llvm/clang/test/OpenMP/target_teams_generic_loop_collapse_codegen.cpp

// 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 _
// expected-no-diagnostics
#ifndef HEADER
#define HEADER

// Test host codegen.
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3

// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -DCK1 -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 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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 -DCK1 -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 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
#ifdef CK1

template <typename T, int X, long long Y>
struct SS{
  T a[X][Y];

  int foo(void) {

    #pragma omp target teams loop collapse(2)
    for(int i = 0; i < X; i++) {
      for(int j = 0; j < Y; j++) {
        a[i][j] = (T)0;
      }
    }

    // discard loop variables not needed here


    return a[0][0];
  }
};

int teams_template_struct(void) {
  SS<int, 123, 456> V;
  return V.foo();

}
#endif // CK1

// Test host codegen.
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11

// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -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 -DCK2 -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 -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -DCK2 -verify -Wno-vla -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 -DCK2 -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 -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
#ifdef CK2

template <typename T, int n, int m>
int tmain(T argc) {
  T a[n][m];
  #pragma omp target teams loop collapse(2)
  for(int i = 0; i < n; i++) {
    for(int j = 0; j < m; j++) {
      a[i][j] = (T)0;
    }
  }
  return 0;
}

int main (int argc, char **argv) {
  int n = 100;
  int m = 2;
  int a[n][m];
  #pragma omp target teams loop collapse(2)
  for(int i = 0; i < n; i++) {
    for(int j = 0; j < m; j++) {
      a[i][j] = 0;
    }
  }
  return tmain<int, 10, 2>(argc);
}








// discard loop variables not needed here


#endif // CK2
#endif // #ifndef HEADER
// CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
// CHECK1-NEXT:    ret i32 [[CALL]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT:    store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT:    store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT:    store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT:    store i64 56088, ptr [[TMP13]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK1-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK1-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK1-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK1:       omp_offload.cont:
// CHECK1-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT:    ret i32 [[TMP20]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[J:%.*]] = 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 [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    store i32 56087, ptr [[DOTOMP_COMB_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:    [[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_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 56087
// 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_COMB_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_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
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_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, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[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(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP1:%.*]] = 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:    [[J:%.*]] = 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 i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK1-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK1-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[CONV2]], 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:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 56087
// 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:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[DIV4:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK1-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 456
// CHECK1-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK1-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK1-NEXT:    store i32 [[ADD7]], ptr [[J]], align 4
// CHECK1-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[J]], align 4
// CHECK1-NEXT:    [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK1-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK1-NEXT:    store i32 0, ptr [[ARRAYIDX9]], 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:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT:    store i32 [[ADD10]], ptr [[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(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK1-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4
// CHECK3-NEXT:    [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(224352) [[V]])
// CHECK3-NEXT:    ret i32 [[CALL]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(224352) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, 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:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK3-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT:    store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT:    store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT:    store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT:    store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT:    store i64 56088, ptr [[TMP13]], align 8
// CHECK3-NEXT:    [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK3-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK3-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK3-NEXT:    [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK3-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK3-NEXT:    [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK3-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK3-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK3:       omp_offload.cont:
// CHECK3-NEXT:    [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
// CHECK3-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK3-NEXT:    ret i32 [[TMP20]]
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTOMP_COMB_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:    [[J:%.*]] = 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 [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    store i32 56087, ptr [[DOTOMP_COMB_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:    [[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_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 56087
// 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_COMB_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_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
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_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, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK3:       omp.inner.for.inc:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[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(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[_TMP1:%.*]] = 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:    [[J:%.*]] = 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 i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK3-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK3-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP2]], 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:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK3-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 56087
// 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:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    br label [[COND_END]]
// CHECK3:       cond.end:
// CHECK3-NEXT:    [[COND:%.*]] = phi i32 [ 56087, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK3:       omp.inner.for.cond:
// CHECK3-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3:       omp.inner.for.body:
// CHECK3-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 456
// CHECK3-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK3-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK3-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[DIV3:%.*]] = sdiv i32 [[TMP12]], 456
// CHECK3-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK3-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK3-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK3-NEXT:    [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK3-NEXT:    store i32 [[ADD6]], ptr [[J]], align 4
// CHECK3-NEXT:    [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK3-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP13]]
// CHECK3-NEXT:    [[TMP14:%.*]] = load i32, ptr [[J]], align 4
// CHECK3-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK3-NEXT:    store i32 0, ptr [[ARRAYIDX7]], 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:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK3-NEXT:    store i32 [[ADD8]], ptr [[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(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK3-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[ARGV_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT:    store i32 100, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 2, ptr [[M]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT:    [[TMP2:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT:    [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK9-NEXT:    [[TMP4:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK9-NEXT:    store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT:    store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP6]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT:    store i32 [[TMP8]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT:    [[TMP11:%.*]] = mul nuw i64 [[TMP10]], 4
// CHECK9-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP7]], ptr [[TMP12]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    store i64 [[TMP7]], ptr [[TMP13]], align 8
// CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP14]], align 8
// CHECK9-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP9]], ptr [[TMP15]], align 8
// CHECK9-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT:    store i64 [[TMP9]], ptr [[TMP16]], align 8
// CHECK9-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT:    store ptr null, ptr [[TMP17]], align 8
// CHECK9-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP18]], align 8
// CHECK9-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT:    store i64 [[TMP1]], ptr [[TMP19]], align 8
// CHECK9-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT:    store ptr null, ptr [[TMP20]], align 8
// CHECK9-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT:    store i64 [[TMP3]], ptr [[TMP21]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT:    store i64 [[TMP3]], ptr [[TMP22]], align 8
// CHECK9-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT:    store ptr null, ptr [[TMP23]], align 8
// CHECK9-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP24]], align 8
// CHECK9-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[VLA]], ptr [[TMP25]], align 8
// CHECK9-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT:    store i64 [[TMP11]], ptr [[TMP26]], align 8
// CHECK9-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT:    store ptr null, ptr [[TMP27]], align 8
// CHECK9-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP31:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT:    store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP32:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT:    store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT:    [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT:    [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT:    [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK9-NEXT:    [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK9-NEXT:    [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT:    store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT:    [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
// CHECK9-NEXT:    [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP36]], align 4
// CHECK9-NEXT:    [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT:    store i32 5, ptr [[TMP37]], align 4
// CHECK9-NEXT:    [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP28]], ptr [[TMP38]], align 8
// CHECK9-NEXT:    [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP29]], ptr [[TMP39]], align 8
// CHECK9-NEXT:    [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr [[TMP30]], ptr [[TMP40]], align 8
// CHECK9-NEXT:    [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes, ptr [[TMP41]], align 8
// CHECK9-NEXT:    [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP42]], align 8
// CHECK9-NEXT:    [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT:    [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT:    store i64 [[ADD]], ptr [[TMP44]], align 8
// CHECK9-NEXT:    [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP45]], align 8
// CHECK9-NEXT:    [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP46]], align 4
// CHECK9-NEXT:    [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP47]], align 4
// CHECK9-NEXT:    [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP48]], align 4
// CHECK9-NEXT:    [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT:    [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0
// CHECK9-NEXT:    br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9:       omp_offload.failed:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i64 [[TMP7]], i64 [[TMP9]], i64 [[TMP1]], i64 [[TMP3]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK9:       omp_offload.cont:
// CHECK9-NEXT:    [[TMP51:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP51]])
// CHECK9-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT:    [[TMP52:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP52]])
// CHECK9-NEXT:    [[TMP53:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT:    ret i32 [[TMP53]]
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP5]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = load i64, ptr [[M_CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined, i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[J12:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_CASTED:%.*]] = alloca i64, 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 i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT:    [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT:    [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT:    [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT:    store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[J]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       land.lhs.true:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT:    br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i64 0, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT:    [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK9-NEXT:    br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT:    store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK9-NEXT:    br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT:    [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT:    [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP21]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT:    [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT:    [[TMP23:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP23]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT:    [[TMP24:%.*]] = load i64, ptr [[M_CASTED]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined.omp_outlined, i64 [[TMP19]], i64 [[TMP20]], i64 [[TMP22]], i64 [[TMP24]], i64 [[TMP0]], i64 [[TMP1]], ptr [[TMP2]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP26:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]]
// CHECK9-NEXT:    store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8
// 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:    [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP28]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[M_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[J12:%.*]] = alloca i32, align 4
// 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 i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT:    store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT:    [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT:    [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT:    [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT:    store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK9-NEXT:    store i32 0, ptr [[J]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9:       land.lhs.true:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT:    br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.then:
// CHECK9-NEXT:    store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[TMP10]], ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT:    store i64 [[TMP11]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT:    [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    [[CMP13:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]]
// CHECK9-NEXT:    br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9:       cond.true:
// CHECK9-NEXT:    [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT:    br label [[COND_END:%.*]]
// CHECK9:       cond.false:
// CHECK9-NEXT:    [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT:    store i64 [[TMP18]], ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT:    [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK9-NEXT:    br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK9-NEXT:    [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT:    [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT:    [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT:    [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK9-NEXT:    [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT:    [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT:    store i32 [[CONV21]], ptr [[I11]], align 4
// CHECK9-NEXT:    [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK9-NEXT:    [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT:    [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT:    [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT:    [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK9-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT:    [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK9-NEXT:    [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT:    [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT:    [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT:    [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT:    [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK9-NEXT:    [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT:    [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT:    [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT:    store i32 [[CONV35]], ptr [[J12]], align 4
// CHECK9-NEXT:    [[TMP27:%.*]] = load i32, ptr [[I11]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT:    [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP28]]
// CHECK9-NEXT:    [[TMP29:%.*]] = load i32, ptr [[J12]], align 4
// CHECK9-NEXT:    [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64
// CHECK9-NEXT:    [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX37]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT:    [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK9-NEXT:    store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8
// 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:    [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP32]])
// CHECK9-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK9:       omp.precond.end:
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[A:%.*]] = alloca [10 x [2 x i32]], align 4
// CHECK9-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT:    store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK9-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT:    store ptr @.offload_sizes.1, ptr [[TMP9]], align 8
// CHECK9-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP10]], align 8
// CHECK9-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT:    store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT:    store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT:    store i64 20, ptr [[TMP13]], align 8
// CHECK9-NEXT:    [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK9-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK9-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK9-NEXT:    [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK9-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK9-NEXT:    [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK9-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK9-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK9-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK9:       omp_offload.failed:
// CHECK9-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[A]]) #[[ATTR3]]
// CHECK9-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK9:       omp_offload.cont:
// CHECK9-NEXT:    ret i32 0
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
// CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined, ptr [[TMP0]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[J:%.*]] = alloca i32, align 4
// 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 [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    store i32 19, ptr [[DOTOMP_COMB_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:    [[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]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 19
// 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_COMB_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_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
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_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, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]])
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[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(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT:    ret void
//
//
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined.omp_outlined
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT:  entry:
// CHECK9-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// CHECK9-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT:    [[_TMP1:%.*]] = 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:    [[J:%.*]] = alloca i32, align 4
// 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 i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP1]] to i32
// CHECK9-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP2]] to i32
// CHECK9-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[CONV2]], 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:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK9-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK9-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 19
// 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:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    br label [[COND_END]]
// CHECK9:       cond.end:
// CHECK9-NEXT:    [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK9-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK9:       omp.inner.for.cond:
// CHECK9-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9:       omp.inner.for.body:
// CHECK9-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK9-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK9-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK9-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[DIV4:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK9-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[DIV4]], 2
// CHECK9-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL5]]
// CHECK9-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK9-NEXT:    [[ADD7:%.*]] = add nsw i32 0, [[MUL6]]
// CHECK9-NEXT:    store i32 [[ADD7]], ptr [[J]], align 4
// CHECK9-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK9-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64
// CHECK9-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT:    [[TMP14:%.*]] = load i32, ptr [[J]], align 4
// CHECK9-NEXT:    [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK9-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK9-NEXT:    store i32 0, ptr [[ARRAYIDX9]], align 4
// CHECK9-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK9:       omp.body.continue:
// CHECK9-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK9:       omp.inner.for.inc:
// CHECK9-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK9-NEXT:    store i32 [[ADD10]], ptr [[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(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK9-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@main
// CHECK11-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[ARGV_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK11-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4
// CHECK11-NEXT:    store i32 100, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 2, ptr [[M]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK11-NEXT:    store ptr [[TMP2]], ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// CHECK11-NEXT:    [[VLA:%.*]] = alloca i32, i32 [[TMP3]], align 4
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[__VLA_EXPR1]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP4]], ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT:    store i32 [[TMP6]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// CHECK11-NEXT:    [[TMP9:%.*]] = mul nuw i32 [[TMP8]], 4
// CHECK11-NEXT:    [[TMP10:%.*]] = sext i32 [[TMP9]] to i64
// CHECK11-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[TMP11]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[TMP12]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP13]], align 4
// CHECK11-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[TMP14]], align 4
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[TMP15]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT:    store ptr null, ptr [[TMP16]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP17]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT:    store i32 [[TMP0]], ptr [[TMP18]], align 4
// CHECK11-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr null, ptr [[TMP19]], align 4
// CHECK11-NEXT:    [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[TMP20]], align 4
// CHECK11-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[TMP21]], align 4
// CHECK11-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT:    store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP23]], align 4
// CHECK11-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[VLA]], ptr [[TMP24]], align 4
// CHECK11-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT:    store i64 [[TMP10]], ptr [[TMP25]], align 4
// CHECK11-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr null, ptr [[TMP26]], align 4
// CHECK11-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP30:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT:    store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP31:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT:    store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT:    [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT:    [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK11-NEXT:    [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK11-NEXT:    [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK11-NEXT:    [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT:    store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT:    [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
// CHECK11-NEXT:    [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP35]], align 4
// CHECK11-NEXT:    [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 5, ptr [[TMP36]], align 4
// CHECK11-NEXT:    [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP27]], ptr [[TMP37]], align 4
// CHECK11-NEXT:    [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP28]], ptr [[TMP38]], align 4
// CHECK11-NEXT:    [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr [[TMP29]], ptr [[TMP39]], align 4
// CHECK11-NEXT:    [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes, ptr [[TMP40]], align 4
// CHECK11-NEXT:    [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP41]], align 4
// CHECK11-NEXT:    [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP42]], align 4
// CHECK11-NEXT:    [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT:    store i64 [[ADD]], ptr [[TMP43]], align 8
// CHECK11-NEXT:    [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP44]], align 8
// CHECK11-NEXT:    [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP45]], align 4
// CHECK11-NEXT:    [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP46]], align 4
// CHECK11-NEXT:    [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP47]], align 4
// CHECK11-NEXT:    [[TMP48:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT:    [[TMP49:%.*]] = icmp ne i32 [[TMP48]], 0
// CHECK11-NEXT:    br i1 [[TMP49]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11:       omp_offload.failed:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81(i32 [[TMP5]], i32 [[TMP7]], i32 [[TMP0]], i32 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK11:       omp_offload.cont:
// CHECK11-NEXT:    [[TMP50:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP50]])
// CHECK11-NEXT:    store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT:    [[TMP51:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP51]])
// CHECK11-NEXT:    [[TMP52:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT:    ret i32 [[TMP52]]
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81
// CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[M]], ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined, i32 [[TMP4]], i32 [[TMP6]], i32 [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I11:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J12:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_CASTED:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[M]], ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK11-NEXT:    [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK11-NEXT:    [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK11-NEXT:    [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT:    store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[J]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       land.lhs.true:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK11-NEXT:    br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i64 0, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK11-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK11-NEXT:    [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK11-NEXT:    br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK11-NEXT:    store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK11-NEXT:    br i1 [[CMP14]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK11-NEXT:    [[TMP20:%.*]] = trunc i64 [[TMP19]] to i32
// CHECK11-NEXT:    [[TMP21:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT:    [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32
// CHECK11-NEXT:    [[TMP23:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP23]], ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP24:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT:    [[TMP25:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP25]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    [[TMP26:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined.omp_outlined, i32 [[TMP20]], i32 [[TMP22]], i32 [[TMP24]], i32 [[TMP26]], i32 [[TMP0]], i32 [[TMP1]], ptr [[TMP2]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP27:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP28:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i64 [[TMP27]], [[TMP28]]
// CHECK11-NEXT:    store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[M_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[VLA_ADDR2:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP3:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I13:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J14:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[M]], ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK11-NEXT:    [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK11-NEXT:    [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK11-NEXT:    [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK11-NEXT:    [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK11-NEXT:    store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[I]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[J]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK11-NEXT:    br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK11:       land.lhs.true:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK11-NEXT:    br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.then:
// CHECK11-NEXT:    store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK11-NEXT:    [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[CONV11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    [[CONV12:%.*]] = zext i32 [[TMP11]] to i64
// CHECK11-NEXT:    store i64 [[CONV11]], ptr [[DOTOMP_LB]], align 8
// CHECK11-NEXT:    store i64 [[CONV12]], ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK11-NEXT:    [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    [[CMP15:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]]
// CHECK11-NEXT:    br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i64 [ [[TMP16]], [[COND_TRUE]] ], [ [[TMP17]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK11-NEXT:    store i64 [[TMP18]], ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT:    [[CMP16:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK11-NEXT:    br i1 [[CMP16]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[SUB17:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK11-NEXT:    [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1
// CHECK11-NEXT:    [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]]
// CHECK11-NEXT:    [[CONV20:%.*]] = sext i32 [[MUL19]] to i64
// CHECK11-NEXT:    [[DIV21:%.*]] = sdiv i64 [[TMP21]], [[CONV20]]
// CHECK11-NEXT:    [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i64 0, [[MUL22]]
// CHECK11-NEXT:    [[CONV23:%.*]] = trunc i64 [[ADD]] to i32
// CHECK11-NEXT:    store i32 [[CONV23]], ptr [[I13]], align 4
// CHECK11-NEXT:    [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[SUB24:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK11-NEXT:    [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK11-NEXT:    [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK11-NEXT:    [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK11-NEXT:    [[DIV28:%.*]] = sdiv i64 [[TMP24]], [[CONV27]]
// CHECK11-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK11-NEXT:    [[SUB29:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK11-NEXT:    [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1
// CHECK11-NEXT:    [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]]
// CHECK11-NEXT:    [[CONV32:%.*]] = sext i32 [[MUL31]] to i64
// CHECK11-NEXT:    [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]]
// CHECK11-NEXT:    [[SUB34:%.*]] = sub nsw i64 [[TMP23]], [[MUL33]]
// CHECK11-NEXT:    [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1
// CHECK11-NEXT:    [[ADD36:%.*]] = add nsw i64 0, [[MUL35]]
// CHECK11-NEXT:    [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32
// CHECK11-NEXT:    store i32 [[CONV37]], ptr [[J14]], align 4
// CHECK11-NEXT:    [[TMP27:%.*]] = load i32, ptr [[I13]], align 4
// CHECK11-NEXT:    [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP1]]
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP28]]
// CHECK11-NEXT:    [[TMP29:%.*]] = load i32, ptr [[J14]], align 4
// CHECK11-NEXT:    [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP29]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX38]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    [[ADD39:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK11-NEXT:    store i64 [[ADD39]], ptr [[DOTOMP_IV]], align 8
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP32]])
// CHECK11-NEXT:    br label [[OMP_PRECOND_END]]
// CHECK11:       omp.precond.end:
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
// CHECK11-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A:%.*]] = alloca [10 x [2 x i32]], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT:    store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT:    store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT:    [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT:    store i32 3, ptr [[TMP5]], align 4
// CHECK11-NEXT:    [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT:    store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT:    store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT:    [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT:    store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT:    store ptr @.offload_sizes.1, ptr [[TMP9]], align 4
// CHECK11-NEXT:    [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT:    store ptr @.offload_maptypes.2, ptr [[TMP10]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT:    store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT:    store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT:    store i64 20, ptr [[TMP13]], align 8
// CHECK11-NEXT:    [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9
// CHECK11-NEXT:    store i64 0, ptr [[TMP14]], align 8
// CHECK11-NEXT:    [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4
// CHECK11-NEXT:    [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11
// CHECK11-NEXT:    store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4
// CHECK11-NEXT:    [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12
// CHECK11-NEXT:    store i32 0, ptr [[TMP17]], align 4
// CHECK11-NEXT:    [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// CHECK11-NEXT:    br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK11:       omp_offload.failed:
// CHECK11-NEXT:    call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[A]]) #[[ATTR3]]
// CHECK11-NEXT:    br label [[OMP_OFFLOAD_CONT]]
// CHECK11:       omp_offload.cont:
// CHECK11-NEXT:    ret i32 0
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
// CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined, ptr [[TMP0]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 19, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 19
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK11-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]])
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT:    ret void
//
//
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.omp_outlined.omp_outlined
// CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(80) [[A:%.*]]) #[[ATTR2]] {
// CHECK11-NEXT:  entry:
// CHECK11-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK11-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    [[J:%.*]] = alloca i32, align 4
// CHECK11-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK11-NEXT:    [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK11-NEXT:    store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK11-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK11-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK11-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK11-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 19
// CHECK11-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK11:       cond.true:
// CHECK11-NEXT:    br label [[COND_END:%.*]]
// CHECK11:       cond.false:
// CHECK11-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    br label [[COND_END]]
// CHECK11:       cond.end:
// CHECK11-NEXT:    [[COND:%.*]] = phi i32 [ 19, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK11-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK11:       omp.inner.for.cond:
// CHECK11-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT:    [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK11-NEXT:    br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK11:       omp.inner.for.body:
// CHECK11-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP10]], 2
// CHECK11-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK11-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK11-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK11-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[DIV3:%.*]] = sdiv i32 [[TMP12]], 2
// CHECK11-NEXT:    [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 2
// CHECK11-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP11]], [[MUL4]]
// CHECK11-NEXT:    [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK11-NEXT:    [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK11-NEXT:    store i32 [[ADD6]], ptr [[J]], align 4
// CHECK11-NEXT:    [[TMP13:%.*]] = load i32, ptr [[I]], align 4
// CHECK11-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP13]]
// CHECK11-NEXT:    [[TMP14:%.*]] = load i32, ptr [[J]], align 4
// CHECK11-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK11-NEXT:    store i32 0, ptr [[ARRAYIDX7]], align 4
// CHECK11-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK11:       omp.body.continue:
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK11:       omp.inner.for.inc:
// CHECK11-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK11-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK11-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK11:       omp.inner.for.end:
// CHECK11-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK11:       omp.loop.exit:
// CHECK11-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]])
// CHECK11-NEXT:    ret void
//