mirrors/llvm-project
0
0
Fork 0
mirror of https://github.com/llvm/llvm-project.git synced 2025-05-01 09:56:07 +00:00
Code Issues Projects Releases Wiki Activity
llvm-project/clang/test/OpenMP/distribute_simd_reduction_codegen.cpp
Johannes Doerfert c5488c8dcc [OpenMP] Properly set static thread limit (w/o analysis) 
We used to have two separate implementations to derive the number of
threads used in a target region. This lead us to sometimes miss out on
user provided thread bounds (num_threads, or thread_limit) when we
looked for "constant default values". If we might miss out on the
presence of those bounds, we cannot set the thread_limit statically
since the runtime will try to honor user input rather than cap it at the
"preferred default". This patch replaces the secondary implementation
with the primary in a mode that will not emit code but just look for the
presence, and potentially upper bounds, of thread limiting clauses.

The runtime test would not pass without this rewrite as we missed some
clauses, set the static limit on the device to the preferred value, but
then violated that value at runtime.

Fixes: https://github.com/llvm/llvm-project/issues/64845

Differential Revision: https://reviews.llvm.org/D158381
2023-08-23 11:12:03 -07:00

1010 lines
66 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 -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 -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 -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 -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 -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 -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 -DCHECK -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5
// RUN: %clang_cc1 -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 -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 --check-prefix=CHECK5
// RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7
// RUN: %clang_cc1 -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 -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 --check-prefix=CHECK7
// RUN: %clang_cc1 -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 -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 -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 -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 --check-prefix=CHECK11
// RUN: %clang_cc1 -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 -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 --check-prefix=CHECK11
// 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
#pragma omp distribute simd 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
#pragma omp distribute simd 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
#pragma omp distribute simd 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 ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[SIVAR_CASTED]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[SIVAR_CASTED]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP1]], ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP1]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 2, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP18]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT: store i32 0, ptr [[TMP19]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK1-NEXT: br i1 [[TMP21]], 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_l64(i64 [[TMP1]]) #[[ATTR3:[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_l64
// CHECK1-SAME: (i64 noundef [[SIVAR:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[SIVAR_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[SIVAR]], ptr [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.omp_outlined, ptr [[SIVAR_ADDR]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[SIVAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[SIVAR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[SIVAR]], ptr [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[SIVAR_ADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: store i32 0, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[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, ptr [[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]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR4:[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 ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[T_VAR_CASTED]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP1]], ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP1]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 2, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP18]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT: store i32 0, ptr [[TMP19]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK1-NEXT: br i1 [[TMP21]], 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 [[TMP1]]) #[[ATTR3]]
// 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]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.omp_outlined, ptr [[T_VAR_ADDR]])
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[T_VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[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, ptr [[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]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]]
// CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1: omp.loop.exit:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR6:[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 ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK3-NEXT: store i32 [[TMP0]], ptr [[SIVAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[SIVAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 2, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT: store i32 0, ptr [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK3-NEXT: br i1 [[TMP21]], 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_l64(i32 [[TMP1]]) #[[ATTR3:[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_l64
// CHECK3-SAME: (i32 noundef [[SIVAR:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[SIVAR_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[SIVAR]], ptr [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.omp_outlined, ptr [[SIVAR_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l64.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[SIVAR_ADDR:%.*]] = alloca ptr, 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: [[SIVAR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[SIVAR]], ptr [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[SIVAR_ADDR]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: store i32 0, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[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, ptr [[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]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK3-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR4:[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 ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK3-NEXT: store i32 [[TMP0]], ptr [[T_VAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[T_VAR_CASTED]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 2, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP17]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP18]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT: store i32 0, ptr [[TMP19]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK3-NEXT: br i1 [[TMP21]], 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 [[TMP1]]) #[[ATTR3]]
// 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]], ptr [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.omp_outlined, ptr [[T_VAR_ADDR]])
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[T_VAR_ADDR:%.*]] = alloca ptr, 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: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[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, ptr [[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]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3: omp.inner.for.cond:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]]
// CHECK3: omp.inner.for.end:
// CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK3: omp.loop.exit:
// CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK3-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK3: .omp.final.then:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void
//
//
// CHECK5-LABEL: define {{[^@]+}}@main
// CHECK5-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[SIVAR:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: store i32 0, ptr [[SIVAR]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK5-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK5-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK5-NEXT: ret i32 [[CALL]]
//
//
// CHECK5-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK5-SAME: () #[[ATTR1:[0-9]+]] comdat {
// CHECK5-NEXT: entry:
// CHECK5-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK5-NEXT: ret i32 0
//
//
// CHECK7-LABEL: define {{[^@]+}}@main
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK7-NEXT: entry:
// CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[SIVAR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[SIVAR]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK7-NEXT: ret i32 [[CALL]]
//
//
// CHECK7-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK7-SAME: () #[[ATTR1:[0-9]+]] comdat {
// CHECK7-NEXT: entry:
// CHECK7-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK7-NEXT: ret i32 0
//
//
// 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, ptr [[RETVAL]], align 4
// CHECK9-NEXT: call void @"_ZZ4mainENK3$_0clEv"(ptr 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]], ptr [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2:[0-9]+]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l45.omp_outlined, ptr [[SIVAR_ADDR]])
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l45.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[SIVAR_ADDR:%.*]] = alloca ptr, 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: [[SIVAR1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[SIVAR]], ptr [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[SIVAR_ADDR]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: store i32 0, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[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, ptr [[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]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9: omp.inner.for.cond:
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK9-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(8) [[REF_TMP]]), !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK9: omp.inner.for.end:
// CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK9: omp.loop.exit:
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 2, ptr [[I]], align 4
// CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK9: .omp.final.done:
// CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK9-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK9-NEXT: ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK9-SAME: () #[[ATTR5:[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: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK11-NEXT: call void @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK11-NEXT: ret i32 0
//
Reference in New Issue View Git Blame Copy Permalink