// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _
// RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -emit-llvm %s -o - | FileCheck %s --check-prefix=IR

// Check same results after serialization round-trip
// RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -emit-pch -o %t %s
// RUN: %clang_cc1 -verify -triple x86_64-pc-linux-gnu -fopenmp -include-pch %t -emit-llvm %s -o - | FileCheck %s --check-prefix=IR-PCH

// expected-no-diagnostics

#ifndef HEADER
#define HEADER
int foo() {
  int i;
  int j;
  int sum[10][10];

  #pragma omp teams loop reduction(+:sum) collapse(2) bind(parallel) \
      order(concurrent) lastprivate(j)
  for(i=0; i<10; i++)
    for(j=0; j<10; j++)
      sum[i][j] += i;

  return 0;
}
#endif
// IR-LABEL: define {{[^@]+}}@_Z3foov
// IR-SAME: () #[[ATTR0:[0-9]+]] {
// IR-NEXT:  entry:
// IR-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-NEXT:    [[J:%.*]] = alloca i32, align 4
// IR-NEXT:    [[SUM:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB4:[0-9]+]], i32 2, ptr @_Z3foov.omp_outlined, ptr [[J]], ptr [[SUM]])
// IR-NEXT:    ret i32 0
//
//
// IR-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined
// IR-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[J:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[SUM:%.*]]) #[[ATTR1:[0-9]+]] {
// IR-NEXT:  entry:
// IR-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[J_ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[SUM_ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[SUM1:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// IR-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// IR-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// IR-NEXT:    [[J3:%.*]] = alloca i32, align 4
// IR-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-NEXT:    [[J4:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// IR-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// IR-NEXT:    store ptr [[J]], ptr [[J_ADDR]], align 8
// IR-NEXT:    store ptr [[SUM]], ptr [[SUM_ADDR]], align 8
// IR-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[J_ADDR]], align 8
// IR-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[SUM_ADDR]], align 8
// IR-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM1]], i32 0, i32 0, i32 0
// IR-NEXT:    [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 100
// IR-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// IR-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// IR:       omp.arrayinit.body:
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// IR-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// IR:       omp.arrayinit.done:
// IR-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// IR-NEXT:    store i32 99, ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// IR-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// IR-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// IR-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP4]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// IR-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
// IR-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// IR:       cond.true:
// IR-NEXT:    br label [[COND_END:%.*]]
// IR:       cond.false:
// IR-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    br label [[COND_END]]
// IR:       cond.end:
// IR-NEXT:    [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// IR-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// IR-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// IR-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// IR:       omp.inner.for.cond:
// IR-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// IR-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// IR-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// IR:       omp.inner.for.body:
// IR-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// IR-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// IR-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-NEXT:    [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// IR-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 4, ptr @_Z3foov.omp_outlined.omp_outlined, i64 [[TMP11]], i64 [[TMP13]], ptr [[J3]], ptr [[SUM1]])
// IR-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// IR:       omp.inner.for.inc:
// IR-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// IR-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// IR-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// IR-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// IR-NEXT:    br label [[OMP_INNER_FOR_COND]]
// IR:       omp.inner.for.end:
// IR-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// IR:       omp.loop.exit:
// IR-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// IR-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2:[0-9]+]], i32 [[TMP17]])
// IR-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// IR-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// IR-NEXT:    br i1 [[TMP19]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// IR:       .omp.lastprivate.then:
// IR-NEXT:    store i32 10, ptr [[J3]], align 4
// IR-NEXT:    [[TMP20:%.*]] = load i32, ptr [[J3]], align 4
// IR-NEXT:    store i32 [[TMP20]], ptr [[TMP0]], align 4
// IR-NEXT:    br label [[DOTOMP_LASTPRIVATE_DONE]]
// IR:       .omp.lastprivate.done:
// IR-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// IR-NEXT:    store ptr [[SUM1]], ptr [[TMP21]], align 8
// IR-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// IR-NEXT:    [[TMP24:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP23]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z3foov.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// IR-NEXT:    switch i32 [[TMP24]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// IR-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// IR-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// IR-NEXT:    ]
// IR:       .omp.reduction.case1:
// IR-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP25]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR:       omp.arraycpy.body:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[SUM1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[TMP26:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], align 4
// IR-NEXT:    [[TMP27:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP26]], [[TMP27]]
// IR-NEXT:    store i32 [[ADD7]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP25]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY]]
// IR:       omp.arraycpy.done10:
// IR-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP23]], ptr @.gomp_critical_user_.reduction.var)
// IR-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR:       .omp.reduction.case2:
// IR-NEXT:    [[TMP28:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY11:%.*]] = icmp eq ptr [[TMP1]], [[TMP28]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY11]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY12:%.*]]
// IR:       omp.arraycpy.body12:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST13:%.*]] = phi ptr [ [[SUM1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST14:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// IR-NEXT:    [[TMP29:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], align 4
// IR-NEXT:    [[TMP30:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 [[TMP29]] monotonic, align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT15]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT16]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT15]], [[TMP28]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY12]]
// IR:       omp.arraycpy.done18:
// IR-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR:       .omp.reduction.default:
// IR-NEXT:    ret void
//
//
// IR-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp_outlined
// IR-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(4) [[J:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[SUM:%.*]]) #[[ATTR1]] {
// IR-NEXT:  entry:
// IR-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// IR-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// IR-NEXT:    [[J_ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[SUM_ADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// IR-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// IR-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// IR-NEXT:    [[J3:%.*]] = alloca i32, align 4
// IR-NEXT:    [[SUM4:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-NEXT:    [[J5:%.*]] = alloca i32, align 4
// IR-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// IR-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// IR-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// IR-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// IR-NEXT:    store ptr [[J]], ptr [[J_ADDR]], align 8
// IR-NEXT:    store ptr [[SUM]], ptr [[SUM_ADDR]], align 8
// IR-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[J_ADDR]], align 8
// IR-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[SUM_ADDR]], align 8
// IR-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// IR-NEXT:    store i32 99, ptr [[DOTOMP_UB]], align 4
// IR-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// IR-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP2]] to i32
// IR-NEXT:    [[TMP3:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// IR-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
// IR-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// IR-NEXT:    store i32 [[CONV2]], ptr [[DOTOMP_UB]], align 4
// IR-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// IR-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// IR-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM4]], i32 0, i32 0, i32 0
// IR-NEXT:    [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 100
// IR-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]]
// IR-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// IR:       omp.arrayinit.body:
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// IR-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// IR:       omp.arrayinit.done:
// IR-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// IR-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// IR-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// IR-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP7]], 99
// IR-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// IR:       cond.true:
// IR-NEXT:    br label [[COND_END:%.*]]
// IR:       cond.false:
// IR-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// IR-NEXT:    br label [[COND_END]]
// IR:       cond.end:
// IR-NEXT:    [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
// IR-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// IR-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// IR-NEXT:    store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// IR-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// IR:       omp.inner.for.cond:
// IR-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// IR-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
// IR-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// IR:       omp.inner.for.body:
// IR-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP12]], 10
// IR-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// IR-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// IR-NEXT:    store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[DIV7:%.*]] = sdiv i32 [[TMP14]], 10
// IR-NEXT:    [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 10
// IR-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP13]], [[MUL8]]
// IR-NEXT:    [[MUL9:%.*]] = mul nsw i32 [[SUB]], 1
// IR-NEXT:    [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
// IR-NEXT:    store i32 [[ADD10]], ptr [[J3]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[TMP16:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64
// IR-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM4]], i64 0, i64 [[IDXPROM]]
// IR-NEXT:    [[TMP17:%.*]] = load i32, ptr [[J3]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP17]] to i64
// IR-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM11]]
// IR-NEXT:    [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP18]], [[TMP15]]
// IR-NEXT:    store i32 [[ADD13]], ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// IR:       omp.body.continue:
// IR-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// IR:       omp.inner.for.inc:
// IR-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP19]], 1
// IR-NEXT:    store i32 [[ADD14]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// IR:       omp.inner.for.end:
// IR-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// IR:       omp.loop.exit:
// IR-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// IR-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]])
// IR-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// IR-NEXT:    store ptr [[SUM4]], ptr [[TMP22]], align 8
// IR-NEXT:    [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
// IR-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP24]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z3foov.omp_outlined.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// IR-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// IR-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// IR-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// IR-NEXT:    ]
// IR:       .omp.reduction.case1:
// IR-NEXT:    [[TMP26:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP26]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE19:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR:       omp.arraycpy.body:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[SUM4]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[TMP27:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 4
// IR-NEXT:    [[TMP28:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
// IR-NEXT:    store i32 [[ADD16]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT17]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE18:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT17]], [[TMP26]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_BODY]]
// IR:       omp.arraycpy.done19:
// IR-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// IR-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR:       .omp.reduction.case2:
// IR-NEXT:    [[TMP29:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY20:%.*]] = icmp eq ptr [[TMP1]], [[TMP29]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY20]], label [[OMP_ARRAYCPY_DONE27:%.*]], label [[OMP_ARRAYCPY_BODY21:%.*]]
// IR:       omp.arraycpy.body21:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST22:%.*]] = phi ptr [ [[SUM4]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT25:%.*]], [[OMP_ARRAYCPY_BODY21]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST23:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY21]] ]
// IR-NEXT:    [[TMP30:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST22]], align 4
// IR-NEXT:    [[TMP31:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST23]], i32 [[TMP30]] monotonic, align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT24]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST23]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT25]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST22]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE26:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT24]], [[TMP29]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_DONE27]], label [[OMP_ARRAYCPY_BODY21]]
// IR:       omp.arraycpy.done27:
// IR-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR:       .omp.reduction.default:
// IR-NEXT:    [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// IR-NEXT:    [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// IR-NEXT:    br i1 [[TMP33]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// IR:       .omp.lastprivate.then:
// IR-NEXT:    store i32 10, ptr [[J3]], align 4
// IR-NEXT:    [[TMP34:%.*]] = load i32, ptr [[J3]], align 4
// IR-NEXT:    store i32 [[TMP34]], ptr [[TMP0]], align 4
// IR-NEXT:    br label [[DOTOMP_LASTPRIVATE_DONE]]
// IR:       .omp.lastprivate.done:
// IR-NEXT:    ret void
//
//
// IR-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp_outlined.omp.reduction.reduction_func
// IR-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// IR-NEXT:  entry:
// IR-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// IR-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// IR-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// IR-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// IR-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// IR-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// IR-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// IR-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// IR-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// IR-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR:       omp.arraycpy.body:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// IR-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// IR:       omp.arraycpy.done2:
// IR-NEXT:    ret void
//
//
// IR-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp.reduction.reduction_func
// IR-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// IR-NEXT:  entry:
// IR-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// IR-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// IR-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// IR-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// IR-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// IR-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// IR-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// IR-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// IR-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// IR-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// IR-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 100
// IR-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR:       omp.arraycpy.body:
// IR-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-NEXT:    [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// IR-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// IR-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// IR:       omp.arraycpy.done2:
// IR-NEXT:    ret void
//
//
// IR-PCH-LABEL: define {{[^@]+}}@_Z3foov
// IR-PCH-SAME: () #[[ATTR0:[0-9]+]] {
// IR-PCH-NEXT:  entry:
// IR-PCH-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[J:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[SUM:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-PCH-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB4:[0-9]+]], i32 2, ptr @_Z3foov.omp_outlined, ptr [[J]], ptr [[SUM]])
// IR-PCH-NEXT:    ret i32 0
//
//
// IR-PCH-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined
// IR-PCH-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[J:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[SUM:%.*]]) #[[ATTR1:[0-9]+]] {
// IR-PCH-NEXT:  entry:
// IR-PCH-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[J_ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[SUM_ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[SUM1:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-PCH-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[J3:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[J4:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// IR-PCH-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[J]], ptr [[J_ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[SUM]], ptr [[SUM_ADDR]], align 8
// IR-PCH-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[J_ADDR]], align 8
// IR-PCH-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[SUM_ADDR]], align 8
// IR-PCH-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM1]], i32 0, i32 0, i32 0
// IR-PCH-NEXT:    [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// IR-PCH:       omp.arrayinit.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// IR-PCH-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// IR-PCH:       omp.arrayinit.done:
// IR-PCH-NEXT:    store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// IR-PCH-NEXT:    store i32 99, ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// IR-PCH-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// IR-PCH-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
// IR-PCH-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP4]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// IR-PCH-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 99
// IR-PCH-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// IR-PCH:       cond.true:
// IR-PCH-NEXT:    br label [[COND_END:%.*]]
// IR-PCH:       cond.false:
// IR-PCH-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    br label [[COND_END]]
// IR-PCH:       cond.end:
// IR-PCH-NEXT:    [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// IR-PCH-NEXT:    store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// IR-PCH-NEXT:    store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// IR-PCH:       omp.inner.for.cond:
// IR-PCH-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// IR-PCH-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// IR-PCH-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// IR-PCH:       omp.inner.for.body:
// IR-PCH-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// IR-PCH-NEXT:    [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// IR-PCH-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// IR-PCH-NEXT:    [[TMP13:%.*]] = zext i32 [[TMP12]] to i64
// IR-PCH-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 4, ptr @_Z3foov.omp_outlined.omp_outlined, i64 [[TMP11]], i64 [[TMP13]], ptr [[J3]], ptr [[SUM1]])
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// IR-PCH:       omp.inner.for.inc:
// IR-PCH-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// IR-PCH-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// IR-PCH-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// IR-PCH-NEXT:    store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND]]
// IR-PCH:       omp.inner.for.end:
// IR-PCH-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// IR-PCH:       omp.loop.exit:
// IR-PCH-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// IR-PCH-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2:[0-9]+]], i32 [[TMP17]])
// IR-PCH-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// IR-PCH-NEXT:    [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0
// IR-PCH-NEXT:    br i1 [[TMP19]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// IR-PCH:       .omp.lastprivate.then:
// IR-PCH-NEXT:    store i32 10, ptr [[J3]], align 4
// IR-PCH-NEXT:    [[TMP20:%.*]] = load i32, ptr [[J3]], align 4
// IR-PCH-NEXT:    store i32 [[TMP20]], ptr [[TMP0]], align 4
// IR-PCH-NEXT:    br label [[DOTOMP_LASTPRIVATE_DONE]]
// IR-PCH:       .omp.lastprivate.done:
// IR-PCH-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// IR-PCH-NEXT:    store ptr [[SUM1]], ptr [[TMP21]], align 8
// IR-PCH-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// IR-PCH-NEXT:    [[TMP24:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP23]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z3foov.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// IR-PCH-NEXT:    switch i32 [[TMP24]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// IR-PCH-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// IR-PCH-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// IR-PCH-NEXT:    ]
// IR-PCH:       .omp.reduction.case1:
// IR-PCH-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP25]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR-PCH:       omp.arraycpy.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[SUM1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[TMP26:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], align 4
// IR-PCH-NEXT:    [[TMP27:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP26]], [[TMP27]]
// IR-PCH-NEXT:    store i32 [[ADD7]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP25]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY]]
// IR-PCH:       omp.arraycpy.done10:
// IR-PCH-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP23]], ptr @.gomp_critical_user_.reduction.var)
// IR-PCH-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR-PCH:       .omp.reduction.case2:
// IR-PCH-NEXT:    [[TMP28:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY11:%.*]] = icmp eq ptr [[TMP1]], [[TMP28]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY11]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY12:%.*]]
// IR-PCH:       omp.arraycpy.body12:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST13:%.*]] = phi ptr [ [[SUM1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST14:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// IR-PCH-NEXT:    [[TMP29:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], align 4
// IR-PCH-NEXT:    [[TMP30:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 [[TMP29]] monotonic, align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT15]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT16]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT15]], [[TMP28]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY12]]
// IR-PCH:       omp.arraycpy.done18:
// IR-PCH-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR-PCH:       .omp.reduction.default:
// IR-PCH-NEXT:    ret void
//
//
// IR-PCH-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp_outlined
// IR-PCH-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(4) [[J:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[SUM:%.*]]) #[[ATTR1]] {
// IR-PCH-NEXT:  entry:
// IR-PCH-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8
// IR-PCH-NEXT:    [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8
// IR-PCH-NEXT:    [[J_ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[SUM_ADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[J3:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[SUM4:%.*]] = alloca [10 x [10 x i32]], align 16
// IR-PCH-NEXT:    [[I:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[J5:%.*]] = alloca i32, align 4
// IR-PCH-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// IR-PCH-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// IR-PCH-NEXT:    store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// IR-PCH-NEXT:    store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[J]], ptr [[J_ADDR]], align 8
// IR-PCH-NEXT:    store ptr [[SUM]], ptr [[SUM_ADDR]], align 8
// IR-PCH-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[J_ADDR]], align 8
// IR-PCH-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[SUM_ADDR]], align 8
// IR-PCH-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// IR-PCH-NEXT:    store i32 99, ptr [[DOTOMP_UB]], align 4
// IR-PCH-NEXT:    [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// IR-PCH-NEXT:    [[CONV:%.*]] = trunc i64 [[TMP2]] to i32
// IR-PCH-NEXT:    [[TMP3:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// IR-PCH-NEXT:    [[CONV2:%.*]] = trunc i64 [[TMP3]] to i32
// IR-PCH-NEXT:    store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// IR-PCH-NEXT:    store i32 [[CONV2]], ptr [[DOTOMP_UB]], align 4
// IR-PCH-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// IR-PCH-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// IR-PCH-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM4]], i32 0, i32 0, i32 0
// IR-PCH-NEXT:    [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// IR-PCH:       omp.arrayinit.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// IR-PCH-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// IR-PCH:       omp.arrayinit.done:
// IR-PCH-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// IR-PCH-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP6]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// IR-PCH-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// IR-PCH-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP7]], 99
// IR-PCH-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// IR-PCH:       cond.true:
// IR-PCH-NEXT:    br label [[COND_END:%.*]]
// IR-PCH:       cond.false:
// IR-PCH-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// IR-PCH-NEXT:    br label [[COND_END]]
// IR-PCH:       cond.end:
// IR-PCH-NEXT:    [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
// IR-PCH-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// IR-PCH-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// IR-PCH-NEXT:    store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// IR-PCH:       omp.inner.for.cond:
// IR-PCH-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// IR-PCH-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
// IR-PCH-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// IR-PCH:       omp.inner.for.body:
// IR-PCH-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[DIV:%.*]] = sdiv i32 [[TMP12]], 10
// IR-PCH-NEXT:    [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// IR-PCH-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// IR-PCH-NEXT:    store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[DIV7:%.*]] = sdiv i32 [[TMP14]], 10
// IR-PCH-NEXT:    [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 10
// IR-PCH-NEXT:    [[SUB:%.*]] = sub nsw i32 [[TMP13]], [[MUL8]]
// IR-PCH-NEXT:    [[MUL9:%.*]] = mul nsw i32 [[SUB]], 1
// IR-PCH-NEXT:    [[ADD10:%.*]] = add nsw i32 0, [[MUL9]]
// IR-PCH-NEXT:    store i32 [[ADD10]], ptr [[J3]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[TMP15:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[TMP16:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64
// IR-PCH-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[SUM4]], i64 0, i64 [[IDXPROM]]
// IR-PCH-NEXT:    [[TMP17:%.*]] = load i32, ptr [[J3]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[IDXPROM11:%.*]] = sext i32 [[TMP17]] to i64
// IR-PCH-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM11]]
// IR-PCH-NEXT:    [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP18]], [[TMP15]]
// IR-PCH-NEXT:    store i32 [[ADD13]], ptr [[ARRAYIDX12]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// IR-PCH:       omp.body.continue:
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// IR-PCH:       omp.inner.for.inc:
// IR-PCH-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    [[ADD14:%.*]] = add nsw i32 [[TMP19]], 1
// IR-PCH-NEXT:    store i32 [[ADD14]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// IR-PCH-NEXT:    br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// IR-PCH:       omp.inner.for.end:
// IR-PCH-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// IR-PCH:       omp.loop.exit:
// IR-PCH-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// IR-PCH-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]])
// IR-PCH-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// IR-PCH-NEXT:    store ptr [[SUM4]], ptr [[TMP22]], align 8
// IR-PCH-NEXT:    [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// IR-PCH-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
// IR-PCH-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP24]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z3foov.omp_outlined.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// IR-PCH-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// IR-PCH-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// IR-PCH-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// IR-PCH-NEXT:    ]
// IR-PCH:       .omp.reduction.case1:
// IR-PCH-NEXT:    [[TMP26:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP26]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE19:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR-PCH:       omp.arraycpy.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[SUM4]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[TMP27:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 4
// IR-PCH-NEXT:    [[TMP28:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[ADD16:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
// IR-PCH-NEXT:    store i32 [[ADD16]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT17]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE18:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT17]], [[TMP26]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_BODY]]
// IR-PCH:       omp.arraycpy.done19:
// IR-PCH-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// IR-PCH-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR-PCH:       .omp.reduction.case2:
// IR-PCH-NEXT:    [[TMP29:%.*]] = getelementptr i32, ptr [[TMP1]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY20:%.*]] = icmp eq ptr [[TMP1]], [[TMP29]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY20]], label [[OMP_ARRAYCPY_DONE27:%.*]], label [[OMP_ARRAYCPY_BODY21:%.*]]
// IR-PCH:       omp.arraycpy.body21:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST22:%.*]] = phi ptr [ [[SUM4]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT25:%.*]], [[OMP_ARRAYCPY_BODY21]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST23:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY21]] ]
// IR-PCH-NEXT:    [[TMP30:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST22]], align 4
// IR-PCH-NEXT:    [[TMP31:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST23]], i32 [[TMP30]] monotonic, align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT24]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST23]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT25]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST22]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE26:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT24]], [[TMP29]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_DONE27]], label [[OMP_ARRAYCPY_BODY21]]
// IR-PCH:       omp.arraycpy.done27:
// IR-PCH-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// IR-PCH:       .omp.reduction.default:
// IR-PCH-NEXT:    [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// IR-PCH-NEXT:    [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0
// IR-PCH-NEXT:    br i1 [[TMP33]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// IR-PCH:       .omp.lastprivate.then:
// IR-PCH-NEXT:    store i32 10, ptr [[J3]], align 4
// IR-PCH-NEXT:    [[TMP34:%.*]] = load i32, ptr [[J3]], align 4
// IR-PCH-NEXT:    store i32 [[TMP34]], ptr [[TMP0]], align 4
// IR-PCH-NEXT:    br label [[DOTOMP_LASTPRIVATE_DONE]]
// IR-PCH:       .omp.lastprivate.done:
// IR-PCH-NEXT:    ret void
//
//
// IR-PCH-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp_outlined.omp.reduction.reduction_func
// IR-PCH-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// IR-PCH-NEXT:  entry:
// IR-PCH-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// IR-PCH-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// IR-PCH-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// IR-PCH-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// IR-PCH-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// IR-PCH-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// IR-PCH-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// IR-PCH-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// IR-PCH-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR-PCH:       omp.arraycpy.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// IR-PCH-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// IR-PCH:       omp.arraycpy.done2:
// IR-PCH-NEXT:    ret void
//
//
// IR-PCH-LABEL: define {{[^@]+}}@_Z3foov.omp_outlined.omp.reduction.reduction_func
// IR-PCH-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// IR-PCH-NEXT:  entry:
// IR-PCH-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// IR-PCH-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// IR-PCH-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// IR-PCH-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// IR-PCH-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// IR-PCH-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// IR-PCH-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// IR-PCH-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// IR-PCH-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// IR-PCH-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 100
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// IR-PCH:       omp.arraycpy.body:
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// IR-PCH-NEXT:    [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// IR-PCH-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// IR-PCH-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// IR-PCH-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// IR-PCH:       omp.arraycpy.done2:
// IR-PCH-NEXT:    ret void
//