caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
4794bbffb228894de3ca0dad50fc027eb94f743e
clang-p2996/clang/test/OpenMP/teams_distribute_reduction_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

965 lines
65 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 -DCHECK -verify -fopenmp -x c++ -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 -DCHECK -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 -DCHECK -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 -DCHECK -verify -fopenmp -x c++ -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 -DCHECK -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 -DCHECK -fopenmp -x c++ -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 -DCHECK -verify -fopenmp-simd -x c++ -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 -DCHECK -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 -DCHECK -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 -DCHECK -verify -fopenmp-simd -x c++ -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 -DCHECK -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 -DCHECK -fopenmp-simd -x c++ -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 -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=CHECK9
// 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++ -std=c++11 -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 -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++ -std=c++11 -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}}"
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
template <typename T>
T tmain() {
T t_var = T();
T vec[] = {1, 2};
#pragma omp target
#pragma omp teams distribute reduction(+: t_var)
for (int i = 0; i < 2; ++i) {
t_var += (T) i;
}
return T();
}
int main() {
static int sivar;
#ifdef LAMBDA
[&]() {
#pragma omp target
#pragma omp teams distribute reduction(+: sivar)
for (int i = 0; i < 2; ++i) {
// Skip global and bound tid vars
sivar += i;
[&]() {
sivar += 4;
}();
}
}();
return 0;
#else
#pragma omp target
#pragma omp teams distribute reduction(+: sivar)
for (int i = 0; i < 2; ++i) {
sivar += i;
}
return tmain<int>();
#endif
}
// Skip global and bound tid vars
// Skip global and bound tid vars
#endif
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SIVAR_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[SIVAR_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, i64* [[SIVAR_CASTED]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64*
// CHECK1-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP9]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP7]], i8*** [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, i64* [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63(i64 [[TMP1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63
// CHECK1-SAME: (i64 noundef [[SIVAR:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[SIVAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[SIVAR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[SIVAR]], i32** [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[SIVAR1]], align 4
// 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: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], i32* [[SIVAR1]], align 4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 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 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i32* [[SIVAR1]] to i8*
// CHECK1-NEXT: store i8* [[TMP13]], i8** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, i8* [[TMP14]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD5]], i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT: [[T_VAR_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i64 8, i1 false)
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[T_VAR_CASTED]] to i32*
// CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[T_VAR_CASTED]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP4]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i64*
// CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP6]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store i8* null, i8** [[TMP7]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, i32* [[TMP11]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store i8** [[TMP9]], i8*** [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.3, i32 0, i32 0), i64** [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i64** [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store i8** null, i8*** [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store i8** null, i8*** [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, i64* [[TMP18]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0
// CHECK1-NEXT: br i1 [[TMP20]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK1: omp_offload.failed:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(i64 [[TMP2]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
// CHECK1-SAME: (i64 noundef [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[T_VAR]], i64* [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[T_VAR_ADDR]] to i32*
// CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[CONV]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK1-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: store i32 0, i32* [[T_VAR1]], align 4
// 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: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], i32* [[T_VAR1]], align 4
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 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 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: [[TMP13:%.*]] = bitcast i32* [[T_VAR1]] to i8*
// CHECK1-NEXT: store i8* [[TMP13]], i8** [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], i32 1, i64 8, i8* [[TMP14]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD5]], i32* [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR7:[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: [[SIVAR_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* @_ZZ4mainE5sivar, align 4
// CHECK3-NEXT: store i32 [[TMP0]], i32* [[SIVAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[SIVAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32*
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP6]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP9]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP7]], i8*** [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i64** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, i64* [[TMP17]], align 8
// CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63(i32 [[TMP1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l63
// CHECK3-SAME: (i32 noundef [[SIVAR:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[SIVAR]], i32* [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[SIVAR_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[SIVAR_ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[SIVAR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, 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: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], 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 i32* [[SIVAR]], i32** [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[SIVAR1]], 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: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], i32* [[SIVAR1]], align 4
// 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: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 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 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i32* [[SIVAR1]] to i8*
// CHECK3-NEXT: store i8* [[TMP13]], i8** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, i8* [[TMP14]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], i32* [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT: [[T_VAR_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, i32* [[T_VAR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = bitcast [2 x i32]* [[VEC]] to i8*
// CHECK3-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP0]], i8* align 4 bitcast ([2 x i32]* @__const._Z5tmainIiET_v.vec to i8*), i32 8, i1 false)
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[T_VAR]], align 4
// CHECK3-NEXT: store i32 [[TMP1]], i32* [[T_VAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[T_VAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP4:%.*]] = bitcast i8** [[TMP3]] to i32*
// CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP4]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = bitcast i8** [[TMP5]] to i32*
// CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP6]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store i8* null, i8** [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, i32* [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, i32* [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store i8** [[TMP8]], i8*** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store i8** [[TMP9]], i8*** [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes.3, i32 0, i32 0), i64** [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes.4, i32 0, i32 0), i64** [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store i8** null, i8*** [[TMP16]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store i8** null, i8*** [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, i64* [[TMP18]], align 8
// CHECK3-NEXT: [[TMP19:%.*]] = call i32 @__tgt_target_kernel(%struct.ident_t* @[[GLOB3]], i64 -1, i32 0, i32 0, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, %struct.__tgt_kernel_arguments* [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0
// CHECK3-NEXT: br i1 [[TMP20]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK3: omp_offload.failed:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(i32 [[TMP2]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret i32 0
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
// CHECK3-SAME: (i32 noundef [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[T_VAR]], i32* [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[T_VAR_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[T_VAR_ADDR:%.*]] = alloca i32*, align 4
// CHECK3-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, 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: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], 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 i32* [[T_VAR]], i32** [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: store i32 0, i32* [[T_VAR1]], 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: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], i32* [[T_VAR1]], align 4
// 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: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 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 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = bitcast i32* [[T_VAR1]] to i8*
// CHECK3-NEXT: store i8* [[TMP13]], i8** [[TMP12]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], i32 1, i32 4, i8* [[TMP14]], void (i8*, i8*)* @.omp.reduction.reduction_func.2, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], i32* [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK3-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 4
// CHECK3-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 4
// CHECK3-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK3-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR7:[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: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK9-NEXT: store i32 0, i32* [[RETVAL]], align 4
// CHECK9-NEXT: call void @"_ZZ4mainENK3$_0clEv"(%class.anon* noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK9-NEXT: ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l45
// CHECK9-SAME: (i64 noundef [[SIVAR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i64 [[SIVAR]], i64* [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[SIVAR_ADDR]] to i32*
// CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_outlined.
// CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[SIVAR_ADDR:%.*]] = alloca i32*, align 8
// CHECK9-NEXT: [[SIVAR1:%.*]] = alloca i32, align 4
// 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: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK9-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
// 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* [[SIVAR]], i32** [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: store i32 0, i32* [[SIVAR1]], align 4
// 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: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 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: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[I]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK9-NEXT: store i32 [[ADD3]], i32* [[SIVAR1]], align 4
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], %class.anon.0* [[REF_TMP]], i32 0, i32 0
// CHECK9-NEXT: store i32* [[SIVAR1]], i32** [[TMP11]], align 8
// CHECK9-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(%class.anon.0* noundef nonnull align 8 dereferenceable(8) [[REF_TMP]])
// 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: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT: store i32 [[ADD4]], i32* [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK9-NEXT: [[TMP14:%.*]] = bitcast i32* [[SIVAR1]] to i8*
// CHECK9-NEXT: store i8* [[TMP14]], i8** [[TMP13]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, i8* [[TMP15]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: switch i32 [[TMP16]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK9-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK9-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK9-NEXT: ]
// CHECK9: .omp.reduction.case1:
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[TMP0]], align 4
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: store i32 [[ADD5]], i32* [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.case2:
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[SIVAR1]], align 4
// CHECK9-NEXT: [[TMP20:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP19]] monotonic, align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]], [8 x i32]* @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.default:
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK9-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8
// CHECK9-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8
// CHECK9-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [1 x i8*]*
// CHECK9-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [1 x i8*]*
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP5]], i64 0, i64 0
// CHECK9-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32*
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[TMP3]], i64 0, i64 0
// CHECK9-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32*
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[TMP11]], align 4
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[TMP8]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4
// 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
//
Reference in New Issue View Git Blame Copy Permalink