK2: invoke.cont:
// 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: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK2-NEXT: store i32 [[ADD3]], 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* @[[GLOB1]], i32 [[TMP3]])
// CHECK2-NEXT: ret void
// CHECK2: terminate.lpad:
// CHECK2-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK2-NEXT: catch i8* null
// CHECK2-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK2-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK2-NEXT: unreachable
//
//
// CHECK2-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK2-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK2-NEXT: entry:
// CHECK2-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK2-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@main
// CHECK5-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK5-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK5-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK5-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK5-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK5-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK5-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: lpad:
// CHECK5-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: cleanup
// CHECK5-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK5-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK5-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK5-NEXT: br label [[EH_RESUME:%.*]]
// CHECK5: omp_offload.cont:
// CHECK5-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK5-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK5-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK5-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK5-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK5-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK5-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK5-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK5: omp_offload.failed2:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK5: omp_offload.cont3:
// CHECK5-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK5-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK5-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK5-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK5: invoke.cont5:
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK5-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK5-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK5: invoke.cont7:
// CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK5-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK5-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK5-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK5-NEXT: ret i32 [[TMP17]]
// CHECK5: eh.resume:
// CHECK5-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK5-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK5-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK5-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK5-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK5-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK5-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK5-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK5-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK5-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: call void @_Z8mayThrowv()
// CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK5-NEXT: ret i8 [[CONV]]
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK5-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK5-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK5-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK5-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK5-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK5-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK5-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont:
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK5-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK5: omp_offload.failed2:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK5: omp_offload.cont3:
// CHECK5-NEXT: ret i32 0
//
//
// CHECK5-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK5-SAME: () #[[ATTR7]] comdat {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK5-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK5: omp_offload.failed:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK5: omp_offload.cont:
// CHECK5-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK5-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK5-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK5-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK5: omp_offload.failed2:
// CHECK5-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK5-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK5: omp_offload.cont3:
// CHECK5-NEXT: ret i32 0
//
//
// CHECK5-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK5-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK5-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK5-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK5-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK5-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK5-SAME: () #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK5-SAME: () #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK5-SAME: () #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK5-SAME: () #[[ATTR3]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK5-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK5-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK5-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK5-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK5: invoke.cont2:
// CHECK5-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK5-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK5-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK5-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: lpad:
// CHECK5-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK5-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK5-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK5-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK5-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK5-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
// CHECK5: terminate.handler:
// CHECK5-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK5-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK5-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK5-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK5-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK5-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK5-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true:
// CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false:
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK5-NEXT: invoke void @_Z3foov()
// CHECK5-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK5: invoke.cont:
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK5-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK5-NEXT: ret void
// CHECK5: terminate.lpad:
// CHECK5-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK5-NEXT: catch i8* null
// CHECK5-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK5-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK5-NEXT: unreachable
//
//
// CHECK5-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK5-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK5-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK5-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK5-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK5-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@main
// CHECK6-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK6-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK6-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK6-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK6-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK6-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK6-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: lpad:
// CHECK6-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: cleanup
// CHECK6-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK6-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK6-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK6-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK6-NEXT: br label [[EH_RESUME:%.*]]
// CHECK6: omp_offload.cont:
// CHECK6-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK6-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK6-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK6-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK6-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK6-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK6-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK6-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK6-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK6-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK6-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK6-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK6: omp_offload.failed2:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK6: omp_offload.cont3:
// CHECK6-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK6-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK6-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK6-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK6: invoke.cont5:
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK6-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK6-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK6: invoke.cont7:
// CHECK6-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK6-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK6-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK6-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK6-NEXT: ret i32 [[TMP17]]
// CHECK6: eh.resume:
// CHECK6-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK6-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK6-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK6-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK6-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK6-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK6-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK6-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK6-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK6-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: call void @_Z8mayThrowv()
// CHECK6-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK6-NEXT: ret i8 [[CONV]]
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK6-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK6-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK6-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK6-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK6-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK6-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK6-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK6-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont:
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK6-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK6: omp_offload.failed2:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK6: omp_offload.cont3:
// CHECK6-NEXT: ret i32 0
//
//
// CHECK6-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK6-SAME: () #[[ATTR7]] comdat {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK6-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK6: omp_offload.failed:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK6: omp_offload.cont:
// CHECK6-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK6-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK6-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK6-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK6: omp_offload.failed2:
// CHECK6-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK6-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK6: omp_offload.cont3:
// CHECK6-NEXT: ret i32 0
//
//
// CHECK6-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK6-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK6-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK6-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK6-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK6-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK6-SAME: () #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK6-SAME: () #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK6-SAME: () #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK6-SAME: () #[[ATTR3]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK6-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK6-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK6-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: [[CALL:%.*]] = invoke signext i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK6-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK6: invoke.cont2:
// CHECK6-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK6-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK6-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK6-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK6-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK6-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: lpad:
// CHECK6-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK6-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK6-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK6-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK6-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK6-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
// CHECK6: terminate.handler:
// CHECK6-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK6-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK6-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK6-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK6-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK6-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK6-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK6-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK6-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK6-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK6-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK6-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK6-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK6-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK6-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK6: cond.true:
// CHECK6-NEXT: br label [[COND_END:%.*]]
// CHECK6: cond.false:
// CHECK6-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: br label [[COND_END]]
// CHECK6: cond.end:
// CHECK6-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK6-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK6-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK6: omp.inner.for.cond:
// CHECK6-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK6-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK6-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK6: omp.inner.for.body:
// CHECK6-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK6-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK6-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK6-NEXT: invoke void @_Z3foov()
// CHECK6-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK6: invoke.cont:
// CHECK6-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK6: omp.body.continue:
// CHECK6-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK6: omp.inner.for.inc:
// CHECK6-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK6-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK6-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK6: omp.inner.for.end:
// CHECK6-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK6: omp.loop.exit:
// CHECK6-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK6-NEXT: ret void
// CHECK6: terminate.lpad:
// CHECK6-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK6-NEXT: catch i8* null
// CHECK6-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK6-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK6-NEXT: unreachable
//
//
// CHECK6-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK6-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK6-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK6-NEXT: ret void
//
//
// CHECK6-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK6-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK6-NEXT: entry:
// CHECK6-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK6-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK9-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK9-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK9-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: lpad:
// CHECK9-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: cleanup
// CHECK9-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK9-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK9-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK9-NEXT: br label [[EH_RESUME:%.*]]
// CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK9-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK9-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK9-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK9: omp_offload.failed2:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK9: omp_offload.cont3:
// CHECK9-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK9-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK9-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK9-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK9: invoke.cont5:
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK9-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK9-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK9: invoke.cont7:
// CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK9-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP17]]
// CHECK9: eh.resume:
// CHECK9-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK9-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK9-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK9-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK9-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK9-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK9-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK9-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: call void @_Z8mayThrowv()
// CHECK9-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK9-NEXT: ret i8 [[CONV]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK9-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK9-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK9-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK9-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK9-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK9-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont:
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK9: omp_offload.failed2:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK9: omp_offload.cont3:
// CHECK9-NEXT: ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK9-SAME: () #[[ATTR7]] comdat {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK9-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont:
// CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK9-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK9-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK9-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK9: omp_offload.failed2:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK9: omp_offload.cont3:
// CHECK9-NEXT: ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK9-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK9-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK9-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK9-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK9-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK9-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK9: invoke.cont2:
// CHECK9-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK9-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK9-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: lpad:
// CHECK9-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK9-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK9-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK9-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK9-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
// CHECK9: terminate.handler:
// CHECK9-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK9-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true:
// CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false:
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: invoke void @_Z3foov()
// CHECK9-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK9: invoke.cont:
// CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9: omp.body.continue:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9: omp.inner.for.inc:
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK9-NEXT: ret void
// CHECK9: terminate.lpad:
// CHECK9-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK9-NEXT: catch i8* null
// CHECK9-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK9-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK9-NEXT: unreachable
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK9-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK9-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@main
// CHECK10-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK10-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK10-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK10-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK10-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK10-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: lpad:
// CHECK10-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: cleanup
// CHECK10-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK10-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK10-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK10-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK10-NEXT: br label [[EH_RESUME:%.*]]
// CHECK10: omp_offload.cont:
// CHECK10-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK10-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK10-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK10-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK10-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK10-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK10-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK10-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK10-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK10-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK10-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK10: omp_offload.failed2:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK10: omp_offload.cont3:
// CHECK10-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK10-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK10-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK10-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK10: invoke.cont5:
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK10-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK10-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK10: invoke.cont7:
// CHECK10-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK10-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK10-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK10-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK10-NEXT: ret i32 [[TMP17]]
// CHECK10: eh.resume:
// CHECK10-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK10-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK10-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK10-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK10-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK10-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK10-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK10-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK10-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK10-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK10-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: call void @_Z8mayThrowv()
// CHECK10-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK10-NEXT: ret i8 [[CONV]]
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK10-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK10-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK10-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK10-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK10-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK10-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK10-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK10-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont:
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK10: omp_offload.failed2:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK10: omp_offload.cont3:
// CHECK10-NEXT: ret i32 0
//
//
// CHECK10-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK10-SAME: () #[[ATTR7]] comdat {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK10-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK10: omp_offload.failed:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK10: omp_offload.cont:
// CHECK10-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK10-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK10-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK10-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK10: omp_offload.failed2:
// CHECK10-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK10-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK10: omp_offload.cont3:
// CHECK10-NEXT: ret i32 0
//
//
// CHECK10-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK10-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK10-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK10-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK10-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK10-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK10-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK10-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK10-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK10-SAME: () #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK10-SAME: () #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK10-SAME: () #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK10-SAME: () #[[ATTR3]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK10-NEXT: ret void
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK10-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK10-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK10-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK10-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK10: invoke.cont2:
// CHECK10-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK10-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK10-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK10-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK10-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK10-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: lpad:
// CHECK10-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK10-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK10-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK10-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK10-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK10-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
// CHECK10: terminate.handler:
// CHECK10-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK10-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK10-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK10-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK10-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK10-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK10-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK10-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK10-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK10-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK10-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK10-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK10-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK10-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK10-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK10: cond.true:
// CHECK10-NEXT: br label [[COND_END:%.*]]
// CHECK10: cond.false:
// CHECK10-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: br label [[COND_END]]
// CHECK10: cond.end:
// CHECK10-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK10-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK10-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK10: omp.inner.for.cond:
// CHECK10-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK10-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK10-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK10: omp.inner.for.body:
// CHECK10-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK10-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK10-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK10-NEXT: invoke void @_Z3foov()
// CHECK10-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK10: invoke.cont:
// CHECK10-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK10: omp.body.continue:
// CHECK10-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK10: omp.inner.for.inc:
// CHECK10-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK10-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK10-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK10: omp.inner.for.end:
// CHECK10-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK10: omp.loop.exit:
// CHECK10-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK10-NEXT: ret void
// CHECK10: terminate.lpad:
// CHECK10-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK10-NEXT: catch i8* null
// CHECK10-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK10-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK10-NEXT: unreachable
//
//
// CHECK10-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK10-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK10-NEXT: entry:
// CHECK10-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK10-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@main
// CHECK13-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK13-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK13-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK13-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK13-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK13-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK13-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK13-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK13: omp_offload.failed:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK13: lpad:
// CHECK13-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: cleanup
// CHECK13-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK13-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK13-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK13-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK13-NEXT: br label [[EH_RESUME:%.*]]
// CHECK13: omp_offload.cont:
// CHECK13-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK13-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK13-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK13-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK13-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK13-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK13-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK13-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK13-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK13-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK13-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK13-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK13-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK13-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK13: omp_offload.failed2:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK13: omp_offload.cont3:
// CHECK13-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK13-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK13-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK13-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK13: invoke.cont5:
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK13-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK13-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK13: invoke.cont7:
// CHECK13-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK13-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK13-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK13-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK13-NEXT: ret i32 [[TMP17]]
// CHECK13: eh.resume:
// CHECK13-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK13-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK13-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK13-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK13-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK13-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK13-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK13-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK13-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK13-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK13-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK13-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: call void @_Z8mayThrowv()
// CHECK13-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK13-NEXT: ret i8 [[CONV]]
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK13-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK13-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK13-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK13-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK13-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK13-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK13-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK13-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK13: omp_offload.failed:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK13: omp_offload.cont:
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK13-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK13: omp_offload.failed2:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK13: omp_offload.cont3:
// CHECK13-NEXT: ret i32 0
//
//
// CHECK13-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK13-SAME: () #[[ATTR7]] comdat {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK13-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK13: omp_offload.failed:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK13: omp_offload.cont:
// CHECK13-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK13-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK13-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK13-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK13: omp_offload.failed2:
// CHECK13-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK13-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK13: omp_offload.cont3:
// CHECK13-NEXT: ret i32 0
//
//
// CHECK13-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK13-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK13-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK13-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK13-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK13-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK13-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK13-SAME: () #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK13-SAME: () #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK13-SAME: () #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK13-SAME: () #[[ATTR3]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK13-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK13-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK13-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK13-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK13: invoke.cont2:
// CHECK13-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK13-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK13-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK13-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK13-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK13-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: lpad:
// CHECK13-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK13-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK13-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK13-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK13-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK13-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
// CHECK13: terminate.handler:
// CHECK13-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK13-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK13-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK13-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK13-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK13-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK13-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK13-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK13-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK13-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK13: cond.true:
// CHECK13-NEXT: br label [[COND_END:%.*]]
// CHECK13: cond.false:
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: br label [[COND_END]]
// CHECK13: cond.end:
// CHECK13-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK13-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK13-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK13-NEXT: invoke void @_Z3foov()
// CHECK13-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK13: invoke.cont:
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK13-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK13: omp.inner.for.end:
// CHECK13-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK13: omp.loop.exit:
// CHECK13-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK13-NEXT: ret void
// CHECK13: terminate.lpad:
// CHECK13-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK13-NEXT: catch i8* null
// CHECK13-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK13-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK13-NEXT: unreachable
//
//
// CHECK13-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK13-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK13-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK13-NEXT: ret void
//
//
// CHECK13-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK13-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK13-NEXT: entry:
// CHECK13-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK13-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@main
// CHECK14-SAME: () #[[ATTR0:[0-9]+]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[S:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK14-NEXT: [[A:%.*]] = alloca i8, align 1
// CHECK14-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK14-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK14-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK14-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK14-NEXT: call void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[S]], i64 0)
// CHECK14-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[S]])
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: store i8 [[CALL]], i8* [[A]], align 1
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK14: omp_offload.failed:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68() #[[ATTR6:[0-9]+]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK14: lpad:
// CHECK14-NEXT: [[TMP2:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: cleanup
// CHECK14-NEXT: [[TMP3:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 0
// CHECK14-NEXT: store i8* [[TMP3]], i8** [[EXN_SLOT]], align 8
// CHECK14-NEXT: [[TMP4:%.*]] = extractvalue { i8*, i32 } [[TMP2]], 1
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK14-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK14-NEXT: br label [[EH_RESUME:%.*]]
// CHECK14: omp_offload.cont:
// CHECK14-NEXT: [[TMP5:%.*]] = load i8, i8* [[A]], align 1
// CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[A_CASTED]] to i8*
// CHECK14-NEXT: store i8 [[TMP5]], i8* [[CONV]], align 1
// CHECK14-NEXT: [[TMP6:%.*]] = load i64, i64* [[A_CASTED]], align 8
// CHECK14-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK14-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64*
// CHECK14-NEXT: store i64 [[TMP6]], i64* [[TMP8]], align 8
// CHECK14-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK14-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64*
// CHECK14-NEXT: store i64 [[TMP6]], i64* [[TMP10]], align 8
// CHECK14-NEXT: [[TMP11:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK14-NEXT: store i8* null, i8** [[TMP11]], align 8
// CHECK14-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK14-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74.region_id, i32 1, i8** [[TMP12]], i8** [[TMP13]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK14-NEXT: br i1 [[TMP15]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK14: omp_offload.failed2:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74(i64 [[TMP6]]) #[[ATTR6]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK14: omp_offload.cont3:
// CHECK14-NEXT: [[TMP16:%.*]] = load i8, i8* [[A]], align 1
// CHECK14-NEXT: [[CONV4:%.*]] = sext i8 [[TMP16]] to i32
// CHECK14-NEXT: [[CALL6:%.*]] = invoke i32 @_Z5tmainIcLi5EEiv()
// CHECK14-NEXT: to label [[INVOKE_CONT5:%.*]] unwind label [[LPAD]]
// CHECK14: invoke.cont5:
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV4]], [[CALL6]]
// CHECK14-NEXT: [[CALL8:%.*]] = invoke i32 @_Z5tmainI1SLi1EEiv()
// CHECK14-NEXT: to label [[INVOKE_CONT7:%.*]] unwind label [[LPAD]]
// CHECK14: invoke.cont7:
// CHECK14-NEXT: [[ADD9:%.*]] = add nsw i32 [[ADD]], [[CALL8]]
// CHECK14-NEXT: store i32 [[ADD9]], i32* [[RETVAL]], align 4
// CHECK14-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[S]]) #[[ATTR6]]
// CHECK14-NEXT: [[TMP17:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK14-NEXT: ret i32 [[TMP17]]
// CHECK14: eh.resume:
// CHECK14-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK14-NEXT: [[SEL:%.*]] = load i32, i32* [[EHSELECTOR_SLOT]], align 4
// CHECK14-NEXT: [[LPAD_VAL:%.*]] = insertvalue { i8*, i32 } undef, i8* [[EXN]], 0
// CHECK14-NEXT: [[LPAD_VAL10:%.*]] = insertvalue { i8*, i32 } [[LPAD_VAL]], i32 [[SEL]], 1
// CHECK14-NEXT: resume { i8*, i32 } [[LPAD_VAL10]]
//
//
// CHECK14-LABEL: define {{[^@]+}}@_ZN1SC1El
// CHECK14-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK14-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK14-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK14-NEXT: call void @_ZN1SC2El(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]], i64 [[TMP0]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@_ZN1ScvcEv
// CHECK14-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK14-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: call void @_Z8mayThrowv()
// CHECK14-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[A]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i8
// CHECK14-NEXT: ret i8 [[CONV]]
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK14-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 2)
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10:[0-9]+]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@__clang_call_terminate
// CHECK14-SAME: (i8* [[TMP0:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK14-NEXT: [[TMP2:%.*]] = call i8* @__cxa_begin_catch(i8* [[TMP0]]) #[[ATTR6]]
// CHECK14-NEXT: call void @_ZSt9terminatev() #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l74
// CHECK14-SAME: (i64 [[A:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = bitcast i64* [[A_ADDR]] to i8*
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i8*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i8* [[CONV]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i8* nonnull align 1 dereferenceable(1) [[A:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[A_ADDR:%.*]] = alloca i8*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i8* [[A]], i8** [[A_ADDR]], align 8
// CHECK14-NEXT: [[TMP0:%.*]] = load i8*, i8** [[A_ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP8:%.*]] = load i8, i8* [[TMP0]], align 1
// CHECK14-NEXT: [[TMP9:%.*]] = sext i8 [[TMP8]] to i32
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP2]], i32 [[TMP9]])
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP11]], i64 [[TMP13]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..3
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@_Z5tmainIcLi5EEiv
// CHECK14-SAME: () #[[ATTR7:[0-9]+]] comdat {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK14: omp_offload.failed:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52() #[[ATTR6]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK14: omp_offload.cont:
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK14-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK14: omp_offload.failed2:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57() #[[ATTR6]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK14: omp_offload.cont3:
// CHECK14-NEXT: ret i32 0
//
//
// CHECK14-LABEL: define {{[^@]+}}@_Z5tmainI1SLi1EEiv
// CHECK14-SAME: () #[[ATTR7]] comdat {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP0:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP1:%.*]] = icmp ne i32 [[TMP0]], 0
// CHECK14-NEXT: br i1 [[TMP1]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK14: omp_offload.failed:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52() #[[ATTR6]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK14: omp_offload.cont:
// CHECK14-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 100)
// CHECK14-NEXT: [[TMP2:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57.region_id, i32 0, i8** null, i8** null, i64* null, i64* null, i8** null, i8** null, i32 0, i32 0)
// CHECK14-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
// CHECK14-NEXT: br i1 [[TMP3]], label [[OMP_OFFLOAD_FAILED2:%.*]], label [[OMP_OFFLOAD_CONT3:%.*]]
// CHECK14: omp_offload.failed2:
// CHECK14-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57() #[[ATTR6]]
// CHECK14-NEXT: br label [[OMP_OFFLOAD_CONT3]]
// CHECK14: omp_offload.cont3:
// CHECK14-NEXT: ret i32 0
//
//
// CHECK14-LABEL: define {{[^@]+}}@_ZN1SD1Ev
// CHECK14-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8:[0-9]+]] comdat align 2 {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK14-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: call void @_ZN1SD2Ev(%struct.S* nonnull align 8 dereferenceable(24) [[THIS1]]) #[[ATTR6]]
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@_ZN1SC2El
// CHECK14-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]], i64 [[A:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK14-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: store i64 [[A]], i64* [[A_ADDR]], align 8
// CHECK14-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[A_ADDR]], align 8
// CHECK14-NEXT: store i64 [[TMP0]], i64* [[A2]], align 8
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l52
// CHECK14-SAME: () #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*))
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 5)
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..5
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIcLi5EEiv_l57
// CHECK14-SAME: () #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*))
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..6
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 23)
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..7
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l52
// CHECK14-SAME: () #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*))
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..8
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 1)
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i64 [[TMP8]], i64 [[TMP10]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..9
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainI1SLi1EEiv_l57
// CHECK14-SAME: () #[[ATTR3]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..10 to void (i32*, i32*, ...)*))
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..10
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 8
// CHECK14-NEXT: [[EXN_SLOT:%.*]] = alloca i8*, align 8
// CHECK14-NEXT: [[EHSELECTOR_SLOT:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK14-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: invoke void @_ZN1SC1El(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]], i64 23)
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: [[CALL:%.*]] = invoke i8 @_ZN1ScvcEv(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]])
// CHECK14-NEXT: to label [[INVOKE_CONT2:%.*]] unwind label [[LPAD:%.*]]
// CHECK14: invoke.cont2:
// CHECK14-NEXT: [[TMP7:%.*]] = sext i8 [[CALL]] to i32
// CHECK14-NEXT: call void @__kmpc_push_num_threads(%struct.ident_t* @[[GLOB3]], i32 [[TMP1]], i32 [[TMP7]])
// CHECK14-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4
// CHECK14-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4
// CHECK14-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK14-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64)* @.omp_outlined..11 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]])
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: lpad:
// CHECK14-NEXT: [[TMP14:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP15:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 0
// CHECK14-NEXT: store i8* [[TMP15]], i8** [[EXN_SLOT]], align 8
// CHECK14-NEXT: [[TMP16:%.*]] = extractvalue { i8*, i32 } [[TMP14]], 1
// CHECK14-NEXT: store i32 [[TMP16]], i32* [[EHSELECTOR_SLOT]], align 4
// CHECK14-NEXT: call void @_ZN1SD1Ev(%struct.S* nonnull align 8 dereferenceable(24) [[REF_TMP]]) #[[ATTR6]]
// CHECK14-NEXT: br label [[TERMINATE_HANDLER:%.*]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP17:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP18:%.*]] = extractvalue { i8*, i32 } [[TMP17]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP18]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
// CHECK14: terminate.handler:
// CHECK14-NEXT: [[EXN:%.*]] = load i8*, i8** [[EXN_SLOT]], align 8
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[EXN]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_outlined..11
// CHECK14-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i64 [[DOTPREVIOUS_LB_:%.*]], i64 [[DOTPREVIOUS_UB_:%.*]]) #[[ATTR3]] personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK14-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK14-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK14-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 99, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP0:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK14-NEXT: [[CONV:%.*]] = trunc i64 [[TMP0]] to i32
// CHECK14-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK14-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK14-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK14-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK14-NEXT: [[TMP2:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK14-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP2]], align 4
// CHECK14-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP3]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK14-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 99
// CHECK14-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK14: cond.true:
// CHECK14-NEXT: br label [[COND_END:%.*]]
// CHECK14: cond.false:
// CHECK14-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: br label [[COND_END]]
// CHECK14: cond.end:
// CHECK14-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK14-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK14-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK14: omp.inner.for.cond:
// CHECK14-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK14-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK14-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK14: omp.inner.for.body:
// CHECK14-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK14-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK14-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK14-NEXT: invoke void @_Z3foov()
// CHECK14-NEXT: to label [[INVOKE_CONT:%.*]] unwind label [[TERMINATE_LPAD:%.*]]
// CHECK14: invoke.cont:
// CHECK14-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK14: omp.body.continue:
// CHECK14-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK14: omp.inner.for.inc:
// CHECK14-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK14-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK14-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK14: omp.inner.for.end:
// CHECK14-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK14: omp.loop.exit:
// CHECK14-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP3]])
// CHECK14-NEXT: ret void
// CHECK14: terminate.lpad:
// CHECK14-NEXT: [[TMP11:%.*]] = landingpad { i8*, i32 }
// CHECK14-NEXT: catch i8* null
// CHECK14-NEXT: [[TMP12:%.*]] = extractvalue { i8*, i32 } [[TMP11]], 0
// CHECK14-NEXT: call void @__clang_call_terminate(i8* [[TMP12]]) #[[ATTR10]]
// CHECK14-NEXT: unreachable
//
//
// CHECK14-LABEL: define {{[^@]+}}@_ZN1SD2Ev
// CHECK14-SAME: (%struct.S* nonnull align 8 dereferenceable(24) [[THIS:%.*]]) unnamed_addr #[[ATTR8]] comdat align 2 {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK14-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK14-NEXT: ret void
//
//
// CHECK14-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK14-SAME: () #[[ATTR9:[0-9]+]] {
// CHECK14-NEXT: entry:
// CHECK14-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK14-NEXT: ret void
//
//