caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
4794bbffb228894de3ca0dad50fc027eb94f743e
clang-p2996/clang/test/OpenMP/distribute_private_codegen.cpp
Joseph Huber 5300263c70 [OpenMP] Add loop tripcount argument to kernel launch and remove push function 
Previously we added the `push_target_tripcount` function to send the
loop tripcount to the device runtime so we knew how to configure the
teams / threads for execute the loop for a teams distribute construct.
This was implemented as a separate function mostly to avoid changing the
interface for backwards compatbility. Now that we've changed it anyway
and the new interface can take an arbitrary number of arguments via the
struct without changing the ABI, we can move this to the new interface.
This will simplify the runtime by removing unnecessary state between
calls.

Depends on D128550

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D128816
2022-07-08 14:44:16 -04:00

1524 lines
103 KiB
C++
Raw Blame History

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -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
template <class T>
struct S {
T f;
S(T a) : f(a) {}
S() : f() {}
operator T() { return T(); }
~S() {}
};
template <typename T>
T tmain() {
S<T> test;
T t_var = T();
T vec[] = {1, 2};
S<T> s_arr[] = {1, 2};
S<T> &var = test;
#pragma omp target
#pragma omp teams
#pragma omp distribute private(t_var, vec, s_arr, s_arr, var, var)
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
}
return T();
}
int main() {
static int svar;
volatile double g;
volatile double &g1 = g;
#ifdef LAMBDA
[&]() {
static float sfvar;
#pragma omp target
#pragma omp teams
#pragma omp distribute private(g, g1, svar, sfvar)
for (int i = 0; i < 2; ++i) {
g = 1;
g1 = 1;
svar = 3;
sfvar = 4.0;
[&]() {
g = 2;
g1 = 2;
svar = 4;
sfvar = 8.0;
}();
}
}();
return 0;
#else
S<float> test;
int t_var = 0;
int vec[] = {1, 2};
S<float> s_arr[] = {1, 2};
S<float> &var = test;
#pragma omp target
#pragma omp teams
#pragma omp distribute private(t_var, vec, s_arr, s_arr, var, var, svar)
for (int i = 0; i < 2; ++i) {
vec[i] = t_var;
s_arr[i] = var;
}
int i;
#pragma omp target
#pragma omp teams
#pragma omp distribute private(i)
for (i = 0; i < 2; ++i) {
;
}
return tmain<int>();
#endif
}
#endif
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[G:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[G1:%.*]] = alloca double*, align 8
// CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT: store double* [[G]], double** [[G1]], align 8
// CHECK1-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK1-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2:[0-9]+]], 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 noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
// 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: [[_TMP1:%.*]] = alloca double*, align 8
// 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: [[G:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[G1:%.*]] = alloca double, align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca double*, align 8
// CHECK1-NEXT: [[SVAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SFVAR:%.*]] = alloca float, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store double* undef, double** [[_TMP1]], align 8
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: store double* [[G1]], double** [[_TMP2]], align 8
// 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_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false:
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_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_UB]], align 4
// CHECK1-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: store double 1.000000e+00, double* [[G]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load double*, double** [[_TMP2]], align 8
// CHECK1-NEXT: store volatile double 1.000000e+00, double* [[TMP8]], align 8
// CHECK1-NEXT: store i32 3, i32* [[SVAR]], align 4
// CHECK1-NEXT: store float 4.000000e+00, float* [[SFVAR]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK1-NEXT: store double* [[G]], double** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
// CHECK1-NEXT: [[TMP11:%.*]] = load double*, double** [[_TMP2]], align 8
// CHECK1-NEXT: store double* [[TMP11]], double** [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
// CHECK1-NEXT: store i32* [[SVAR]], i32** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3
// CHECK1-NEXT: store float* [[SFVAR]], float** [[TMP13]], align 8
// CHECK1-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(32) [[REF_TMP]])
// 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: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK1-NEXT: store i32 [[ADD4]], 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_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[G:%.*]] = alloca double, align 8
// CHECK3-NEXT: [[G1:%.*]] = alloca double*, align 4
// CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK3-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT: store double* [[G]], double** [[G1]], align 4
// CHECK3-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK3-NEXT: ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l68
// CHECK3-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP1:%.*]] = alloca double*, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[G:%.*]] = alloca double, align 8
// CHECK3-NEXT: [[G1:%.*]] = alloca double, align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca double*, align 4
// CHECK3-NEXT: [[SVAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[SFVAR:%.*]] = alloca float, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 4
// CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store double* undef, double** [[_TMP1]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: store double* [[G1]], double** [[_TMP2]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK3: cond.true:
// CHECK3-NEXT: br label [[COND_END:%.*]]
// CHECK3: cond.false:
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK3-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK3-NEXT: store double 1.000000e+00, double* [[G]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load double*, double** [[_TMP2]], align 4
// CHECK3-NEXT: store volatile double 1.000000e+00, double* [[TMP8]], align 4
// CHECK3-NEXT: store i32 3, i32* [[SVAR]], align 4
// CHECK3-NEXT: store float 4.000000e+00, float* [[SFVAR]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK3-NEXT: store double* [[G]], double** [[TMP9]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 1
// CHECK3-NEXT: [[TMP11:%.*]] = load double*, double** [[_TMP2]], align 4
// CHECK3-NEXT: store double* [[TMP11]], double** [[TMP10]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 2
// CHECK3-NEXT: store i32* [[SVAR]], i32** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 3
// CHECK3-NEXT: store float* [[SFVAR]], float** [[TMP13]], align 4
// CHECK3-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 4 dereferenceable(16) [[REF_TMP]])
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK3: omp.body.continue:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3: omp.inner.for.inc:
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK3-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[G:%.*]] = alloca double, align 8
// CHECK9-NEXT: [[G1:%.*]] = alloca double*, align 8
// CHECK9-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK9-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK9-NEXT: [[VAR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: store double* [[G]], double** [[G1]], align 8
// CHECK9-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i64 8, i1 false)
// CHECK9-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 0
// CHECK9-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float noundef 1.000000e+00)
// CHECK9-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i64 1
// CHECK9-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float noundef 2.000000e+00)
// CHECK9-NEXT: store %struct.S* [[TEST]], %struct.S** [[VAR]], align 8
// CHECK9-NEXT: store %struct.S* undef, %struct.S** [[_TMP1]], align 8
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, i32* [[TMP1]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 0, i32* [[TMP2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store i8** null, i8*** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store i8** null, i8*** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store i64* null, i64** [[TMP5]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store i64* null, i64** [[TMP6]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store i8** null, i8*** [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store i8** null, i8*** [[TMP8]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 2, i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l93.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK9-NEXT: br i1 [[TMP11]], 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_l93() #[[ATTR4:[0-9]+]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont:
// CHECK9-NEXT: [[KERNEL_ARGS3:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, i32* [[TMP12]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 1
// CHECK9-NEXT: store i32 0, i32* [[TMP13]], align 4
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 2
// CHECK9-NEXT: store i8** null, i8*** [[TMP14]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 3
// CHECK9-NEXT: store i8** null, i8*** [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 4
// CHECK9-NEXT: store i64* null, i64** [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 5
// CHECK9-NEXT: store i64* null, i64** [[TMP17]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 6
// CHECK9-NEXT: store i8** null, i8*** [[TMP18]], align 8
// CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 7
// CHECK9-NEXT: store i8** null, i8*** [[TMP19]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 8
// CHECK9-NEXT: store i64 2, i64* [[TMP20]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]])
// CHECK9-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK9-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED4:%.*]], label [[OMP_OFFLOAD_CONT5:%.*]]
// CHECK9: omp_offload.failed4:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102() #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT5]]
// CHECK9: omp_offload.cont5:
// CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK9-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9: arraydestroy.body:
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP23]], [[OMP_OFFLOAD_CONT5]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE6:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9: arraydestroy.done6:
// CHECK9-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP24]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[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 @_ZN1SIfEC2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store float [[A]], float* [[A_ADDR]], align 4
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK9-NEXT: call void @_ZN1SIfEC2Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l93
// 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* @[[GLOB2]], 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 noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[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: [[_TMP1:%.*]] = alloca %struct.S*, align 8
// 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: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK9-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[SVAR:%.*]] = 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 %struct.S* undef, %struct.S** [[_TMP1]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 1, 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: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT: br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK9: arrayctor.loop:
// CHECK9-NEXT: [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK9-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK9-NEXT: [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i64 1
// CHECK9-NEXT: [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK9-NEXT: br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK9: arrayctor.cont:
// CHECK9-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK9-NEXT: store %struct.S* [[VAR]], %struct.S** [[_TMP2]], align 8
// 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_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_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_UB]], align 4
// CHECK9-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK9: omp.inner.for.cond.cleanup:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: store i32 [[TMP8]], i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load %struct.S*, %struct.S** [[_TMP2]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4
// CHECK9-NEXT: [[IDXPROM4:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i64 0, i64 [[IDXPROM4]]
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast %struct.S* [[ARRAYIDX5]] to i8*
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast %struct.S* [[TMP10]] to i8*
// CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP12]], i8* align 4 [[TMP13]], i64 4, i1 false)
// 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: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK9-NEXT: store i32 [[ADD6]], 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: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP16]])
// CHECK9-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN7]], i64 2
// CHECK9-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9: arraydestroy.body:
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK9-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9: arraydestroy.done8:
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] 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 @_ZN1SIfED2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[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_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: [[I1:%.*]] = 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_LB]], align 4
// CHECK9-NEXT: store i32 1, 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: [[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_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_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_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// 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: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], 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 [[TMP1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK9-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK9-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK9-NEXT: [[VAR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK9-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK9-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 0
// CHECK9-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 noundef signext 1)
// CHECK9-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i64 1
// CHECK9-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 noundef signext 2)
// CHECK9-NEXT: store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 8
// CHECK9-NEXT: store %struct.S.0* undef, %struct.S.0** [[_TMP1]], align 8
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, i32* [[TMP1]], align 4
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 0, i32* [[TMP2]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store i8** null, i8*** [[TMP3]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store i8** null, i8*** [[TMP4]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store i64* null, i64** [[TMP5]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store i64* null, i64** [[TMP6]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store i8** null, i8*** [[TMP7]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store i8** null, i8*** [[TMP8]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 2, i64* [[TMP9]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK9-NEXT: br i1 [[TMP11]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9: omp_offload.failed:
// CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49() #[[ATTR4]]
// CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK9: omp_offload.cont:
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9: arraydestroy.body:
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK9-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9: arraydestroy.done2:
// CHECK9-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP13]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] 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: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: store float 0.000000e+00, float* [[F]], align 4
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca float, align 4
// CHECK9-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store float [[A]], float* [[A_ADDR]], align 4
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK9-NEXT: store float [[TMP0]], float* [[F]], align 4
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK9-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] 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 {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: call void @_ZN1SIiEC2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK9-NEXT: call void @_ZN1SIiEC2Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]], i32 noundef signext [[TMP0]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49
// CHECK9-SAME: () #[[ATTR3]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[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: [[_TMP1:%.*]] = alloca %struct.S.0*, align 8
// 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: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK9-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK9-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK9-NEXT: [[_TMP2:%.*]] = alloca %struct.S.0*, align 8
// 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 %struct.S.0* undef, %struct.S.0** [[_TMP1]], align 8
// CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 1, 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: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i64 2
// CHECK9-NEXT: br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK9: arrayctor.loop:
// CHECK9-NEXT: [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK9-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK9-NEXT: [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i64 1
// CHECK9-NEXT: [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK9-NEXT: br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK9: arrayctor.cont:
// CHECK9-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK9-NEXT: store %struct.S.0* [[VAR]], %struct.S.0** [[_TMP2]], align 8
// 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_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// 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_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_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_UB]], align 4
// CHECK9-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK9-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK9: omp.inner.for.cond.cleanup:
// CHECK9-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: store i32 [[TMP8]], i32* [[ARRAYIDX]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP2]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4
// CHECK9-NEXT: [[IDXPROM4:%.*]] = sext i32 [[TMP11]] to i64
// CHECK9-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i64 0, i64 [[IDXPROM4]]
// CHECK9-NEXT: [[TMP12:%.*]] = bitcast %struct.S.0* [[ARRAYIDX5]] to i8*
// CHECK9-NEXT: [[TMP13:%.*]] = bitcast %struct.S.0* [[TMP10]] to i8*
// CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP12]], i8* align 4 [[TMP13]], i64 4, i1 false)
// 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: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK9-NEXT: store i32 [[ADD6]], 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: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP16]])
// CHECK9-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAY_BEGIN7:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN7]], i64 2
// CHECK9-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK9: arraydestroy.body:
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK9-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK9-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK9-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN7]]
// CHECK9-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK9: arraydestroy.done8:
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: call void @_ZN1SIiED2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: store i32 0, i32* [[F]], align 4
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef signext [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP0]], i32* [[F]], align 4
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK9-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 8
// CHECK9-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 8
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK9-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK9-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@main
// CHECK11-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[G:%.*]] = alloca double, align 8
// CHECK11-NEXT: [[G1:%.*]] = alloca double*, align 4
// CHECK11-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK11-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK11-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK11-NEXT: [[VAR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: store double* [[G]], double** [[G1]], align 4
// CHECK11-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK11-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const.main.vec to i8*), i32 8, i1 false)
// CHECK11-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], float noundef 1.000000e+00)
// CHECK11-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYINIT_BEGIN]], i32 1
// CHECK11-NEXT: call void @_ZN1SIfEC1Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float noundef 2.000000e+00)
// CHECK11-NEXT: store %struct.S* [[TEST]], %struct.S** [[VAR]], align 4
// CHECK11-NEXT: store %struct.S* undef, %struct.S** [[_TMP1]], align 4
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, i32* [[TMP1]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 0, i32* [[TMP2]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store i8** null, i8*** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store i8** null, i8*** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store i64* null, i64** [[TMP5]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store i64* null, i64** [[TMP6]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store i8** null, i8*** [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store i8** null, i8*** [[TMP8]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 2, i64* [[TMP9]], align 8
// CHECK11-NEXT: [[TMP10:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l93.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK11-NEXT: br i1 [[TMP11]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l93() #[[ATTR4:[0-9]+]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont:
// CHECK11-NEXT: [[KERNEL_ARGS3:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, i32* [[TMP12]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 1
// CHECK11-NEXT: store i32 0, i32* [[TMP13]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 2
// CHECK11-NEXT: store i8** null, i8*** [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 3
// CHECK11-NEXT: store i8** null, i8*** [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 4
// CHECK11-NEXT: store i64* null, i64** [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 5
// CHECK11-NEXT: store i64* null, i64** [[TMP17]], align 4
// CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 6
// CHECK11-NEXT: store i8** null, i8*** [[TMP18]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 7
// CHECK11-NEXT: store i8** null, i8*** [[TMP19]], align 4
// CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]], i32 0, i32 8
// CHECK11-NEXT: store i64 2, i64* [[TMP20]], align 8
// CHECK11-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS3]])
// CHECK11-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK11-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED4:%.*]], label [[OMP_OFFLOAD_CONT5:%.*]]
// CHECK11: omp_offload.failed4:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102() #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT5]]
// CHECK11: omp_offload.cont5:
// CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK11-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
// CHECK11-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK11: arraydestroy.body:
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP23]], [[OMP_OFFLOAD_CONT5]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK11-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK11-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE6:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK11: arraydestroy.done6:
// CHECK11-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP24]]
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIfEC2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca float, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: store float [[A]], float* [[A_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIfEC2Ef(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l93
// CHECK11-SAME: () #[[ATTR3:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK11-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S], align 4
// CHECK11-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK11-NEXT: [[_TMP2:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[SVAR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT: store %struct.S* undef, %struct.S** [[_TMP1]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN]], i32 2
// CHECK11-NEXT: br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK11: arrayctor.loop:
// CHECK11-NEXT: [[ARRAYCTOR_CUR:%.*]] = phi %struct.S* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK11-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK11-NEXT: [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYCTOR_CUR]], i32 1
// CHECK11-NEXT: [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK11-NEXT: br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK11: arrayctor.cont:
// CHECK11-NEXT: call void @_ZN1SIfEC1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK11-NEXT: store %struct.S* [[VAR]], %struct.S** [[_TMP2]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11: cond.true:
// CHECK11-NEXT: br label [[COND_END:%.*]]
// CHECK11: cond.false:
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK11-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK11: omp.inner.for.cond.cleanup:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i32 0, i32 [[TMP9]]
// CHECK11-NEXT: store i32 [[TMP8]], i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = load %struct.S*, %struct.S** [[_TMP2]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4
// CHECK11-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast %struct.S* [[ARRAYIDX4]] to i8*
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast %struct.S* [[TMP10]] to i8*
// CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP12]], i8* align 4 [[TMP13]], i32 4, i1 false)
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK11-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit:
// CHECK11-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP16]])
// CHECK11-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAY_BEGIN6:%.*]] = getelementptr inbounds [2 x %struct.S], [2 x %struct.S]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAY_BEGIN6]], i32 2
// CHECK11-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK11: arraydestroy.body:
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S* [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], %struct.S* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK11-NEXT: call void @_ZN1SIfED1Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN6]]
// CHECK11-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE7:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK11: arraydestroy.done7:
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIfED2Ev(%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l102
// CHECK11-SAME: () #[[ATTR3]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[I1:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11: cond.true:
// CHECK11-NEXT: br label [[COND_END:%.*]]
// CHECK11: cond.false:
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK11-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit:
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK11-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK11-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK11-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK11-NEXT: [[VAR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK11-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i32 8, i1 false)
// CHECK11-NEXT: [[ARRAYINIT_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_BEGIN]], i32 noundef 1)
// CHECK11-NEXT: [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYINIT_BEGIN]], i32 1
// CHECK11-NEXT: call void @_ZN1SIiEC1Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 noundef 2)
// CHECK11-NEXT: store %struct.S.0* [[TEST]], %struct.S.0** [[VAR]], align 4
// CHECK11-NEXT: store %struct.S.0* undef, %struct.S.0** [[_TMP1]], align 4
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, i32* [[TMP1]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 0, i32* [[TMP2]], align 4
// CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store i8** null, i8*** [[TMP3]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store i8** null, i8*** [[TMP4]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store i64* null, i64** [[TMP5]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store i64* null, i64** [[TMP6]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store i8** null, i8*** [[TMP7]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store i8** null, i8*** [[TMP8]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 2, i64* [[TMP9]], align 8
// CHECK11-NEXT: [[TMP10:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB2]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP11:%.*]] = icmp ne i32 [[TMP10]], 0
// CHECK11-NEXT: br i1 [[TMP11]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11: omp_offload.failed:
// CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49() #[[ATTR4]]
// CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK11: omp_offload.cont:
// CHECK11-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK11-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
// CHECK11-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK11: arraydestroy.body:
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP12]], [[OMP_OFFLOAD_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK11-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN]]
// CHECK11-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK11: arraydestroy.done2:
// CHECK11-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR4]]
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP13]]
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT: store float 0.000000e+00, float* [[F]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca float, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: store float [[A]], float* [[A_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], %struct.S* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP0:%.*]] = load float, float* [[A_ADDR]], align 4
// CHECK11-NEXT: store float [[TMP0]], float* [[F]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK11-SAME: (%struct.S* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S*, align 4
// CHECK11-NEXT: store %struct.S* [[THIS]], %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S*, %struct.S** [[THIS_ADDR]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIiEC2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIiEC2Ei(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]], i32 noundef [[TMP0]])
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l49
// CHECK11-SAME: () #[[ATTR3]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..2
// CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[_TMP1:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK11-NEXT: [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK11-NEXT: [[VAR:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK11-NEXT: [[_TMP2:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT: store %struct.S.0* undef, %struct.S.0** [[_TMP1]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN]], i32 2
// CHECK11-NEXT: br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK11: arrayctor.loop:
// CHECK11-NEXT: [[ARRAYCTOR_CUR:%.*]] = phi %struct.S.0* [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK11-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK11-NEXT: [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYCTOR_CUR]], i32 1
// CHECK11-NEXT: [[ARRAYCTOR_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK11-NEXT: br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK11: arrayctor.cont:
// CHECK11-NEXT: call void @_ZN1SIiEC1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[VAR]])
// CHECK11-NEXT: store %struct.S.0* [[VAR]], %struct.S.0** [[_TMP2]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 1
// CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11: cond.true:
// CHECK11-NEXT: br label [[COND_END:%.*]]
// CHECK11: cond.false:
// CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: br label [[COND_END]]
// CHECK11: cond.end:
// CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP4]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11: omp.inner.for.cond:
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK11-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK11: omp.inner.for.cond.cleanup:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK11: omp.inner.for.body:
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 1
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], [2 x i32]* [[VEC]], i32 0, i32 [[TMP9]]
// CHECK11-NEXT: store i32 [[TMP8]], i32* [[ARRAYIDX]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = load %struct.S.0*, %struct.S.0** [[_TMP2]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4
// CHECK11-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT: [[TMP12:%.*]] = bitcast %struct.S.0* [[ARRAYIDX4]] to i8*
// CHECK11-NEXT: [[TMP13:%.*]] = bitcast %struct.S.0* [[TMP10]] to i8*
// CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP12]], i8* align 4 [[TMP13]], i32 4, i1 false)
// CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11: omp.body.continue:
// CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11: omp.inner.for.inc:
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK11-NEXT: store i32 [[ADD5]], i32* [[DOTOMP_IV]], align 4
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK11: omp.inner.for.end:
// CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11: omp.loop.exit:
// CHECK11-NEXT: [[TMP15:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP15]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP16]])
// CHECK11-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[VAR]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAY_BEGIN6:%.*]] = getelementptr inbounds [2 x %struct.S.0], [2 x %struct.S.0]* [[S_ARR]], i32 0, i32 0
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAY_BEGIN6]], i32 2
// CHECK11-NEXT: br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK11: arraydestroy.body:
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi %struct.S.0* [ [[TMP17]], [[OMP_LOOP_EXIT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK11-NEXT: [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], %struct.S.0* [[ARRAYDESTROY_ELEMENTPAST]], i32 -1
// CHECK11-NEXT: call void @_ZN1SIiED1Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR4]]
// CHECK11-NEXT: [[ARRAYDESTROY_DONE:%.*]] = icmp eq %struct.S.0* [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN6]]
// CHECK11-NEXT: br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE7:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK11: arraydestroy.done7:
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: call void @_ZN1SIiED2Ev(%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR4]]
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT: store i32 0, i32* [[F]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: store i32 [[A]], i32* [[A_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[F:%.*]] = getelementptr inbounds [[STRUCT_S_0:%.*]], %struct.S.0* [[THIS1]], i32 0, i32 0
// CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
// CHECK11-NEXT: store i32 [[TMP0]], i32* [[F]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK11-SAME: (%struct.S.0* noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] comdat align 2 {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.S.0*, align 4
// CHECK11-NEXT: store %struct.S.0* [[THIS]], %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: [[THIS1:%.*]] = load %struct.S.0*, %struct.S.0** [[THIS_ADDR]], align 4
// CHECK11-NEXT: ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK11-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK11-NEXT: entry:
// CHECK11-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK11-NEXT: ret void
//
Reference in New Issue View Git Blame Copy Permalink