[Attributor] Set floating point loads and stores as nofree in AANoFreeFloating

[jhuber6]

Summary

The current implementation of AANoFreeFloating will incorrectly list floating
point loads and stores as may-free. This prevents other attributor instances
like HeapToStack from pushing some allocations to the stack.

[jhuber6]

Fixing tests.

[jdoerfert]

LG, two nits

llvm/test/Transforms/Attributor/ArgumentPromotion/byval.ll

163 ↗

(On Diff #350956)

Run the update script on this file again until the diff disappears, unfortunate non-determinism somewhere.

llvm/test/Transforms/OpenMP/hide_mem_transfer_latency.ll

1 ↗

(On Diff #350956)

Maybe keep the existing options as part of this commit.

[jhuber6]

Fixing nits.

llvm/lib/Transforms/IPO/AttributorAttributes.cpp

auto Pred = [&](const Use &U, bool &Follow) -> bool {
         return NoFreeArg.isAssumedNoFree();
       }
 
       if (isa<GetElementPtrInst>(UserI) || isa<BitCastInst>(UserI) ||
           isa<PHINode>(UserI) || isa<SelectInst>(UserI)) {
         Follow = true;
         return true;
       }
-      if (isa<ReturnInst>(UserI))
+      if (isa<StoreInst>(UserI) || isa<LoadInst>(UserI) ||
+          isa<ReturnInst>(UserI))
         return true;
 
       // Unknown user.
       return false;
     };
     if (!A.checkForAllUses(Pred, *this, AssociatedValue))
       return indicatePessimisticFixpoint();
 

llvm/test/Transforms/Attributor/ArgumentPromotion/alignment.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
 ; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=9 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=9 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
 ; RUN: opt -attributor-cgscc -enable-new-pm=0 -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_NPM,IS__CGSCC____,IS________OPM,IS__CGSCC_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_OPM,IS__CGSCC____,IS________NPM,IS__CGSCC_NPM
 
 define void @f() {
 ; IS________OPM-LABEL: define {{[^@]+}}@f() {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[A:%.*]] = alloca i32, align 1
-; IS________OPM-NEXT:    call void @g(i32* noalias nocapture noundef nonnull readonly dereferenceable(4) [[A]])
+; IS________OPM-NEXT:    call void @g(i32* noalias nocapture nofree noundef nonnull readonly dereferenceable(4) [[A]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@f() {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[A:%.*]] = alloca i32, align 1
 ; IS________NPM-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 1
 ; IS________NPM-NEXT:    call void @g(i32 [[TMP0]])
 ; IS________NPM-NEXT:    ret void
 ;
 entry:
   %a = alloca i32, align 1
   call void @g(i32* %a)
   ret void
 }
 
 define internal void @g(i32* %a) {
 ; IS________OPM-LABEL: define {{[^@]+}}@g
-; IS________OPM-SAME: (i32* noalias nocapture noundef nonnull readonly dereferenceable(4) [[A:%.*]]) {
+; IS________OPM-SAME: (i32* noalias nocapture nofree noundef nonnull readonly dereferenceable(4) [[A:%.*]]) {
 ; IS________OPM-NEXT:    [[AA:%.*]] = load i32, i32* [[A]], align 1
 ; IS________OPM-NEXT:    call void @z(i32 [[AA]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@g
 ; IS__TUNIT_NPM-SAME: (i32 [[TMP0:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:    [[A_PRIV:%.*]] = alloca i32, align 4
 ; IS__TUNIT_NPM-NEXT:    store i32 [[TMP0]], i32* [[A_PRIV]], align 4

llvm/test/Transforms/Attributor/ArgumentPromotion/dbg.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
 ; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=2 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
 ; RUN: opt -attributor-cgscc -enable-new-pm=0 -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_NPM,IS__CGSCC____,IS________OPM,IS__CGSCC_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_OPM,IS__CGSCC____,IS________NPM,IS__CGSCC_NPM
 
 declare void @sink(i32)
 
 define internal void @test(i32** %X) !dbg !2 {
 ; CHECK-LABEL: define {{[^@]+}}@test
-; CHECK-SAME: (i32** nocapture noundef nonnull readonly align 8 dereferenceable(8) [[X:%.*]]) !dbg [[DBG3:![0-9]+]] {
+; CHECK-SAME: (i32** nocapture nofree noundef nonnull readonly align 8 dereferenceable(8) [[X:%.*]]) !dbg [[DBG3:![0-9]+]] {
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32*, i32** [[X]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 8
 ; CHECK-NEXT:    call void @sink(i32 [[TMP2]])
 ; CHECK-NEXT:    ret void
 ;
   %1 = load i32*, i32** %X, align 8
   %2 = load i32, i32* %1, align 8
   call void @sink(i32 %2)
   ret void
 }
 
 %struct.pair = type { i32, i32 }
 
 define internal void @test_byval(%struct.pair* byval(%struct.pair) %P) {
 ; CHECK-LABEL: define {{[^@]+}}@test_byval() {
 ; CHECK-NEXT:    call void @sink(i32 noundef 0)
 ; CHECK-NEXT:    ret void
 ;
   call void @sink(i32 0)
   ret void
 }
 
 define void @caller(i32** %Y, %struct.pair* %P) {
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@caller
-; IS__TUNIT____-SAME: (i32** nocapture readonly [[Y:%.*]], %struct.pair* nocapture nofree readnone [[P:%.*]]) {
-; IS__TUNIT____-NEXT:    call void @test(i32** nocapture readonly align 8 [[Y]]), !dbg [[DBG4:![0-9]+]]
+; IS__TUNIT____-SAME: (i32** nocapture nofree readonly [[Y:%.*]], %struct.pair* nocapture nofree readnone [[P:%.*]]) {
+; IS__TUNIT____-NEXT:    call void @test(i32** nocapture nofree readonly align 8 [[Y]]), !dbg [[DBG4:![0-9]+]]
 ; IS__TUNIT____-NEXT:    call void @test_byval(), !dbg [[DBG5:![0-9]+]]
 ; IS__TUNIT____-NEXT:    ret void
 ;
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@caller
-; IS__CGSCC____-SAME: (i32** nocapture noundef nonnull readonly align 8 dereferenceable(8) [[Y:%.*]], %struct.pair* nocapture nofree readnone [[P:%.*]]) {
-; IS__CGSCC____-NEXT:    call void @test(i32** nocapture noundef nonnull readonly align 8 dereferenceable(8) [[Y]]), !dbg [[DBG4:![0-9]+]]
+; IS__CGSCC____-SAME: (i32** nocapture nofree noundef nonnull readonly align 8 dereferenceable(8) [[Y:%.*]], %struct.pair* nocapture nofree readnone [[P:%.*]]) {
+; IS__CGSCC____-NEXT:    call void @test(i32** nocapture nofree noundef nonnull readonly align 8 dereferenceable(8) [[Y]]), !dbg [[DBG4:![0-9]+]]
 ; IS__CGSCC____-NEXT:    call void @test_byval(), !dbg [[DBG5:![0-9]+]]
 ; IS__CGSCC____-NEXT:    ret void
 ;
   call void @test(i32** %Y), !dbg !1
 
   call void @test_byval(%struct.pair* byval(%struct.pair) %P), !dbg !6
   ret void
 }

llvm/test/Transforms/Attributor/ArgumentPromotion/fp80.ll

 ; CHECK: @[[B:[a-zA-Z0-9_$"\\.-]+]] = internal global [[STRUCT_S:%.*]] { double 3.140000e+00, i16 9439, i8 25, [5 x i8] undef }, align 16
 ; CHECK: @[[A:[a-zA-Z0-9_$"\\.-]+]] = internal global [[STRUCT_FOO:%.*]] { i32 1, i64 2 }, align 8
 ;.
 define void @run() {
 ; IS________OPM: Function Attrs: nofree noreturn nosync nounwind readnone
 ; IS________OPM-LABEL: define {{[^@]+}}@run
 ; IS________OPM-SAME: () #[[ATTR0:[0-9]+]] {
 ; IS________OPM-NEXT:  entry:
-; IS________OPM-NEXT:    [[TMP0:%.*]] = call i64 @CaptureAStruct(%struct.Foo* nocapture nofree noundef nonnull readonly byval(%struct.Foo) align 8 dereferenceable(16) @a) #[[ATTR0]]
+; IS________OPM-NEXT:    [[TMP0:%.*]] = call i64 @CaptureAStruct(%struct.Foo* nocapture nofree noundef nonnull readonly byval([[STRUCT_FOO:%.*]]) align 8 dereferenceable(16) @a) #[[ATTR0]]
 ; IS________OPM-NEXT:    unreachable
 ;
 ; IS__TUNIT_NPM: Function Attrs: nofree noreturn nosync nounwind readnone
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@run
 ; IS__TUNIT_NPM-SAME: () #[[ATTR0:[0-9]+]] {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[A_CAST:%.*]] = bitcast %struct.Foo* @a to i32*
 ; IS__TUNIT_NPM-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_CAST]], align 8

llvm/test/Transforms/Attributor/ArgumentPromotion/profile.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
 ; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=3 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=3 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_OPM,NOT_CGSCC_NPM,NOT_TUNIT_OPM,IS__TUNIT____,IS________NPM,IS__TUNIT_NPM
 ; RUN: opt -attributor-cgscc -enable-new-pm=0 -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_NPM,IS__CGSCC____,IS________OPM,IS__CGSCC_OPM
 ; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_TUNIT_NPM,NOT_TUNIT_OPM,NOT_CGSCC_OPM,IS__CGSCC____,IS________NPM,IS__CGSCC_NPM
 target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128"
 
 ; Checks if !prof metadata is corret in deadargelim.
 
 define void @caller() #0 {
 ; IS________OPM-LABEL: define {{[^@]+}}@caller() {
 ; IS________OPM-NEXT:    [[X:%.*]] = alloca i32, align 4
 ; IS________OPM-NEXT:    store i32 42, i32* [[X]], align 4
-; IS________OPM-NEXT:    call void @promote_i32_ptr(i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[X]]), !prof [[PROF0:![0-9]+]]
+; IS________OPM-NEXT:    call void @promote_i32_ptr(i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[X]]), !prof [[PROF0:![0-9]+]]
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@caller() {
 ; IS________NPM-NEXT:    [[X:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    store i32 42, i32* [[X]], align 4
 ; IS________NPM-NEXT:    [[TMP1:%.*]] = load i32, i32* [[X]], align 4
 ; IS________NPM-NEXT:    call void @promote_i32_ptr(i32 [[TMP1]]), !prof [[PROF0:![0-9]+]]
 ; IS________NPM-NEXT:    ret void
 ;
   %x = alloca i32
   store i32 42, i32* %x
   call void @promote_i32_ptr(i32* %x), !prof !0
   ret void
 }
 
 define internal void @promote_i32_ptr(i32* %xp) {
 ; IS________OPM-LABEL: define {{[^@]+}}@promote_i32_ptr
-; IS________OPM-SAME: (i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[XP:%.*]]) {
+; IS________OPM-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[XP:%.*]]) {
 ; IS________OPM-NEXT:    [[X:%.*]] = load i32, i32* [[XP]], align 4
 ; IS________OPM-NEXT:    call void @use_i32(i32 [[X]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@promote_i32_ptr
 ; IS________NPM-SAME: (i32 [[TMP0:%.*]]) {
 ; IS________NPM-NEXT:    [[XP_PRIV:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    store i32 [[TMP0]], i32* [[XP_PRIV]], align 4

llvm/test/Transforms/Attributor/ArgumentPromotion/tail.ll

 }
 
 define void @zed(%pair* byval(%pair) %Data) {
 ; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@zed
 ; IS__TUNIT_OPM-SAME: (%pair* noalias nocapture nonnull readonly byval([[PAIR:%.*]]) dereferenceable(8) [[DATA:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:    call void @bar(%pair* noalias nocapture nonnull readonly byval([[PAIR]]) dereferenceable(8) [[DATA]])
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
-; IS________NPM-LABEL: define {{[^@]+}}@zed
-; IS________NPM-SAME: (%pair* noalias nocapture nonnull readonly byval([[PAIR:%.*]]) dereferenceable(8) [[DATA:%.*]]) {
-; IS________NPM-NEXT:    [[DATA_CAST:%.*]] = bitcast %pair* [[DATA]] to i32*
-; IS________NPM-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DATA_CAST]], align 1
-; IS________NPM-NEXT:    [[DATA_0_1:%.*]] = getelementptr [[PAIR]], %pair* [[DATA]], i32 0, i32 1
-; IS________NPM-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DATA_0_1]], align 1
-; IS________NPM-NEXT:    call void @bar(i32 [[TMP1]], i32 [[TMP2]])
-; IS________NPM-NEXT:    ret void
+; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@zed
+; IS__TUNIT_NPM-SAME: (%pair* noalias nocapture nonnull readonly byval([[PAIR:%.*]]) dereferenceable(8) [[DATA:%.*]]) {
+; IS__TUNIT_NPM-NEXT:    [[DATA_CAST:%.*]] = bitcast %pair* [[DATA]] to i32*
+; IS__TUNIT_NPM-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DATA_CAST]], align 1
+; IS__TUNIT_NPM-NEXT:    [[DATA_0_1:%.*]] = getelementptr [[PAIR]], %pair* [[DATA]], i32 0, i32 1
+; IS__TUNIT_NPM-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DATA_0_1]], align 1
+; IS__TUNIT_NPM-NEXT:    call void @bar(i32 [[TMP1]], i32 [[TMP2]])
+; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@zed
 ; IS__CGSCC_OPM-SAME: (%pair* noalias nocapture noundef nonnull readonly byval([[PAIR:%.*]]) dereferenceable(8) [[DATA:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:    call void @bar(%pair* noalias nocapture noundef nonnull readonly byval([[PAIR]]) dereferenceable(8) [[DATA]])
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
+; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@zed
+; IS__CGSCC_NPM-SAME: (%pair* noalias nocapture nofree nonnull readonly byval([[PAIR:%.*]]) dereferenceable(8) [[DATA:%.*]]) {
+; IS__CGSCC_NPM-NEXT:    [[DATA_CAST:%.*]] = bitcast %pair* [[DATA]] to i32*
+; IS__CGSCC_NPM-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DATA_CAST]], align 1
+; IS__CGSCC_NPM-NEXT:    [[DATA_0_1:%.*]] = getelementptr [[PAIR]], %pair* [[DATA]], i32 0, i32 1
+; IS__CGSCC_NPM-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DATA_0_1]], align 1
+; IS__CGSCC_NPM-NEXT:    call void @bar(i32 [[TMP1]], i32 [[TMP2]])
+; IS__CGSCC_NPM-NEXT:    ret void
+;
   call void @bar(%pair* byval(%pair) %Data)
   ret void
 }

llvm/test/Transforms/Attributor/IPConstantProp/openmp_parallel_for.ll

 ; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@foo
 ; IS__TUNIT_OPM-SAME: (i32 [[N:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:  entry:
 ; IS__TUNIT_OPM-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
 ; IS__TUNIT_OPM-NEXT:    [[P:%.*]] = alloca float, align 4
 ; IS__TUNIT_OPM-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
 ; IS__TUNIT_OPM-NEXT:    store float 3.000000e+00, float* [[P]], align 4
 ; IS__TUNIT_OPM-NEXT:    store i32 7, i32* [[N_ADDR]], align 4
-; IS__TUNIT_OPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 undef)
+; IS__TUNIT_OPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 undef)
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@foo
 ; IS__TUNIT_NPM-SAME: (i32 [[N:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
 ; IS__TUNIT_NPM-NEXT:    [[P:%.*]] = alloca float, align 4
 ; IS__TUNIT_NPM-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
 ; IS__TUNIT_NPM-NEXT:    store float 3.000000e+00, float* [[P]], align 4
 ; IS__TUNIT_NPM-NEXT:    store i32 7, i32* [[N_ADDR]], align 4
-; IS__TUNIT_NPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 undef)
+; IS__TUNIT_NPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 undef)
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@foo
 ; IS__CGSCC_OPM-SAME: (i32 [[N:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:  entry:
 ; IS__CGSCC_OPM-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
 ; IS__CGSCC_OPM-NEXT:    [[P:%.*]] = alloca float, align 4
 ; IS__CGSCC_OPM-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
 ; IS__CGSCC_OPM-NEXT:    store float 3.000000e+00, float* [[P]], align 4
 ; IS__CGSCC_OPM-NEXT:    store i32 7, i32* [[N_ADDR]], align 4
-; IS__CGSCC_OPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 noundef 4617315517961601024)
+; IS__CGSCC_OPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 noundef 4617315517961601024)
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
 ; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@foo
 ; IS__CGSCC_NPM-SAME: (i32 [[N:%.*]]) {
 ; IS__CGSCC_NPM-NEXT:  entry:
 ; IS__CGSCC_NPM-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
 ; IS__CGSCC_NPM-NEXT:    [[P:%.*]] = alloca float, align 4
 ; IS__CGSCC_NPM-NEXT:    store i32 [[N]], i32* [[N_ADDR]], align 4
 ; IS__CGSCC_NPM-NEXT:    store float 3.000000e+00, float* [[P]], align 4
 ; IS__CGSCC_NPM-NEXT:    store i32 7, i32* [[N_ADDR]], align 4
-; IS__CGSCC_NPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 noundef 4617315517961601024)
+; IS__CGSCC_NPM-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], float* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 noundef 4617315517961601024)
 ; IS__CGSCC_NPM-NEXT:    ret void
 ;
 entry:
   %N.addr = alloca i32, align 4
   %p = alloca float, align 4
   store i32 %N, i32* %N.addr, align 4
   store float 3.000000e+00, float* %p, align 4
   store i32 7, i32* %N.addr, align 4
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, float*, i64)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nonnull %N.addr, float* nonnull %p, i64 4617315517961601024)
   ret void
 }
 
 define internal void @.omp_outlined.(i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) %N, float* dereferenceable(4) %p, i64 %q) {
 ; IS________OPM-LABEL: define {{[^@]+}}@.omp_outlined.
-; IS________OPM-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N:%.*]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
+; IS________OPM-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[Q_ADDR:%.*]] = alloca i64, align 8
 ; IS________OPM-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 ; IS________OPM-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 ; IS________OPM-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 ; IS________OPM-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 ; IS________OPM-NEXT:    store i64 4617315517961601024, i64* [[Q_ADDR]], align 8
 ; IS________OPM-NEXT:    [[CONV:%.*]] = bitcast i64* [[Q_ADDR]] to double*

 ; IS________OPM:       omp.loop.exit:
 ; IS________OPM-NEXT:    [[TMP12:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
 ; IS________OPM-NEXT:    call void @__kmpc_for_static_fini(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 [[TMP12]])
 ; IS________OPM-NEXT:    br label [[OMP_PRECOND_END]]
 ; IS________OPM:       omp.precond.end:
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@.omp_outlined.
-; IS________NPM-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[N:%.*]], float* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
+; IS________NPM-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N:%.*]], float* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[Q_ADDR:%.*]] = alloca i64, align 8
 ; IS________NPM-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
 ; IS________NPM-NEXT:    store i64 4617315517961601024, i64* [[Q_ADDR]], align 8
 ; IS________NPM-NEXT:    [[CONV:%.*]] = bitcast i64* [[Q_ADDR]] to double*

llvm/test/Transforms/Attributor/callbacks.ll

 ; IS__TUNIT_OPM-SAME: (i32* align 256 [[A:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:  entry:
 ; IS__TUNIT_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 undef, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 undef, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@t0_caller
 ; IS__TUNIT_NPM-SAME: (i32* align 256 [[A:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 undef, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 undef, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@t0_caller
 ; IS__CGSCC_OPM-SAME: (i32* align 256 [[A:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:  entry:
 ; IS__CGSCC_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 noundef 99, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 noundef 99, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
 ; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@t0_caller
 ; IS__CGSCC_NPM-SAME: (i32* align 256 [[A:%.*]]) {
 ; IS__CGSCC_NPM-NEXT:  entry:
 ; IS__CGSCC_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 noundef 99, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* align 256 [[A]], i64 noundef 99, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_NPM-NEXT:    ret void
 ;
 entry:
   %b = alloca i32, align 32
   %c = alloca i32*, align 64
   %ptr = alloca i32, align 128
   %0 = bitcast i32* %b to i8*
   store i32 42, i32* %b, align 4
   store i32* %b, i32** %c, align 8
   call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t0_callback_broker(i32* null, i32* %ptr, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32**)* @t0_callback_callee to void (i32*, i32*, ...)*), i32* %a, i64 99, i32** %c)
   ret void
 }
 
 ; Note that the first two arguments are provided by the callback_broker according to the callback in !1 below!
 ; The others are annotated with alignment information, amongst others, or even replaced by the constants passed to the call.
 define internal void @t0_callback_callee(i32* %is_not_null, i32* %ptr, i32* %a, i64 %b, i32** %c) {
 ;
 ; IS________OPM-LABEL: define {{[^@]+}}@t0_callback_callee
-; IS________OPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________OPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________OPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________OPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________OPM-NEXT:    tail call void @t0_check(i32* align 256 [[A]], i64 noundef 99, i32* [[TMP0]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@t0_callback_callee
-; IS________NPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________NPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________NPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________NPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________NPM-NEXT:    tail call void @t0_check(i32* align 256 [[A]], i64 noundef 99, i32* [[TMP0]])
 ; IS________NPM-NEXT:    ret void
 ;
 entry:

 ; IS__TUNIT_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:  entry:
 ; IS__TUNIT_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@t1_caller
 ; IS__TUNIT_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@t1_caller
 ; IS__CGSCC_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:  entry:
 ; IS__CGSCC_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
 ; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@t1_caller
 ; IS__CGSCC_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_NPM-NEXT:  entry:
 ; IS__CGSCC_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_NPM-NEXT:    ret void
 ;
 entry:
   %b = alloca i32, align 32
   %c = alloca i32*, align 64
   %ptr = alloca i32, align 128
   %0 = bitcast i32* %b to i8*
   store i32 42, i32* %b, align 4
   store i32* %b, i32** %c, align 8
   call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t1_callback_broker(i32* null, i32* %ptr, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32**)* @t1_callback_callee to void (i32*, i32*, ...)*), i32* %a, i64 99, i32** %c)
   ret void
 }
 
 ; Note that the first two arguments are provided by the callback_broker according to the callback in !1 below!
 ; The others are annotated with alignment information, amongst others, or even replaced by the constants passed to the call.
 define internal void @t1_callback_callee(i32* %is_not_null, i32* %ptr, i32* %a, i64 %b, i32** %c) {
 ;
 ; IS________OPM: Function Attrs: nosync
 ; IS________OPM-LABEL: define {{[^@]+}}@t1_callback_callee
-; IS________OPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
+; IS________OPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________OPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________OPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________OPM-NEXT:    tail call void @t1_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM: Function Attrs: nosync
 ; IS________NPM-LABEL: define {{[^@]+}}@t1_callback_callee
-; IS________NPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
+; IS________NPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________NPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________NPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________NPM-NEXT:    tail call void @t1_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________NPM-NEXT:    ret void
 ;
 entry:

 ; IS__TUNIT_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:  entry:
 ; IS__TUNIT_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@t2_caller
 ; IS__TUNIT_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@t2_caller
 ; IS__CGSCC_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:  entry:
 ; IS__CGSCC_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
 ; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@t2_caller
 ; IS__CGSCC_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_NPM-NEXT:  entry:
 ; IS__CGSCC_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_NPM-NEXT:    ret void
 ;
 entry:
   %b = alloca i32, align 32
   %c = alloca i32*, align 64
   %ptr = alloca i32, align 128
   %0 = bitcast i32* %b to i8*
   store i32 42, i32* %b, align 4
   store i32* %b, i32** %c, align 8
   call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t2_callback_broker(i32* null, i32* %ptr, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32**)* @t2_callback_callee to void (i32*, i32*, ...)*), i32* %a, i64 99, i32** %c)
   ret void
 }
 
 ; Note that the first two arguments are provided by the callback_broker according to the callback in !1 below!
 ; The others are annotated with alignment information, amongst others, or even replaced by the constants passed to the call.
 ;
 ; FIXME: We should derive noalias for %a and add a "fake use" of %a in all potentially synchronizing calls.
 define internal void @t2_callback_callee(i32* %is_not_null, i32* %ptr, i32* %a, i64 %b, i32** %c) {
 ;
 ; IS________OPM-LABEL: define {{[^@]+}}@t2_callback_callee
-; IS________OPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________OPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________OPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________OPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________OPM-NEXT:    tail call void @t2_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@t2_callback_callee
-; IS________NPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________NPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________NPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________NPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________NPM-NEXT:    tail call void @t2_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________NPM-NEXT:    ret void
 ;
 entry:

 ; IS__TUNIT_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_OPM-NEXT:  entry:
 ; IS__TUNIT_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
-; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@t3_caller
 ; IS__TUNIT_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__TUNIT_NPM-NEXT:  entry:
 ; IS__TUNIT_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__TUNIT_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__TUNIT_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__TUNIT_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__TUNIT_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__TUNIT_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
-; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__TUNIT_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 undef, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC_OPM-LABEL: define {{[^@]+}}@t3_caller
 ; IS__CGSCC_OPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_OPM-NEXT:  entry:
 ; IS__CGSCC_OPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_OPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_OPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_OPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_OPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_OPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
-; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_OPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_OPM-NEXT:    ret void
 ;
 ; IS__CGSCC_NPM-LABEL: define {{[^@]+}}@t3_caller
 ; IS__CGSCC_NPM-SAME: (i32* noalias nocapture align 256 [[A:%.*]]) {
 ; IS__CGSCC_NPM-NEXT:  entry:
 ; IS__CGSCC_NPM-NEXT:    [[B:%.*]] = alloca i32, align 32
 ; IS__CGSCC_NPM-NEXT:    [[C:%.*]] = alloca i32*, align 64
 ; IS__CGSCC_NPM-NEXT:    [[PTR:%.*]] = alloca i32, align 128
 ; IS__CGSCC_NPM-NEXT:    [[TMP0:%.*]] = bitcast i32* [[B]] to i8*
 ; IS__CGSCC_NPM-NEXT:    store i32 42, i32* [[B]], align 32
 ; IS__CGSCC_NPM-NEXT:    store i32* [[B]], i32** [[C]], align 64
-; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
-; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
+; IS__CGSCC_NPM-NEXT:    call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* noalias nocapture noundef align 536870912 null, i32* noalias nocapture noundef nonnull align 128 dereferenceable(4) [[PTR]], void (i32*, i32*, ...)* nocapture noundef bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* nocapture align 256 [[A]], i64 noundef 99, i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C]])
 ; IS__CGSCC_NPM-NEXT:    ret void
 ;
 entry:
   %b = alloca i32, align 32
   %c = alloca i32*, align 64
   %ptr = alloca i32, align 128
   %0 = bitcast i32* %b to i8*
   store i32 42, i32* %b, align 4
   store i32* %b, i32** %c, align 8
   call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* null, i32* %ptr, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* %a, i64 99, i32** %c)
   call void (i32*, i32*, void (i32*, i32*, ...)*, ...) @t3_callback_broker(i32* null, i32* %ptr, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32**)* @t3_callback_callee to void (i32*, i32*, ...)*), i32* %a, i64 99, i32** %c)
   ret void
 }
 
 ; Note that the first two arguments are provided by the callback_broker according to the callback in !1 below!
 ; The others are annotated with alignment information, amongst others, or even replaced by the constants passed to the call.
 ;
 ; FIXME: We should derive noalias for %a and add a "fake use" of %a in all potentially synchronizing calls.
 define internal void @t3_callback_callee(i32* %is_not_null, i32* %ptr, i32* %a, i64 %b, i32** %c) {
 ;
 ; IS________OPM-LABEL: define {{[^@]+}}@t3_callback_callee
-; IS________OPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________OPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________OPM-NEXT:  entry:
 ; IS________OPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________OPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________OPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________OPM-NEXT:    tail call void @t3_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@t3_callback_callee
-; IS________NPM-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
+; IS________NPM-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[IS_NOT_NULL:%.*]], i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[PTR:%.*]], i32* nocapture align 256 [[A:%.*]], i64 [[B:%.*]], i32** noalias nocapture nofree noundef nonnull readonly align 64 dereferenceable(8) [[C:%.*]]) {
 ; IS________NPM-NEXT:  entry:
 ; IS________NPM-NEXT:    [[PTR_VAL:%.*]] = load i32, i32* [[PTR]], align 8
 ; IS________NPM-NEXT:    store i32 [[PTR_VAL]], i32* [[IS_NOT_NULL]], align 4
 ; IS________NPM-NEXT:    [[TMP0:%.*]] = load i32*, i32** [[C]], align 64
 ; IS________NPM-NEXT:    tail call void @t3_check(i32* nocapture align 256 [[A]], i64 noundef 99, i32* nocapture [[TMP0]])
 ; IS________NPM-NEXT:    ret void
 ;
 entry:

llvm/test/Transforms/Attributor/heap_to_stack.ll

;
   store i8 %v, i8* %1
   tail call void @no_sync_func(i8* %1)
   tail call void @free(i8* nonnull dereferenceable(1) %1)
   ret void
 }
 
 define void @test16b(i8 %v, i8** %P) {
 ; CHECK-LABEL: define {{[^@]+}}@test16b
-; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]]) {
+; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture nofree writeonly [[P:%.*]]) {
 ; CHECK-NEXT:    [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 noundef 4)
 ; CHECK-NEXT:    store i8* [[TMP1]], i8** [[P]], align 8
 ; CHECK-NEXT:    tail call void @no_sync_func(i8* nocapture nofree [[TMP1]])
 ; CHECK-NEXT:    tail call void @free(i8* nocapture [[TMP1]])
 ; CHECK-NEXT:    ret void
 ;
   %1 = tail call noalias i8* @malloc(i64 4)
   store i8* %1, i8** %P
   tail call void @no_sync_func(i8* %1)
   tail call void @free(i8* %1)
   ret void
 }
 
 define void @test16c(i8 %v, i8** %P) {
 ; IS________OPM-LABEL: define {{[^@]+}}@test16c
-; IS________OPM-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]]) {
+; IS________OPM-SAME: (i8 [[V:%.*]], i8** nocapture nofree writeonly [[P:%.*]]) {
 ; IS________OPM-NEXT:    [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 noundef 4)
 ; IS________OPM-NEXT:    store i8* [[TMP1]], i8** [[P]], align 8
 ; IS________OPM-NEXT:    tail call void @no_sync_func(i8* nocapture nofree [[TMP1]]) #[[ATTR5]]
 ; IS________OPM-NEXT:    tail call void @free(i8* nocapture [[TMP1]])
 ; IS________OPM-NEXT:    ret void
 ;
 ; IS________NPM-LABEL: define {{[^@]+}}@test16c
-; IS________NPM-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]]) {
+; IS________NPM-SAME: (i8 [[V:%.*]], i8** nocapture nofree writeonly [[P:%.*]]) {
 ; IS________NPM-NEXT:    [[TMP1:%.*]] = alloca i8, i64 4, align 1
 ; IS________NPM-NEXT:    store i8* [[TMP1]], i8** [[P]], align 8
 ; IS________NPM-NEXT:    tail call void @no_sync_func(i8* nocapture nofree [[TMP1]]) #[[ATTR6]]
 ; IS________NPM-NEXT:    ret void
 ;
   %1 = tail call noalias i8* @malloc(i64 4)
   store i8* %1, i8** %P
   tail call void @no_sync_func(i8* %1) nounwind
   tail call void @free(i8* %1)
   ret void
 }
 
 define void @test16d(i8 %v, i8** %P) {
 ; CHECK-LABEL: define {{[^@]+}}@test16d
-; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture writeonly [[P:%.*]]) {
+; CHECK-SAME: (i8 [[V:%.*]], i8** nocapture nofree writeonly [[P:%.*]]) {
 ; CHECK-NEXT:    [[TMP1:%.*]] = tail call noalias i8* @malloc(i64 noundef 4)
 ; CHECK-NEXT:    store i8* [[TMP1]], i8** [[P]], align 8
 ; CHECK-NEXT:    ret void
 ;
   %1 = tail call noalias i8* @malloc(i64 4)
   store i8* %1, i8** %P
   ret void
 }

llvm/test/Transforms/Attributor/memory_locations.ll

 return:                                           ; preds = %if.end, %if.then
   %retval.0 = phi i8* [ %call, %if.then ], [ null, %if.end ]
   ret i8* %retval.0
 }
 
 define dso_local i8* @internal_argmem_only_read(i32* %arg) {
 ; CHECK: Function Attrs: inaccessiblemem_or_argmemonly
 ; CHECK-LABEL: define {{[^@]+}}@internal_argmem_only_read
-; CHECK-SAME: (i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1:[0-9]+]] {
+; CHECK-SAME: (i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP:%.*]] = load i32, i32* [[ARG]], align 4
 ; CHECK-NEXT:    [[CONV:%.*]] = sext i32 [[TMP]] to i64
 ; CHECK-NEXT:    [[CALL:%.*]] = call noalias i8* @malloc(i64 [[CONV]])
 ; CHECK-NEXT:    ret i8* [[CALL]]
 ;
 entry:
   %tmp = load i32, i32* %arg, align 4
   %conv = sext i32 %tmp to i64
   %call = call i8* @malloc(i64 %conv)
   ret i8* %call
 }
 
 define dso_local i8* @internal_argmem_only_write(i32* %arg) {
 ; CHECK: Function Attrs: inaccessiblemem_or_argmemonly
 ; CHECK-LABEL: define {{[^@]+}}@internal_argmem_only_write
-; CHECK-SAME: (i32* nocapture noundef nonnull writeonly align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
+; CHECK-SAME: (i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    store i32 10, i32* [[ARG]], align 4
 ; CHECK-NEXT:    [[CALL:%.*]] = call noalias dereferenceable_or_null(10) i8* @malloc(i64 noundef 10)
 ; CHECK-NEXT:    ret i8* [[CALL]]
 ;
 entry:
   store i32 10, i32* %arg, align 4
   %call = call dereferenceable_or_null(10) i8* @malloc(i64 10)
   ret i8* %call
 }
 
 define dso_local i8* @internal_argmem_only_rec(i32* %arg) {
 ; IS__TUNIT____: Function Attrs: inaccessiblemem_or_argmemonly
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@internal_argmem_only_rec
-; IS__TUNIT____-SAME: (i32* nocapture [[ARG:%.*]]) #[[ATTR1]] {
+; IS__TUNIT____-SAME: (i32* nocapture nofree [[ARG:%.*]]) #[[ATTR1]] {
 ; IS__TUNIT____-NEXT:  entry:
-; IS__TUNIT____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture align 4 [[ARG]])
+; IS__TUNIT____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture nofree align 4 [[ARG]])
 ; IS__TUNIT____-NEXT:    ret i8* [[CALL]]
 ;
 ; IS__CGSCC____: Function Attrs: inaccessiblemem_or_argmemonly
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@internal_argmem_only_rec
-; IS__CGSCC____-SAME: (i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
+; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
 ; IS__CGSCC____-NEXT:  entry:
-; IS__CGSCC____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ARG]])
+; IS__CGSCC____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ARG]])
 ; IS__CGSCC____-NEXT:    ret i8* [[CALL]]
 ;
 entry:
   %call = call i8* @internal_argmem_only_rec_1(i32* %arg)
   ret i8* %call
 }
 
 define internal i8* @internal_argmem_only_rec_1(i32* %arg) {
 ; IS__TUNIT____: Function Attrs: inaccessiblemem_or_argmemonly
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@internal_argmem_only_rec_1
-; IS__TUNIT____-SAME: (i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
+; IS__TUNIT____-SAME: (i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
 ; IS__TUNIT____-NEXT:  entry:
 ; IS__TUNIT____-NEXT:    [[TMP:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__TUNIT____-NEXT:    [[CMP:%.*]] = icmp eq i32 [[TMP]], 0
 ; IS__TUNIT____-NEXT:    br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
 ; IS__TUNIT____:       if.then:
 ; IS__TUNIT____-NEXT:    br label [[RETURN:%.*]]
 ; IS__TUNIT____:       if.end:
 ; IS__TUNIT____-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__TUNIT____-NEXT:    [[CMP1:%.*]] = icmp eq i32 [[TMP1]], 1
 ; IS__TUNIT____-NEXT:    br i1 [[CMP1]], label [[IF_THEN2:%.*]], label [[IF_END3:%.*]]
 ; IS__TUNIT____:       if.then2:
 ; IS__TUNIT____-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARG]], i64 -1
-; IS__TUNIT____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_2(i32* nocapture nonnull align 4 dereferenceable(4) [[ADD_PTR]])
+; IS__TUNIT____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_2(i32* nocapture nofree nonnull align 4 dereferenceable(4) [[ADD_PTR]])
 ; IS__TUNIT____-NEXT:    br label [[RETURN]]
 ; IS__TUNIT____:       if.end3:
 ; IS__TUNIT____-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__TUNIT____-NEXT:    [[CONV:%.*]] = sext i32 [[TMP2]] to i64
 ; IS__TUNIT____-NEXT:    [[CALL4:%.*]] = call noalias i8* @malloc(i64 [[CONV]])
 ; IS__TUNIT____-NEXT:    br label [[RETURN]]
 ; IS__TUNIT____:       return:
 ; IS__TUNIT____-NEXT:    [[RETVAL_0:%.*]] = phi i8* [ null, [[IF_THEN]] ], [ [[CALL]], [[IF_THEN2]] ], [ [[CALL4]], [[IF_END3]] ]
 ; IS__TUNIT____-NEXT:    ret i8* [[RETVAL_0]]
 ;
 ; IS__CGSCC____: Function Attrs: inaccessiblemem_or_argmemonly
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@internal_argmem_only_rec_1
-; IS__CGSCC____-SAME: (i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
+; IS__CGSCC____-SAME: (i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
 ; IS__CGSCC____-NEXT:  entry:
 ; IS__CGSCC____-NEXT:    [[TMP:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__CGSCC____-NEXT:    [[CMP:%.*]] = icmp eq i32 [[TMP]], 0
 ; IS__CGSCC____-NEXT:    br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
 ; IS__CGSCC____:       if.then:
 ; IS__CGSCC____-NEXT:    br label [[RETURN:%.*]]
 ; IS__CGSCC____:       if.end:
 ; IS__CGSCC____-NEXT:    [[TMP1:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__CGSCC____-NEXT:    [[CMP1:%.*]] = icmp eq i32 [[TMP1]], 1
 ; IS__CGSCC____-NEXT:    br i1 [[CMP1]], label [[IF_THEN2:%.*]], label [[IF_END3:%.*]]
 ; IS__CGSCC____:       if.then2:
 ; IS__CGSCC____-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARG]], i64 -1
-; IS__CGSCC____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_2(i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ADD_PTR]])
+; IS__CGSCC____-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_2(i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ADD_PTR]])
 ; IS__CGSCC____-NEXT:    br label [[RETURN]]
 ; IS__CGSCC____:       if.end3:
 ; IS__CGSCC____-NEXT:    [[TMP2:%.*]] = load i32, i32* [[ARG]], align 4
 ; IS__CGSCC____-NEXT:    [[CONV:%.*]] = sext i32 [[TMP2]] to i64
 ; IS__CGSCC____-NEXT:    [[CALL4:%.*]] = call noalias i8* @malloc(i64 [[CONV]])
 ; IS__CGSCC____-NEXT:    br label [[RETURN]]
 ; IS__CGSCC____:       return:
 ; IS__CGSCC____-NEXT:    [[RETVAL_0:%.*]] = phi i8* [ null, [[IF_THEN]] ], [ [[CALL]], [[IF_THEN2]] ], [ [[CALL4]], [[IF_END3]] ]

 return:                                           ; preds = %if.end3, %if.then2, %if.then
   %retval.0 = phi i8* [ null, %if.then ], [ %call, %if.then2 ], [ %call4, %if.end3 ]
   ret i8* %retval.0
 }
 
 define internal i8* @internal_argmem_only_rec_2(i32* %arg) {
 ; CHECK: Function Attrs: inaccessiblemem_or_argmemonly
 ; CHECK-LABEL: define {{[^@]+}}@internal_argmem_only_rec_2
-; CHECK-SAME: (i32* nocapture noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
+; CHECK-SAME: (i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[ARG:%.*]]) #[[ATTR1]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    store i32 0, i32* [[ARG]], align 4
 ; CHECK-NEXT:    [[ADD_PTR:%.*]] = getelementptr inbounds i32, i32* [[ARG]], i64 -1
-; CHECK-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture nonnull align 4 dereferenceable(4) [[ADD_PTR]])
+; CHECK-NEXT:    [[CALL:%.*]] = call noalias i8* @internal_argmem_only_rec_1(i32* nocapture nofree nonnull align 4 dereferenceable(4) [[ADD_PTR]])
 ; CHECK-NEXT:    ret i8* [[CALL]]
 ;
 entry:
   store i32 0, i32* %arg, align 4
   %add.ptr = getelementptr inbounds i32, i32* %arg, i64 -1
   %call = call i8* @internal_argmem_only_rec_1(i32* nonnull %add.ptr)
   ret i8* %call
 }

llvm/test/Transforms/Attributor/nocapture-1.ll

 declare i8* @maybe_returned_ptr(i8* readonly %ptr) readonly nounwind willreturn
 declare i8 @maybe_returned_val(i8* %ptr) readonly nounwind willreturn
 declare void @val_use(i8 %ptr) readonly nounwind willreturn
 
 ; FIXME: Both pointers should be nocapture
 define void @ptr_uses(i8* %ptr, i8* %wptr) {
 ; IS__TUNIT____: Function Attrs: nounwind willreturn
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@ptr_uses
-; IS__TUNIT____-SAME: (i8* [[PTR:%.*]], i8* nocapture noundef nonnull writeonly dereferenceable(1) [[WPTR:%.*]]) #[[ATTR12:[0-9]+]] {
+; IS__TUNIT____-SAME: (i8* [[PTR:%.*]], i8* nocapture nofree noundef nonnull writeonly dereferenceable(1) [[WPTR:%.*]]) #[[ATTR12:[0-9]+]] {
 ; IS__TUNIT____-NEXT:    store i8 0, i8* [[WPTR]], align 1
 ; IS__TUNIT____-NEXT:    ret void
 ;
 ; IS__CGSCC____: Function Attrs: nounwind willreturn
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@ptr_uses
-; IS__CGSCC____-SAME: (i8* [[PTR:%.*]], i8* nocapture noundef nonnull writeonly dereferenceable(1) [[WPTR:%.*]]) #[[ATTR14]] {
+; IS__CGSCC____-SAME: (i8* [[PTR:%.*]], i8* nocapture nofree noundef nonnull writeonly dereferenceable(1) [[WPTR:%.*]]) #[[ATTR14]] {
 ; IS__CGSCC____-NEXT:    store i8 0, i8* [[WPTR]], align 1
 ; IS__CGSCC____-NEXT:    ret void
 ;
   %call_ptr = call i8* @maybe_returned_ptr(i8* %ptr)
   %call_val = call i8 @maybe_returned_val(i8* %call_ptr)
   call void @val_use(i8 %call_val)
   store i8 0, i8* %wptr
   ret void

llvm/test/Transforms/Attributor/nofree.ll

 
 declare double @cos(double) nobuiltin nounwind nofree
 
 ; FIXME: %a should be nofree.
 ; TEST 13 NoFree argument - positive.
 define noalias i32* @test13(i64* nocapture readonly %a) {
 ; CHECK: Function Attrs: nounwind
 ; CHECK-LABEL: define {{[^@]+}}@test13
-; CHECK-SAME: (i64* nocapture noundef nonnull readonly align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR0]] {
+; CHECK-SAME: (i64* nocapture nofree noundef nonnull readonly align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i64, i64* [[A]], align 8
 ; CHECK-NEXT:    [[CALL:%.*]] = tail call noalias i8* @malloc(i64 [[TMP0]]) #[[ATTR2]]
 ; CHECK-NEXT:    [[TMP1:%.*]] = bitcast i8* [[CALL]] to i32*
 ; CHECK-NEXT:    ret i32* [[TMP1]]
 ;
 entry:
   %0 = load i64, i64* %a, align 8

llvm/test/Transforms/Attributor/nonnull.ll

f:
   ret void
 }
 
 ; The callsite must execute in order for the attribute to transfer to the parent.
 ; The volatile load can't trap, so we can guarantee that we'll get to the call.
 
 define i8 @parent6(i8* %a, i8* %b) {
 ; CHECK-LABEL: define {{[^@]+}}@parent6
-; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* noundef [[B:%.*]]) {
+; CHECK-SAME: (i8* nonnull [[A:%.*]], i8* nofree noundef [[B:%.*]]) {
 ; CHECK-NEXT:    [[C:%.*]] = load volatile i8, i8* [[B]], align 1
 ; CHECK-NEXT:    call void @use1nonnull(i8* nonnull [[A]])
 ; CHECK-NEXT:    ret i8 [[C]]
 ;
 
   %c = load volatile i8, i8* %b
   call void @use1nonnull(i8* %a)
   ret i8 %c

llvm/test/Transforms/Attributor/norecurse.ll

 ; CHECK: Function Attrs
 ; CHECK-NEXT: declare i32 @k()
 declare i32 @k() readnone
 
 define void @intrinsic(i8* %dest, i8* %src, i32 %len) {
 ; CHECK: Function Attrs: argmemonly nofree nosync nounwind willreturn
 ; CHECK-LABEL: define {{[^@]+}}@intrinsic
 ; CHECK-SAME: (i8* nocapture nofree writeonly [[DEST:%.*]], i8* nocapture nofree readonly [[SRC:%.*]], i32 [[LEN:%.*]]) #[[ATTR5:[0-9]+]] {
-; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* noalias nocapture nofree writeonly [[DEST]], i8* noalias nocapture nofree readonly [[SRC]], i32 [[LEN]], i1 noundef false) #[[ATTR10:[0-9]+]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i32(i8* noalias nocapture nofree writeonly [[DEST]], i8* noalias nocapture nofree readonly [[SRC]], i32 [[LEN]], i1 noundef false) #[[ATTR11:[0-9]+]]
 ; CHECK-NEXT:    ret void
 ;
   call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 %len, i1 false)
   ret void
 }
 
 ; CHECK: Function Attrs
 ; CHECK-NEXT: declare void @llvm.memcpy.p0i8.p0i8.i32
 declare void @llvm.memcpy.p0i8.p0i8.i32(i8*, i8*, i32, i1)
 
 define internal i32 @called_by_norecurse() {
 ; IS__TUNIT____: Function Attrs: nosync readnone
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@called_by_norecurse
 ; IS__TUNIT____-SAME: () #[[ATTR3]] {
 ; IS__TUNIT____-NEXT:    [[A:%.*]] = call i32 @k()
 ; IS__TUNIT____-NEXT:    ret i32 undef
 ;
 ; IS__CGSCC____: Function Attrs: norecurse nosync readnone
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@called_by_norecurse
-; IS__CGSCC____-SAME: () #[[ATTR6:[0-9]+]] {
+; IS__CGSCC____-SAME: () #[[ATTR7:[0-9]+]] {
 ; IS__CGSCC____-NEXT:    [[A:%.*]] = call i32 @k()
 ; IS__CGSCC____-NEXT:    ret i32 undef
 ;
   %a = call i32 @k()
   ret i32 %a
 }
 define void @m() norecurse {
 ; IS__TUNIT____: Function Attrs: norecurse nosync readnone
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@m
-; IS__TUNIT____-SAME: () #[[ATTR6:[0-9]+]] {
+; IS__TUNIT____-SAME: () #[[ATTR7:[0-9]+]] {
 ; IS__TUNIT____-NEXT:    [[A:%.*]] = call i32 @called_by_norecurse() #[[ATTR3]]
 ; IS__TUNIT____-NEXT:    ret void
 ;
 ; IS__CGSCC____: Function Attrs: norecurse nosync readnone
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@m
-; IS__CGSCC____-SAME: () #[[ATTR6]] {
+; IS__CGSCC____-SAME: () #[[ATTR7]] {
 ; IS__CGSCC____-NEXT:    [[A:%.*]] = call i32 @called_by_norecurse()
 ; IS__CGSCC____-NEXT:    ret void
 ;
   %a = call i32 @called_by_norecurse()
   ret void
 }
 
 define internal i32 @called_by_norecurse_indirectly() {

 ; IS__TUNIT____: Function Attrs: nosync readnone
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@o
 ; IS__TUNIT____-SAME: () #[[ATTR3]] {
 ; IS__TUNIT____-NEXT:    [[A:%.*]] = call i32 @called_by_norecurse_indirectly() #[[ATTR3]]
 ; IS__TUNIT____-NEXT:    ret i32 [[A]]
 ;
 ; IS__CGSCC____: Function Attrs: norecurse nosync readnone
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@o
-; IS__CGSCC____-SAME: () #[[ATTR6]] {
+; IS__CGSCC____-SAME: () #[[ATTR7]] {
 ; IS__CGSCC____-NEXT:    [[A:%.*]] = call i32 @called_by_norecurse_indirectly()
 ; IS__CGSCC____-NEXT:    ret i32 [[A]]
 ;
   %a = call i32 @called_by_norecurse_indirectly()
   ret i32 %a
 }
 define i32 @p() norecurse {
 ; IS__TUNIT____: Function Attrs: norecurse nosync readnone
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@p
-; IS__TUNIT____-SAME: () #[[ATTR6]] {
+; IS__TUNIT____-SAME: () #[[ATTR7]] {
 ; IS__TUNIT____-NEXT:    [[A:%.*]] = call i32 @o() #[[ATTR3]]
 ; IS__TUNIT____-NEXT:    ret i32 [[A]]
 ;
 ; IS__CGSCC____: Function Attrs: norecurse nosync readnone
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@p
-; IS__CGSCC____-SAME: () #[[ATTR6]] {
+; IS__CGSCC____-SAME: () #[[ATTR7]] {
 ; IS__CGSCC____-NEXT:    [[A:%.*]] = call i32 @o()
 ; IS__CGSCC____-NEXT:    ret i32 [[A]]
 ;
   %a = call i32 @o()
   ret i32 %a
 }
 
 define void @f(i32 %x)  {
 ; CHECK: Function Attrs: nofree nosync nounwind readnone
 ; CHECK-LABEL: define {{[^@]+}}@f
-; CHECK-SAME: (i32 [[X:%.*]]) #[[ATTR7:[0-9]+]] {
+; CHECK-SAME: (i32 [[X:%.*]]) #[[ATTR8:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[X_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[X]], i32* [[X_ADDR]], align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[X_ADDR]], align 4
 ; CHECK-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP0]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
 ; CHECK:       if.then:
-; CHECK-NEXT:    call void @g() #[[ATTR8:[0-9]+]]
+; CHECK-NEXT:    call void @g() #[[ATTR9:[0-9]+]]
 ; CHECK-NEXT:    br label [[IF_END]]
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
 ;
 entry:
   %x.addr = alloca i32, align 4
   store i32 %x, i32* %x.addr, align 4
   %0 = load i32, i32* %x.addr, align 4
   %tobool = icmp ne i32 %0, 0
   br i1 %tobool, label %if.then, label %if.end
 
 if.then:
   call void @g() norecurse
   br label %if.end
 
 if.end:
   ret void
 }
 
 define void @g() norecurse {
 ; CHECK: Function Attrs: nofree norecurse nosync nounwind readnone
 ; CHECK-LABEL: define {{[^@]+}}@g
-; CHECK-SAME: () #[[ATTR8]] {
+; CHECK-SAME: () #[[ATTR9]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @f(i32 noundef 0) #[[ATTR7]]
+; CHECK-NEXT:    call void @f(i32 noundef 0) #[[ATTR8]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
   call void @f(i32 0)
   ret void
 }
 
 define linkonce_odr i32 @leaf_redefinable() {

 ;
   %3 = tail call i32 %0(i32 %1) #2
   ret i32 %3
 }
 
 define i32 @eval_func2(i32 (i32)* , i32) local_unnamed_addr null_pointer_is_valid{
 ; CHECK: Function Attrs: null_pointer_is_valid
 ; CHECK-LABEL: define {{[^@]+}}@eval_func2
-; CHECK-SAME: (i32 (i32)* nocapture nofree noundef [[TMP0:%.*]], i32 [[TMP1:%.*]]) local_unnamed_addr #[[ATTR9:[0-9]+]] {
+; CHECK-SAME: (i32 (i32)* nocapture nofree noundef [[TMP0:%.*]], i32 [[TMP1:%.*]]) local_unnamed_addr #[[ATTR10:[0-9]+]] {
 ; CHECK-NEXT:    [[TMP3:%.*]] = tail call i32 [[TMP0]](i32 [[TMP1]])
 ; CHECK-NEXT:    ret i32 [[TMP3]]
 ;
   %3 = tail call i32 %0(i32 %1) #2
   ret i32 %3
 }
 
 ; Call an unknown function in a dead block.

 
 ;.
 ; IS__TUNIT____: attributes #[[ATTR0]] = { nofree nosync nounwind readnone willreturn }
 ; IS__TUNIT____: attributes #[[ATTR1]] = { nofree noreturn nosync nounwind readnone willreturn }
 ; IS__TUNIT____: attributes #[[ATTR2]] = { nofree noreturn nosync nounwind readnone }
 ; IS__TUNIT____: attributes #[[ATTR3]] = { nosync readnone }
 ; IS__TUNIT____: attributes #[[ATTR4:[0-9]+]] = { readnone }
 ; IS__TUNIT____: attributes #[[ATTR5]] = { argmemonly nofree nosync nounwind willreturn }
-; IS__TUNIT____: attributes #[[ATTR6]] = { norecurse nosync readnone }
-; IS__TUNIT____: attributes #[[ATTR7]] = { nofree nosync nounwind readnone }
-; IS__TUNIT____: attributes #[[ATTR8]] = { nofree norecurse nosync nounwind readnone }
-; IS__TUNIT____: attributes #[[ATTR9]] = { null_pointer_is_valid }
-; IS__TUNIT____: attributes #[[ATTR10]] = { willreturn }
+; IS__TUNIT____: attributes #[[ATTR6:[0-9]+]] = { argmemonly nofree nounwind willreturn }
+; IS__TUNIT____: attributes #[[ATTR7]] = { norecurse nosync readnone }
+; IS__TUNIT____: attributes #[[ATTR8]] = { nofree nosync nounwind readnone }
+; IS__TUNIT____: attributes #[[ATTR9]] = { nofree norecurse nosync nounwind readnone }
+; IS__TUNIT____: attributes #[[ATTR10]] = { null_pointer_is_valid }
+; IS__TUNIT____: attributes #[[ATTR11]] = { willreturn }
 ;.
 ; IS__CGSCC____: attributes #[[ATTR0]] = { nofree norecurse nosync nounwind readnone willreturn }
 ; IS__CGSCC____: attributes #[[ATTR1]] = { nofree norecurse noreturn nosync nounwind readnone willreturn }
 ; IS__CGSCC____: attributes #[[ATTR2]] = { nofree noreturn nosync nounwind readnone }
 ; IS__CGSCC____: attributes #[[ATTR3]] = { nosync readnone }
 ; IS__CGSCC____: attributes #[[ATTR4:[0-9]+]] = { readnone }
 ; IS__CGSCC____: attributes #[[ATTR5]] = { argmemonly nofree nosync nounwind willreturn }
-; IS__CGSCC____: attributes #[[ATTR6]] = { norecurse nosync readnone }
-; IS__CGSCC____: attributes #[[ATTR7]] = { nofree nosync nounwind readnone }
-; IS__CGSCC____: attributes #[[ATTR8]] = { nofree norecurse nosync nounwind readnone }
-; IS__CGSCC____: attributes #[[ATTR9]] = { null_pointer_is_valid }
-; IS__CGSCC____: attributes #[[ATTR10]] = { willreturn }
+; IS__CGSCC____: attributes #[[ATTR6:[0-9]+]] = { argmemonly nofree nounwind willreturn }
+; IS__CGSCC____: attributes #[[ATTR7]] = { norecurse nosync readnone }
+; IS__CGSCC____: attributes #[[ATTR8]] = { nofree nosync nounwind readnone }
+; IS__CGSCC____: attributes #[[ATTR9]] = { nofree norecurse nosync nounwind readnone }
+; IS__CGSCC____: attributes #[[ATTR10]] = { null_pointer_is_valid }
+; IS__CGSCC____: attributes #[[ATTR11]] = { willreturn }
 ;.

llvm/test/Transforms/Attributor/range.ll

t:
   ret i1 %r
 f:
   ret i1 false
 }
 
 
 define void @spam(i32* %arg, i32* %arg1) {
 ; CHECK-LABEL: define {{[^@]+}}@spam
-; CHECK-SAME: (i32* nocapture noundef nonnull readonly align 8 dereferenceable(4) [[ARG:%.*]], i32* nocapture nofree readnone [[ARG1:%.*]]) {
+; CHECK-SAME: (i32* nocapture nofree noundef nonnull readonly align 8 dereferenceable(4) [[ARG:%.*]], i32* nocapture nofree readnone [[ARG1:%.*]]) {
 ; CHECK-NEXT:  bb:
 ; CHECK-NEXT:    [[TMP:%.*]] = load i32, i32* [[ARG]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ult i32 [[TMP]], 4
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[BB3:%.*]], label [[BB4:%.*]]
 ; CHECK:       bb3:
 ; CHECK-NEXT:    call fastcc void @wobble(i32 signext [[TMP]])
 ; CHECK-NEXT:    br label [[BB5:%.*]]
 ; CHECK:       bb4:

llvm/test/Transforms/Attributor/readattrs.ll

;
   store i8* %q, i8** %p
   ret void
 }
 
 declare void @test6_1()
 ; This is not a missed optz'n.
 define void @test6_2(i8** %p, i8* %q) {
 ; CHECK-LABEL: define {{[^@]+}}@test6_2
-; CHECK-SAME: (i8** nocapture noundef nonnull writeonly align 8 dereferenceable(8) [[P:%.*]], i8* [[Q:%.*]]) {
+; CHECK-SAME: (i8** nocapture nofree noundef nonnull writeonly align 8 dereferenceable(8) [[P:%.*]], i8* nofree [[Q:%.*]]) {
 ; CHECK-NEXT:    store i8* [[Q]], i8** [[P]], align 8
 ; CHECK-NEXT:    call void @test6_1()
 ; CHECK-NEXT:    ret void
 ;
   store i8* %q, i8** %p
   call void @test6_1()
   ret void
 }

llvm/test/Transforms/Attributor/value-simplify.ll

;
   ret void
 }
 ; FIXME: Alignment and dereferenceability are not propagated to the argument
 define void @complicated_args_sret(%struct.X** %b) {
 ;
 ; IS__TUNIT_OPM: Function Attrs: argmemonly nofree nosync nounwind willreturn writeonly
 ; IS__TUNIT_OPM-LABEL: define {{[^@]+}}@complicated_args_sret
 ; IS__TUNIT_OPM-SAME: (%struct.X** nocapture nofree writeonly [[B:%.*]]) #[[ATTR3]] {
-; IS__TUNIT_OPM-NEXT:    call void @test_sret(%struct.X* noalias nocapture nofree noundef writeonly sret(%struct.X) align 536870912 null, %struct.X** nocapture nofree writeonly align 8 [[B]]) #[[ATTR6:[0-9]+]]
+; IS__TUNIT_OPM-NEXT:    call void @test_sret(%struct.X* noalias nocapture nofree noundef writeonly sret([[STRUCT_X:%.*]]) align 536870912 null, %struct.X** nocapture nofree writeonly align 8 [[B]]) #[[ATTR6:[0-9]+]]
 ; IS__TUNIT_OPM-NEXT:    ret void
 ;
 ; IS__TUNIT_NPM: Function Attrs: argmemonly nofree nosync nounwind willreturn writeonly
 ; IS__TUNIT_NPM-LABEL: define {{[^@]+}}@complicated_args_sret
 ; IS__TUNIT_NPM-SAME: (%struct.X** nocapture nofree writeonly [[B:%.*]]) #[[ATTR3]] {
-; IS__TUNIT_NPM-NEXT:    call void @test_sret(%struct.X* noalias nocapture nofree noundef writeonly sret(%struct.X) align 536870912 null, %struct.X** nocapture nofree writeonly align 8 [[B]]) #[[ATTR5:[0-9]+]]
+; IS__TUNIT_NPM-NEXT:    call void @test_sret(%struct.X* noalias nocapture nofree noundef writeonly sret([[STRUCT_X:%.*]]) align 536870912 null, %struct.X** nocapture nofree writeonly align 8 [[B]]) #[[ATTR5:[0-9]+]]
 ; IS__TUNIT_NPM-NEXT:    ret void
 ;
 ; IS__CGSCC____: Function Attrs: argmemonly nofree norecurse nosync nounwind willreturn writeonly
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@complicated_args_sret
 ; IS__CGSCC____-SAME: (%struct.X** nocapture nofree noundef nonnull writeonly align 8 dereferenceable(8) [[B:%.*]]) #[[ATTR2]] {
 ; IS__CGSCC____-NEXT:    unreachable
 ;
   call void @test_sret(%struct.X* sret(%struct.X) null, %struct.X** %b)

llvm/test/Transforms/Attributor/willreturn.ll

 ;     if(*p){
 ;       exit(1);
 ;     }
 ;     return;
 ; }
 define void @conditional_exit(i32 %0, i32* nocapture readonly %1) local_unnamed_addr #0 {
 ; IS__TUNIT____: Function Attrs: noinline nounwind uwtable
 ; IS__TUNIT____-LABEL: define {{[^@]+}}@conditional_exit
-; IS__TUNIT____-SAME: (i32 [[TMP0:%.*]], i32* nocapture readonly [[TMP1:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] {
+; IS__TUNIT____-SAME: (i32 [[TMP0:%.*]], i32* nocapture nofree readonly [[TMP1:%.*]]) local_unnamed_addr #[[ATTR6:[0-9]+]] {
 ; IS__TUNIT____-NEXT:    [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
 ; IS__TUNIT____-NEXT:    br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
 ; IS__TUNIT____:       4:
 ; IS__TUNIT____-NEXT:    tail call void @exit(i32 noundef 0) #[[ATTR4]]
 ; IS__TUNIT____-NEXT:    unreachable
 ; IS__TUNIT____:       5:
 ; IS__TUNIT____-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4
 ; IS__TUNIT____-NEXT:    [[TMP7:%.*]] = icmp eq i32 [[TMP6]], 0
 ; IS__TUNIT____-NEXT:    br i1 [[TMP7]], label [[TMP9:%.*]], label [[TMP8:%.*]]
 ; IS__TUNIT____:       8:
 ; IS__TUNIT____-NEXT:    tail call void @exit(i32 noundef 1) #[[ATTR4]]
 ; IS__TUNIT____-NEXT:    unreachable
 ; IS__TUNIT____:       9:
 ; IS__TUNIT____-NEXT:    ret void
 ;
 ; IS__CGSCC____: Function Attrs: noinline nounwind uwtable
 ; IS__CGSCC____-LABEL: define {{[^@]+}}@conditional_exit
-; IS__CGSCC____-SAME: (i32 [[TMP0:%.*]], i32* nocapture readonly [[TMP1:%.*]]) local_unnamed_addr #[[ATTR7:[0-9]+]] {
+; IS__CGSCC____-SAME: (i32 [[TMP0:%.*]], i32* nocapture nofree readonly [[TMP1:%.*]]) local_unnamed_addr #[[ATTR7:[0-9]+]] {
 ; IS__CGSCC____-NEXT:    [[TMP3:%.*]] = icmp eq i32 [[TMP0]], 0
 ; IS__CGSCC____-NEXT:    br i1 [[TMP3]], label [[TMP5:%.*]], label [[TMP4:%.*]]
 ; IS__CGSCC____:       4:
 ; IS__CGSCC____-NEXT:    tail call void @exit(i32 noundef 0) #[[ATTR5]]
 ; IS__CGSCC____-NEXT:    unreachable
 ; IS__CGSCC____:       5:
 ; IS__CGSCC____-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP1]], align 4
 ; IS__CGSCC____-NEXT:    [[TMP7:%.*]] = icmp eq i32 [[TMP6]], 0

llvm/test/Transforms/OpenMP/parallel_deletion.ll

 ;    #pragma omp parallel
 ;      {}
 ;    }
 ;
 ; We delete all but the first of the parallel regions in this test.
 define void @delete_parallel_0() {
 ; CHECK-LABEL: define {{[^@]+}}@delete_parallel_0() {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0:@.*]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined.willreturn to void (i32*, i32*, ...)*))
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0:[0-9]+]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined.willreturn to void (i32*, i32*, ...)*))
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@delete_parallel_0() {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0:[0-9]+]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined.willreturn to void (i32*, i32*, ...)*))
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@delete_parallel_0() {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0:[0-9]+]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined.willreturn to void (i32*, i32*, ...)*))
+; CHECK2-NEXT:    ret void
 entry:
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined.willreturn to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined.willreturn.0 to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined.willreturn.1 to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined.willreturn.2 to void (i32*, i32*, ...)*))
   ret void
 }
 
 define internal void @.omp_outlined.willreturn(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.willreturn
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR0:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR0:[0-9]+]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @unknown() [[ATTR0]]
+; CHECK-NEXT:    call void @unknown() #[[ATTR0]]
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.willreturn
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @unknown() #[[ATTR0]]
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.willreturn
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @unknown() #[[ATTR0]]
+; CHECK2-NEXT:    ret void
 entry:
   call void @unknown() willreturn
   ret void
 }
 
 define internal void @.omp_outlined.willreturn.0(i32* noalias %.global_tid., i32* noalias %.bound_tid.) willreturn {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.0
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR1:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @readonly() [[ATTR4:#.*]]
+; CHECK-NEXT:    call void @readonly() #[[ATTR4:[0-9]+]]
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.0
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @readonly() #[[ATTR4:[0-9]+]]
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.0
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @readonly() #[[ATTR4:[0-9]+]]
+; CHECK2-NEXT:    ret void
 entry:
   call void @readonly()
   ret void
 }
 
 define internal void @.omp_outlined.willreturn.1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.1
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR2:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @readnone() [[ATTR0]]
+; CHECK-NEXT:    call void @readnone() #[[ATTR0]]
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.1
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @readnone() #[[ATTR0]]
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.1
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR2:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @readnone() #[[ATTR0]]
+; CHECK2-NEXT:    ret void
 entry:
   call void @readnone() willreturn
   ret void
 }
 
 define internal void @.omp_outlined.willreturn.2(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.2
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR3:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.2
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.willreturn.2
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    ret void
 entry:
   ret void
 }
 
 ;    void delete_parallel_1(void) {
 ;    #pragma omp parallel
 ;      { unknown(); }
 ;    #pragma omp parallel
 ;      { readonly(); }
 ;    #pragma omp parallel
 ;      { readnone(); }
 ;    #pragma omp parallel
 ;      {}
 ;    }
 ;
 ; We delete only the last parallel regions in this test because the others might not return.
 define void @delete_parallel_1() {
 ; CHECK-LABEL: define {{[^@]+}}@delete_parallel_1() {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..0 to void (i32*, i32*, ...)*))
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
-; CHECK-NEXT:    ret void
-;
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..0 to void (i32*, i32*, ...)*))
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
+; CHECK-NEXT:    ret void
+;
+; CHECK1-LABEL: define {{[^@]+}}@delete_parallel_1() {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..0 to void (i32*, i32*, ...)*))
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@delete_parallel_1() {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..0 to void (i32*, i32*, ...)*))
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 0, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
+; CHECK2-NEXT:    ret void
 entry:
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..0 to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*))
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 0, void (i32*, i32*, ...)* bitcast (void (i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*))
   ret void
 }
 
 define internal void @.omp_outlined.(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.
 ; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    call void @unknown()
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @unknown()
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @unknown()
+; CHECK2-NEXT:    ret void
 entry:
   call void @unknown()
   ret void
 }
 
 define internal void @.omp_outlined..0(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..0
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR4]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    call void @readonly() [[ATTR4]]
+; CHECK-NEXT:    call void @readonly() #[[ATTR4]]
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..0
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @readonly() #[[ATTR4]]
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..0
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR4]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @readonly() #[[ATTR4]]
+; CHECK2-NEXT:    ret void
 entry:
   call void @readonly()
   ret void
 }
 
 define internal void @.omp_outlined..1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..1
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR5:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR5:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    call void @readnone()
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR5:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    call void @readnone()
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR5:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    call void @readnone()
+; CHECK2-NEXT:    ret void
 entry:
   call void @readnone()
   ret void
 }
 
 define internal void @.omp_outlined..2(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..2
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) [[ATTR3]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]]) #[[ATTR3]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    ret void
 entry:
   ret void
 }
 
 ;    void delete_parallel_2(void) {
 ;      int a = 0;
 ;    #pragma omp parallel
 ;      {

 ;
 ; FIXME: We do not realize that `a` is dead and all accesses to it can be removed
 ;        making the parallel regions readonly and deletable.
 define void @delete_parallel_2() {
 ; CHECK-LABEL: define {{[^@]+}}@delete_parallel_2() {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP:%.*]] = bitcast i32* [[A]] to i8*
-; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 dereferenceable(4) [[TMP]]) [[ATTR0]]
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 dereferenceable(4) [[TMP]]) #[[ATTR0]]
 ; CHECK-NEXT:    store i32 0, i32* [[A]], align 4
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
 ; CHECK-NEXT:    [[TMP1:%.*]] = bitcast i32* [[A]] to i8*
 ; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP1]])
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@delete_parallel_2() {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP:%.*]] = bitcast i32* [[A]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 dereferenceable(4) [[TMP]]) #[[ATTR0]]
+; CHECK1-NEXT:    store i32 0, i32* [[A]], align 4
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = bitcast i32* [[A]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP1]])
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@delete_parallel_2() {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP:%.*]] = bitcast i32* [[A]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 dereferenceable(4) [[TMP]]) #[[ATTR0]]
+; CHECK2-NEXT:    store i32 0, i32* [[A]], align 4
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = bitcast i32* [[A]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP1]])
+; CHECK2-NEXT:    ret void
 entry:
   %a = alloca i32, align 4
   %tmp = bitcast i32* %a to i8*
   call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %tmp)
   store i32 0, i32* %a, align 4
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nonnull %a)
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nonnull %a)
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nonnull %a)
   call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @0, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nonnull %a)
   %tmp1 = bitcast i32* %a to i8*
   call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %tmp1)
   ret void
 }
 
 define internal void @.omp_outlined..3(i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) %a) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..3
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) [[ATTR6:#.*]] {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR6:[0-9]+]] {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @omp_get_thread_num() [[ATTR12:#.*]]
+; CHECK-NEXT:    [[CALL:%.*]] = call i32 @omp_get_thread_num() #[[ATTR14:[0-9]+]]
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
 ; CHECK-NEXT:    br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[TMP:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP]], 1
 ; CHECK-NEXT:    store i32 [[INC]], i32* [[A]], align 4
 ; CHECK-NEXT:    br label [[IF_END]]
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR6:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[CALL:%.*]] = call i32 @omp_get_thread_num() #[[ATTR14:[0-9]+]]
+; CHECK1-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
+; CHECK1-NEXT:    br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
+; CHECK1:       if.then:
+; CHECK1-NEXT:    [[TMP:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP]], 1
+; CHECK1-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK1-NEXT:    br label [[IF_END]]
+; CHECK1:       if.end:
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR6:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[CALL:%.*]] = call i32 @omp_get_thread_num() #[[ATTR14:[0-9]+]]
+; CHECK2-NEXT:    [[CMP:%.*]] = icmp eq i32 [[CALL]], 0
+; CHECK2-NEXT:    br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
+; CHECK2:       if.then:
+; CHECK2-NEXT:    [[TMP:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP]], 1
+; CHECK2-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK2-NEXT:    br label [[IF_END]]
+; CHECK2:       if.end:
+; CHECK2-NEXT:    ret void
 entry:
   %call = call i32 @omp_get_thread_num()
   %cmp = icmp eq i32 %call, 0
   br i1 %cmp, label %if.then, label %if.end
 
 if.then:                                          ; preds = %entry
   %tmp = load i32, i32* %a, align 4
   %inc = add nsw i32 %tmp, 1
   store i32 %inc, i32* %a, align 4
   br label %if.end
 
 if.end:                                           ; preds = %if.then, %entry
   ret void
 }
 
 define internal void @.omp_outlined..4(i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) %a) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..4
-; CHECK-SAME: (i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* noundef nonnull [[GLOB0]], i32 [[TMP]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
 ; CHECK:       omp_if.then:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
 ; CHECK-NEXT:    store i32 [[INC]], i32* [[A]], align 4
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* noundef nonnull [[GLOB0]], i32 [[TMP]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
 ; CHECK-NEXT:    br label [[OMP_IF_END]]
 ; CHECK:       omp_if.end:
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK1-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
+; CHECK1:       omp_if.then:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK1-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK1-NEXT:    br label [[OMP_IF_END]]
+; CHECK1:       omp_if.end:
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK2-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
+; CHECK2:       omp_if.then:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK2-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK2-NEXT:    br label [[OMP_IF_END]]
+; CHECK2:       omp_if.end:
+; CHECK2-NEXT:    ret void
 entry:
   %tmp = load i32, i32* %.global_tid., align 4
   %tmp1 = call i32 @__kmpc_master(%struct.ident_t* nonnull @0, i32 %tmp)
   %tmp2 = icmp eq i32 %tmp1, 0
   br i1 %tmp2, label %omp_if.end, label %omp_if.then
 
 omp_if.then:                                      ; preds = %entry
   %tmp3 = load i32, i32* %a, align 4

 declare i32 @omp_get_thread_num() inaccessiblememonly nofree nosync nounwind readonly
 
 declare i32 @__kmpc_master(%struct.ident_t*, i32)
 
 declare void @__kmpc_end_master(%struct.ident_t*, i32)
 
 define internal void @.omp_outlined..5(i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) %a) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..5
-; CHECK-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull [[GLOB0]]) [[ATTR12]]
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @[[GLOB0]]) #[[ATTR14]]
 ; CHECK-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_single(%struct.ident_t* noundef nonnull [[GLOB0]], i32 [[TMP]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
 ; CHECK:       omp_if.then:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
 ; CHECK-NEXT:    store i32 [[INC]], i32* [[A]], align 4
-; CHECK-NEXT:    call void @__kmpc_end_single(%struct.ident_t* noundef nonnull [[GLOB0]], i32 [[TMP]])
+; CHECK-NEXT:    call void @__kmpc_end_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
 ; CHECK-NEXT:    br label [[OMP_IF_END]]
 ; CHECK:       omp_if.end:
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* noundef nonnull [[GLOB1:@.*]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* noundef nonnull @[[GLOB1:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK1-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @[[GLOB0]]) #[[ATTR14]]
+; CHECK1-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
+; CHECK1:       omp_if.then:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK1-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK1-NEXT:    call void @__kmpc_end_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK1-NEXT:    br label [[OMP_IF_END]]
+; CHECK1:       omp_if.end:
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* noundef nonnull @[[GLOB1:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK2-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @[[GLOB0]]) #[[ATTR14]]
+; CHECK2-NEXT:    [[TMP:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_IF_END:%.*]], label [[OMP_IF_THEN:%.*]]
+; CHECK2:       omp_if.then:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK2-NEXT:    store i32 [[INC]], i32* [[A]], align 4
+; CHECK2-NEXT:    call void @__kmpc_end_single(%struct.ident_t* noundef nonnull @[[GLOB0]], i32 [[TMP]])
+; CHECK2-NEXT:    br label [[OMP_IF_END]]
+; CHECK2:       omp_if.end:
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* noundef nonnull @[[GLOB1:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    ret void
 entry:
   %omp_global_thread_num = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @0)
   %tmp = load i32, i32* %.global_tid., align 4
   %tmp1 = call i32 @__kmpc_single(%struct.ident_t* nonnull @0, i32 %tmp)
   %tmp2 = icmp eq i32 %tmp1, 0
   br i1 %tmp2, label %omp_if.end, label %omp_if.then
 
 omp_if.then:                                      ; preds = %entry
   %tmp3 = load i32, i32* %a, align 4
   %inc = add nsw i32 %tmp3, 1
   store i32 %inc, i32* %a, align 4
   call void @__kmpc_end_single(%struct.ident_t* nonnull @0, i32 %tmp)
   br label %omp_if.end
 
 omp_if.end:                                       ; preds = %entry, %omp_if.then
   call void @__kmpc_barrier(%struct.ident_t* nonnull @1, i32 %omp_global_thread_num) #6
   ret void
 }
 
 define internal void @.omp_outlined..6(i32* noalias %.global_tid., i32* noalias %.bound_tid., i32* dereferenceable(4) %a) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..6
-; CHECK-SAME: (i32* noalias nocapture noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A1:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
 ; CHECK-NEXT:    [[TMP:%.*]] = bitcast i32* [[A1]] to i8*
-; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 [[TMP]]) [[ATTR0]]
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 [[TMP]]) #[[ATTR0]]
 ; CHECK-NEXT:    store i32 1, i32* [[A1]], align 4
 ; CHECK-NEXT:    [[TMP1:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i32**
 ; CHECK-NEXT:    store i32* [[A1]], i32** [[TMP1]], align 8
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
 ; CHECK-NEXT:    [[TMP3:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
-; CHECK-NEXT:    [[TMP4:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* noundef nonnull [[GLOB2:@.*]], i32 [[TMP2]], i32 noundef 1, i64 noundef 8, i8* noundef nonnull align 8 [[TMP3]], void (i8*, i8*)* noundef nonnull @.omp.reduction.reduction_func, [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK-NEXT:    [[TMP4:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 noundef 1, i64 noundef 8, i8* noundef nonnull align 8 [[TMP3]], void (i8*, i8*)* noundef nonnull @.omp.reduction.reduction_func, [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
 ; CHECK-NEXT:    switch i32 [[TMP4]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
 ; CHECK-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
 ; CHECK-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
 ; CHECK-NEXT:    ]
 ; CHECK:       .omp.reduction.case1:
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A1]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[A]], align 4
-; CHECK-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* noundef nonnull [[GLOB2]], i32 [[TMP2]], [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2]], i32 [[TMP2]], [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
 ; CHECK-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
 ; CHECK:       .omp.reduction.case2:
 ; CHECK-NEXT:    [[TMP7:%.*]] = load i32, i32* [[A1]], align 4
 ; CHECK-NEXT:    [[TMP8:%.*]] = atomicrmw add i32* [[A]], i32 [[TMP7]] monotonic, align 4
 ; CHECK-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
 ; CHECK:       .omp.reduction.default:
 ; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i32* [[A1]] to i8*
 ; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP9]])
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK1-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A1:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
+; CHECK1-NEXT:    [[TMP:%.*]] = bitcast i32* [[A1]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 [[TMP]]) #[[ATTR0]]
+; CHECK1-NEXT:    store i32 1, i32* [[A1]], align 4
+; CHECK1-NEXT:    [[TMP1:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i32**
+; CHECK1-NEXT:    store i32* [[A1]], i32** [[TMP1]], align 8
+; CHECK1-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP3:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
+; CHECK1-NEXT:    [[TMP4:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 noundef 1, i64 noundef 8, i8* noundef nonnull align 8 [[TMP3]], void (i8*, i8*)* noundef nonnull @.omp.reduction.reduction_func, [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK1-NEXT:    switch i32 [[TMP4]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
+; CHECK1-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
+; CHECK1-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
+; CHECK1-NEXT:    ]
+; CHECK1:       .omp.reduction.case1:
+; CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A1]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[A]], align 4
+; CHECK1-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2]], i32 [[TMP2]], [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
+; CHECK1:       .omp.reduction.case2:
+; CHECK1-NEXT:    [[TMP7:%.*]] = load i32, i32* [[A1]], align 4
+; CHECK1-NEXT:    [[TMP8:%.*]] = atomicrmw add i32* [[A]], i32 [[TMP7]] monotonic, align 4
+; CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
+; CHECK1:       .omp.reduction.default:
+; CHECK1-NEXT:    [[TMP9:%.*]] = bitcast i32* [[A1]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP9]])
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK2-SAME: (i32* noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A1:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x i8*], align 8
+; CHECK2-NEXT:    [[TMP:%.*]] = bitcast i32* [[A1]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull align 4 [[TMP]]) #[[ATTR0]]
+; CHECK2-NEXT:    store i32 1, i32* [[A1]], align 4
+; CHECK2-NEXT:    [[TMP1:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i32**
+; CHECK2-NEXT:    store i32* [[A1]], i32** [[TMP1]], align 8
+; CHECK2-NEXT:    [[TMP2:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP3:%.*]] = bitcast [1 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8*
+; CHECK2-NEXT:    [[TMP4:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 noundef 1, i64 noundef 8, i8* noundef nonnull align 8 [[TMP3]], void (i8*, i8*)* noundef nonnull @.omp.reduction.reduction_func, [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK2-NEXT:    switch i32 [[TMP4]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
+; CHECK2-NEXT:    i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
+; CHECK2-NEXT:    i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
+; CHECK2-NEXT:    ]
+; CHECK2:       .omp.reduction.case1:
+; CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A1]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[A]], align 4
+; CHECK2-NEXT:    call void @__kmpc_end_reduce_nowait(%struct.ident_t* noundef nonnull @[[GLOB2]], i32 [[TMP2]], [8 x i32]* noundef nonnull @.gomp_critical_user_.reduction.var)
+; CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
+; CHECK2:       .omp.reduction.case2:
+; CHECK2-NEXT:    [[TMP7:%.*]] = load i32, i32* [[A1]], align 4
+; CHECK2-NEXT:    [[TMP8:%.*]] = atomicrmw add i32* [[A]], i32 [[TMP7]] monotonic, align 4
+; CHECK2-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
+; CHECK2:       .omp.reduction.default:
+; CHECK2-NEXT:    [[TMP9:%.*]] = bitcast i32* [[A1]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP9]])
+; CHECK2-NEXT:    ret void
 entry:
   %a1 = alloca i32, align 4
   %.omp.reduction.red_list = alloca [1 x i8*], align 8
   %tmp = bitcast i32* %a1 to i8*
   call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %tmp)
   store i32 1, i32* %a1, align 4
   %tmp1 = bitcast [1 x i8*]* %.omp.reduction.red_list to i32**
   store i32* %a1, i32** %tmp1, align 8

 .omp.reduction.default:                           ; preds = %.omp.reduction.case2, %.omp.reduction.case1, %entry
   %tmp9 = bitcast i32* %a1 to i8*
   call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %tmp9)
   ret void
 }
 
 define internal void @.omp.reduction.reduction_func(i8* %arg, i8* %arg1) {
 ; CHECK-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
-; CHECK-SAME: (i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG:%.*]], i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG1:%.*]]) [[ATTR9:#.*]] {
+; CHECK-SAME: (i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG:%.*]], i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG1:%.*]]) #[[ATTR10:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP:%.*]] = bitcast i8* [[ARG1]] to i32**
 ; CHECK-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
 ; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i8* [[ARG]] to i32**
 ; CHECK-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP3]], align 8
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
 ; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP2]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[TMP4]], align 4
 ; CHECK-NEXT:    ret void
 ;
+; CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
+; CHECK1-SAME: (i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG:%.*]], i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG1:%.*]]) #[[ATTR10:[0-9]+]] {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP:%.*]] = bitcast i8* [[ARG1]] to i32**
+; CHECK1-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
+; CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i8* [[ARG]] to i32**
+; CHECK1-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP3]], align 8
+; CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
+; CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP2]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[TMP4]], align 4
+; CHECK1-NEXT:    ret void
+; CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
+; CHECK2-SAME: (i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG:%.*]], i8* nocapture nofree nonnull readonly align 8 dereferenceable(8) [[ARG1:%.*]]) #[[ATTR10:[0-9]+]] {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP:%.*]] = bitcast i8* [[ARG1]] to i32**
+; CHECK2-NEXT:    [[TMP2:%.*]] = load i32*, i32** [[TMP]], align 8
+; CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i8* [[ARG]] to i32**
+; CHECK2-NEXT:    [[TMP4:%.*]] = load i32*, i32** [[TMP3]], align 8
+; CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[TMP4]], align 4
+; CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[TMP2]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP5]], [[TMP6]]
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[TMP4]], align 4
+; CHECK2-NEXT:    ret void
 entry:
   %tmp = bitcast i8* %arg1 to i32**
   %tmp2 = load i32*, i32** %tmp, align 8
   %tmp3 = bitcast i8* %arg to i32**
   %tmp4 = load i32*, i32** %tmp3, align 8
   %tmp5 = load i32, i32* %tmp4, align 4
   %tmp6 = load i32, i32* %tmp2, align 4
   %add = add nsw i32 %tmp5, %tmp6

llvm/test/Transforms/OpenMP/parallel_region_merging.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs
-; RUN: opt -S -attributor -openmp-opt-cgscc -openmp-opt-enable-merging -enable-new-pm=0 < %s | FileCheck %s
-; RUN: opt -S -aa-pipeline= -passes='attributor,cgscc(openmp-opt-cgscc)' -openmp-opt-enable-merging  < %s | FileCheck %s
+; RUN: opt -S -attributor -openmp-opt-cgscc -openmp-opt-enable-merging -enable-new-pm=0 < %s | FileCheck %s --check-prefix=CHECK1
+; RUN: opt -S -aa-pipeline= -passes='attributor,cgscc(openmp-opt-cgscc)' -openmp-opt-enable-merging  < %s | FileCheck %s --check-prefix=CHECK2
 ; #include <omp.h>
 ; void foo();
 ; void use(int);
 ; void usef(float);
 ; void merge(int a) {
 ; #pragma omp parallel
 ;     {
 ;         use(a);

 !0 = !{i32 1, !"wchar_size", i32 4}
 !1 = !{!2}
 !2 = !{i64 2, i64 -1, i64 -1, i1 true}
 ; CHECK-LABEL: define {{[^@]+}}@merge
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1:@.*]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) [[ATTR0:#.*]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0:[0-9]+]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2:@.*]], i32 [[OMP_GLOBAL_THREAD_NUM]])
-; CHECK-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..1
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_proc_bind
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 [[TMP0]], i32 noundef 3)
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..2
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..3
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_num_threads
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 [[TMP0]], i32 [[A]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..4
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..5
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_seq_call
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    call void (...) @foo()
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..6
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..7
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..8
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..9
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_float
-; CHECK-SAME: (float [[F:%.*]], float* nocapture writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
 ; CHECK-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
 ; CHECK-NEXT:    store float [[ADD]], float* [[P]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..10
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
 ; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
 ; CHECK-NEXT:    call void @use(i32 [[CONV]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..11
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
 ; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
 ; CHECK-NEXT:    call void @use(i32 [[CONV]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
 ; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
 ; CHECK-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
 ; CHECK-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..12
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..13
 ; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
 ; CHECK-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
 ; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
 ; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
 ; CHECK-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
 ; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..14
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..15
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
 ; CHECK-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
 ; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
 ; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
 ; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..16
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..17
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP1]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions
 ; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
-; CHECK-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..18
-; CHECK-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
 ; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
 ; CHECK-NEXT:    ret void
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..19
-; CHECK-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
 ; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
 ; CHECK-NEXT:    ret void
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
 ; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
 ; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
 ; CHECK-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
 ; CHECK-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..20
-; CHECK-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
 ; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
 ; CHECK-NEXT:    ret void
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..21
-; CHECK-SAME: (i32* noalias nocapture readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
 ; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
 ; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
 ; CHECK:       if.then:
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
-; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
 ; CHECK-NEXT:    ret void
 ; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_3
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_3..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
-; CHECK-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
-; CHECK-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..22
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..23
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..24
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_3_seq
 ; CHECK-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
 ; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
 ; CHECK:       omp_region.end:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
 ; CHECK-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
 ; CHECK:       omp_region.end4:
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split.split.split:
-; CHECK-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
 ; CHECK:       omp_region.body5:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
 ; CHECK:       seq.par.merged2:
 ; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
 ; CHECK-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
 ; CHECK-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split.split.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
 ; CHECK:       omp_region.body5.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END4]]
 ; CHECK:       omp_region.body:
 ; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
 ; CHECK:       seq.par.merged:
 ; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
 ; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
 ; CHECK-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
 ; CHECK:       omp.par.merged.split:
 ; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
 ; CHECK:       omp_region.body.split:
-; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* [[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
 ; CHECK-NEXT:    br label [[OMP_REGION_END]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..25
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..26
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..27
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_3_seq_call
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    call void (...) @foo()
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    call void (...) @foo()
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..28
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..29
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..30
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 [[TMP0]], i32 noundef 3)
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..31
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..32
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..33
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@unmergable_3_num_threads
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..34
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..35
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..36
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1
 ; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
 ; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
 ; CHECK:       omp_parallel:
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* [[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
 ; CHECK:       omp.par.outlined.exit:
 ; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
 ; CHECK:       omp.par.exit.split:
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
 ; CHECK:       entry.split.split:
 ; CHECK-NEXT:    call void (...) @foo()
-; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) [[ATTR0]] {
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
 ; CHECK-NEXT:  omp.par.entry:
 ; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
 ; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
 ; CHECK:       omp.par.outlined.exit.exitStub:
 ; CHECK-NEXT:    ret void
 ; CHECK:       omp.par.region:
 ; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
 ; CHECK:       omp.par.merged:
-; CHECK-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
-; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
-; CHECK-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
 ; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
 ; CHECK:       entry.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
 ; CHECK:       omp.par.region.split:
 ; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
 ; CHECK:       omp.par.pre_finalize:
 ; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..37
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..38
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..39
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..7
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..8
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..9
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float
+; CHECK-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
+; CHECK-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
+; CHECK-NEXT:    store float [[ADD]], float* [[P]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..10
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..11
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
+; CHECK-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..12
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..13
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
+; CHECK-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
+; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
+; CHECK-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..14
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..15
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..16
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..17
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP1]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..18
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..19
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
+; CHECK-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
+; CHECK-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..20
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..21
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..22
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..23
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..24
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq
+; CHECK-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
+; CHECK:       omp_region.end4:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body5:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
+; CHECK:       seq.par.merged2:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
+; CHECK-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
+; CHECK:       omp_region.body5.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    br label [[OMP_REGION_END4]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..25
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..26
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..27
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..28
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..29
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..30
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..31
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..32
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..33
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..34
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..35
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..36
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..37
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..38
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..39
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..7
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..8
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..9
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float
+; CHECK-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
+; CHECK-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
+; CHECK-NEXT:    store float [[ADD]], float* [[P]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..10
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..11
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
+; CHECK-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..12
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..13
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
+; CHECK-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
+; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
+; CHECK-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..14
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..15
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..16
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..17
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP1]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..18
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..19
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
+; CHECK-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
+; CHECK-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..20
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..21
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..22
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..23
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..24
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq
+; CHECK-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
+; CHECK:       omp_region.end4:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body5:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
+; CHECK:       seq.par.merged2:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
+; CHECK-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
+; CHECK:       omp_region.body5.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    br label [[OMP_REGION_END4]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..25
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..26
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..27
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..28
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..29
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..30
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..31
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..32
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..33
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..34
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..35
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..36
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..37
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..38
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..39
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..7
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..8
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..9
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float
+; CHECK-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
+; CHECK-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
+; CHECK-NEXT:    store float [[ADD]], float* [[P]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..10
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..11
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK-NEXT:    call void @use(i32 [[CONV]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
+; CHECK-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..12
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..13
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
+; CHECK-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
+; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
+; CHECK-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..14
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..15
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..16
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..17
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP1]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..18
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..19
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
+; CHECK-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
+; CHECK-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
+; CHECK-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..20
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..21
+; CHECK-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK:       if.then:
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK-NEXT:    ret void
+; CHECK:       if.end:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..22
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..23
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..24
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq
+; CHECK-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK:       omp_region.end:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
+; CHECK:       omp_region.end4:
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split.split:
+; CHECK-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK:       omp_region.body5:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
+; CHECK:       seq.par.merged2:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
+; CHECK-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split.split.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
+; CHECK:       omp_region.body5.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK-NEXT:    br label [[OMP_REGION_END4]]
+; CHECK:       omp_region.body:
+; CHECK-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK:       seq.par.merged:
+; CHECK-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
+; CHECK-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK:       omp.par.merged.split:
+; CHECK-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK:       omp_region.body.split:
+; CHECK-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    br label [[OMP_REGION_END]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..25
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..26
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..27
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_seq_call
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..28
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..29
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..30
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..31
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..32
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..33
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@unmergable_3_num_threads
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..34
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..35
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..36
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1
+; CHECK-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK:       omp_parallel:
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK:       omp.par.outlined.exit:
+; CHECK-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK:       omp.par.exit.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
+; CHECK-NEXT:    call void (...) @foo()
+; CHECK-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
+; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT:  omp.par.entry:
+; CHECK-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK:       omp.par.outlined.exit.exitStub:
+; CHECK-NEXT:    ret void
+; CHECK:       omp.par.region:
+; CHECK-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK:       omp.par.merged:
+; CHECK-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK:       entry.split:
+; CHECK-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK:       omp.par.region.split:
+; CHECK-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK:       omp.par.pre_finalize:
+; CHECK-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..37
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK-NEXT:    ret void
+; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..38
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
-;
-;
 ; CHECK-LABEL: define {{[^@]+}}@.omp_outlined..39
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
 ; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
+; CHECK1-LABEL: define {{[^@]+}}@merge
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_proc_bind
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_num_threads
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_seq_call
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (...) @foo()
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..7
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..8
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..9
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_float
+; CHECK1-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
+; CHECK1-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
+; CHECK1-NEXT:    store float [[ADD]], float* [[P]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..10
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK1-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK1-NEXT:    call void @use(i32 [[CONV]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..11
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK1-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK1-NEXT:    call void @use(i32 [[CONV]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_firstprivate
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK1-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
+; CHECK1-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..12
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..13
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
+; CHECK1-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_sink_lt
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK1-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
+; CHECK1-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
+; CHECK1-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
+; CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..14
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..15
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_par_use
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK1-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK1-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..16
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..17
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP1]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_cancellable_regions
+; CHECK1-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..18
+; CHECK1-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK1-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK1-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK1:       if.then:
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK1-NEXT:    ret void
+; CHECK1:       if.end:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..19
+; CHECK1-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK1-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK1-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK1:       if.then:
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK1-NEXT:    ret void
+; CHECK1:       if.end:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
+; CHECK1-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
+; CHECK1-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
+; CHECK1-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
+; CHECK1-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..20
+; CHECK1-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK1-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK1-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK1:       if.then:
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK1-NEXT:    ret void
+; CHECK1:       if.end:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..21
+; CHECK1-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK1-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK1-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK1:       if.then:
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK1-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK1-NEXT:    ret void
+; CHECK1:       if.end:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_3
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_3..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..22
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..23
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..24
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_3_seq
+; CHECK1-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK1-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK1:       omp_region.end:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK1-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK1-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
+; CHECK1:       omp_region.end4:
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split.split.split:
+; CHECK1-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK1:       omp_region.body5:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
+; CHECK1:       seq.par.merged2:
+; CHECK1-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK1-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
+; CHECK1-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split.split.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
+; CHECK1:       omp_region.body5.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END4]]
+; CHECK1:       omp_region.body:
+; CHECK1-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK1:       seq.par.merged:
+; CHECK1-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
+; CHECK1-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK1:       omp.par.merged.split:
+; CHECK1-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK1:       omp_region.body.split:
+; CHECK1-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..25
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..26
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..27
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_3_seq_call
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (...) @foo()
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (...) @foo()
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..28
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..29
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..30
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..31
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..32
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..33
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@unmergable_3_num_threads
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..34
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..35
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..36
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_2_unmergable_1
+; CHECK1-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK1:       omp_parallel:
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK1:       omp.par.outlined.exit:
+; CHECK1-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK1:       omp.par.exit.split:
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK1:       entry.split.split:
+; CHECK1-NEXT:    call void (...) @foo()
+; CHECK1-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
+; CHECK1-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK1-NEXT:  omp.par.entry:
+; CHECK1-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK1-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK1:       omp.par.outlined.exit.exitStub:
+; CHECK1-NEXT:    ret void
+; CHECK1:       omp.par.region:
+; CHECK1-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK1:       omp.par.merged:
+; CHECK1-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK1-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK1-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK1-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK1:       entry.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK1:       omp.par.region.split:
+; CHECK1-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK1:       omp.par.pre_finalize:
+; CHECK1-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..37
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..38
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..39
+; CHECK1-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK1-NEXT:  entry:
+; CHECK1-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK1-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK1-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1:[0-9]+]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0:[0-9]+]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined.
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..1
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_proc_bind
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..2
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..3
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_num_threads
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[A]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..4
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..5
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_seq_call
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (...) @foo()
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..6
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..7
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..8
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..9
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_float
+; CHECK2-SAME: (float [[F:%.*]], float* nocapture nofree writeonly [[P:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[F_ADDR:%.*]] = alloca float, align 4
+; CHECK2-NEXT:    store float [[F]], float* [[F_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*, float*)* @merge_seq_float..omp_par to void (i32*, i32*, ...)*), float* [[F_ADDR]], float* [[P]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_float..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], float* [[F_ADDR:%.*]], float* [[P:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load float, float* [[F_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = fadd float [[TMP3]], 0x40091EB860000000
+; CHECK2-NEXT:    store float [[ADD]], float* [[P]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..10
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK2-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK2-NEXT:    call void @use(i32 [[CONV]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..11
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[F:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load float, float* [[F]], align 4
+; CHECK2-NEXT:    [[CONV:%.*]] = fptosi float [[TMP0]] to i32
+; CHECK2-NEXT:    call void @use(i32 [[CONV]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_firstprivate
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]] = alloca i64, align 8
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64*)* @merge_seq_firstprivate..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_firstprivate..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK2-NEXT:    call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP3]], 1
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[A_CASTED_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[ADD]] to i64
+; CHECK2-NEXT:    store i64 [[A_CASTED_SROA_0_0_INSERT_EXT]], i64* [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..12
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..13
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i64 [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[A]] to i32
+; CHECK2-NEXT:    call void @use(i32 [[A_ADDR_SROA_0_0_EXTRACT_TRUNC]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_sink_lt
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_seq_sink_lt..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_sink_lt..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.start.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK2-NEXT:    [[TMP4:%.*]] = ptrtoint i32* [[B]] to i64
+; CHECK2-NEXT:    [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32
+; CHECK2-NEXT:    store i32 [[TMP5]], i32* [[B]], align 4
+; CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[TMP3]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..14
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..15
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_par_use
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[B:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    [[LT_CAST3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.start.p0i8(i64 -1, i8* [[LT_CAST3]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_seq_par_use..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[B]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    [[LT_CAST:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK2-NEXT:    call void @llvm.lifetime.end.p0i8(i64 noundef 4, i8* noundef nonnull [[LT_CAST]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_seq_par_use..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[B:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = bitcast i32* [[B]] to i8*
+; CHECK2-NEXT:    [[TMP4:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP4]], 1
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[B]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..16
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..17
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[B:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[B]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP1]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_cancellable_regions
+; CHECK2-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..18
+; CHECK2-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK2-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK2-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK2:       if.then:
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK2-NEXT:    ret void
+; CHECK2:       if.end:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..19
+; CHECK2-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK2-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK2-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK2:       if.then:
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK2-NEXT:    ret void
+; CHECK2:       if.end:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq
+; CHECK2-SAME: (i32 [[CANCEL1:%.*]], i32 [[CANCEL2:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[CANCEL1_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[CANCEL2_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[CANCEL1]], i32* [[CANCEL1_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[CANCEL2]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @merge_cancellable_regions_seq..omp_par to void (i32*, i32*, ...)*), i32* [[CANCEL1_ADDR]], i32* [[CANCEL2_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_cancellable_regions_seq..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[CANCEL1_ADDR:%.*]], i32* [[CANCEL2_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP3:%.*]] = load i32, i32* [[CANCEL1_ADDR]], align 4
+; CHECK2-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP3]], 0
+; CHECK2-NEXT:    [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL_NOT]] to i32
+; CHECK2-NEXT:    store i32 [[LNOT_EXT]], i32* [[CANCEL2_ADDR]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..20
+; CHECK2-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL1:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL1]], align 4
+; CHECK2-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK2-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK2:       if.then:
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK2-NEXT:    ret void
+; CHECK2:       if.end:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..21
+; CHECK2-SAME: (i32* noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[CANCEL2:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[CANCEL2]], align 4
+; CHECK2-NEXT:    [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK2-NEXT:    br i1 [[TOBOOL_NOT]], label [[IF_END:%.*]], label [[IF_THEN:%.*]]
+; CHECK2:       if.then:
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[DOTGLOBAL_TID_]], align 4
+; CHECK2-NEXT:    [[TMP2:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull @[[GLOB1]], i32 [[TMP1]], i32 noundef 1)
+; CHECK2-NEXT:    ret void
+; CHECK2:       if.end:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_3
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_3..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_3..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..22
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..23
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..24
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_3_seq
+; CHECK2-SAME: (i32 [[A:%.*]], i32 [[B:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[ADD1_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[ADD_SEQ_OUTPUT_ALLOC:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM7:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*, i32*)* @merge_3_seq..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    [[ADD1_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[ADD1_SEQ_OUTPUT_LOAD]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_3_seq..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]], i32* [[ADD_SEQ_OUTPUT_ALLOC:%.*]], i32* [[ADD1_SEQ_OUTPUT_ALLOC:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    [[TMP2:%.*]] = icmp ne i32 [[TMP1]], 0
+; CHECK2-NEXT:    br i1 [[TMP2]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
+; CHECK2:       omp_region.end:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK2-NEXT:    [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
+; CHECK2-NEXT:    br i1 [[TMP4]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
+; CHECK2:       omp_region.end4:
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split.split.split:
+; CHECK2-NEXT:    call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+; CHECK2:       omp_region.body5:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED2:%.*]]
+; CHECK2:       seq.par.merged2:
+; CHECK2-NEXT:    [[TMP5:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD_SEQ_OUTPUT_LOAD:%.*]] = load i32, i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK2-NEXT:    [[ADD1:%.*]] = add nsw i32 [[ADD_SEQ_OUTPUT_LOAD]], [[TMP5]]
+; CHECK2-NEXT:    store i32 [[ADD1]], i32* [[ADD1_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split.split.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY5_SPLIT:%.*]]
+; CHECK2:       omp_region.body5.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END4]]
+; CHECK2:       omp_region.body:
+; CHECK2-NEXT:    br label [[SEQ_PAR_MERGED:%.*]]
+; CHECK2:       seq.par.merged:
+; CHECK2-NEXT:    [[TMP6:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
+; CHECK2-NEXT:    store i32 [[ADD]], i32* [[ADD_SEQ_OUTPUT_ALLOC]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED_SPLIT:%.*]]
+; CHECK2:       omp.par.merged.split:
+; CHECK2-NEXT:    br label [[OMP_REGION_BODY_SPLIT:%.*]]
+; CHECK2:       omp_region.body.split:
+; CHECK2-NEXT:    call void @__kmpc_end_master(%struct.ident_t* @[[GLOB1]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    br label [[OMP_REGION_END]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..25
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..26
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..27
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_3_seq_call
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (...) @foo()
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (...) @foo()
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..28
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..29
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..30
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_3_proc_bind
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void @__kmpc_push_proc_bind(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 noundef 3)
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..31
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..32
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..33
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@unmergable_3_num_threads
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]])
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[TMP1:%.*]] = load i32, i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    call void @__kmpc_push_num_threads(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 [[TMP0]], i32 [[TMP1]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..34
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..35
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..36
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_2_unmergable_1
+; CHECK2-SAME: (i32 [[A:%.*]]) local_unnamed_addr {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    store i32 [[A]], i32* [[A_ADDR]], align 4
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    br label [[OMP_PARALLEL:%.*]]
+; CHECK2:       omp_parallel:
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_2_unmergable_1..omp_par to void (i32*, i32*, ...)*), i32* [[A_ADDR]])
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT:%.*]]
+; CHECK2:       omp.par.outlined.exit:
+; CHECK2-NEXT:    br label [[OMP_PAR_EXIT_SPLIT:%.*]]
+; CHECK2:       omp.par.exit.split:
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK2:       entry.split.split:
+; CHECK2-NEXT:    call void (...) @foo()
+; CHECK2-NEXT:    call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 1, void (i32*, i32*, ...)* noundef bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@merge_2_unmergable_1..omp_par
+; CHECK2-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[A_ADDR:%.*]]) #[[ATTR0]] {
+; CHECK2-NEXT:  omp.par.entry:
+; CHECK2-NEXT:    [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[TID_ADDR]], align 4
+; CHECK2-NEXT:    store i32 [[TMP0]], i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION:%.*]]
+; CHECK2:       omp.par.outlined.exit.exitStub:
+; CHECK2-NEXT:    ret void
+; CHECK2:       omp.par.region:
+; CHECK2-NEXT:    br label [[OMP_PAR_MERGED:%.*]]
+; CHECK2:       omp.par.merged:
+; CHECK2-NEXT:    call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]])
+; CHECK2-NEXT:    call void @__kmpc_barrier(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
+; CHECK2-NEXT:    call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]])
+; CHECK2-NEXT:    br label [[ENTRY_SPLIT:%.*]]
+; CHECK2:       entry.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_REGION_SPLIT:%.*]]
+; CHECK2:       omp.par.region.split:
+; CHECK2-NEXT:    br label [[OMP_PAR_PRE_FINALIZE:%.*]]
+; CHECK2:       omp.par.pre_finalize:
+; CHECK2-NEXT:    br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..37
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..38
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
+;
+;
+; CHECK2-LABEL: define {{[^@]+}}@.omp_outlined..39
+; CHECK2-SAME: (i32* noalias nocapture nofree readnone [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[A:%.*]]) {
+; CHECK2-NEXT:  entry:
+; CHECK2-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK2-NEXT:    call void @use(i32 [[TMP0]])
+; CHECK2-NEXT:    ret void
 ;