[OpenMPOpt][WIP] Expand parallel region merging

[ggeorgakoudis]

Summary

The existing implementation of parallel region merging applies only to
consecutive parallel regions that have speculatable sequential
instructions in-between. This patch lifts this limitation to expand
merging with any sequential instructions in-between, except calls to
unmergable OpenMP runtime functions. In-between sequential instructions
in the merged region are sequentialized in a "master" region and any
output values are broadcasted to the following parallel regions and the
sequential region continuation of the merged region.

[jdoerfert]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

616

Don't we have to proof the absence of these calls and not "look for them". I mean foobar is not one of them but could call one of them, right?

[ggeorgakoudis]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

616

Yup, you're absolutely right. I was short sighted. We either have to check that there is no callpath from an in-between instruction to one of those functions and abort merging the affected regions, or entirely bail out from merging when there is any one of them declared, as we were doing before. I probably need to include omp_proc_bind and look for others too.

Also it got me thinking. What if there's a kpmc_fork_call inside a function call in the sequential region? Can a fork call be called within a master sequential region? Need to find out more.

[jdoerfert]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

616

You cannot have "any" OpenMP runtime calls in the guarded parts. One easy way is to disallow any call. Next step is to add an Attributor abstract attribute (as part of OpenMPOpt) which will deduce if a function might call an OpenMP runtime function, e.g., if it never calls an external function we don't know it cannot call an OpenMP runtime function. You will also be able to use the omp assume attribtues that will be introduced shortly.

[ggeorgakoudis]

Simplify check for unmergable regions, add flag for extractor sinking in OMPIRBuilder outlining.

[ggeorgakoudis]

Update to correctly handle inputs to sequential regions.

[jdoerfert]

Apologies for the delayed review.

Update the commit title and message.

I think we are basically set here. I left some minor comments, all but the Input alloca question and the deleted file are not worth a new round of review, I just want confirmation on those two points.

llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

188

If this flag is not set there should not be any instructions, right? If so, we should add an assertion for that.

Style:
I would not use the iterators as you never actually reach the end() anyway. while (true) { I = BB.front(); or while(BB.size() > 1) {... or something.

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

641

Add a comment describing what this is doing on a high-level.

673

Conversion to pointer happens in the IRBuilder (after D92189 landed). This means we can omit the input alloca stuff below, right?

777

Range loop :)

783

We usually recommend to compute the end iterator only once.

885

Add some comment explaining what this is doing on a high-level.

905

A lot of heuristics could be used until we have an AAOpenMPRuntimeCalls attribute that determines if a call can result in an OpenMP runtime call.
Lifetime intrinisics is fine, any intrinsic should be OK actually.

915

Reverse the order ("early exits") to reduce indention:

`
if terminator
  ...
if !call
  ...
call handling.

928–931

If BB->end() stays the same capture it; It = .., End = ...; It != End

985

Can we make a helper for this, pass he OMPRTL kind and it does the 3 steps.

llvm/test/Transforms/OpenMP/parallel_region_merging_legacy_pm.ll

412 ↗

(On Diff #306308)

Why was this file deleted?

[ggeorgakoudis]

Updated for comments.

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

673

That's correct. Removed.

905

Makes sense. Updated to accept any intrinsic call as mergable.

llvm/test/Transforms/OpenMP/parallel_region_merging_legacy_pm.ll

412 ↗

(On Diff #306308)

There were different test files for the legacy and the new PMs because of them processing changes in the CG in opposite order, so function names for outlined functions wouldn't match. This has been resolved in D90566,

[jdoerfert]

I Left some more comments that can be addressed before the commit. Otherwise LGTM.

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

691

If you just want the users of instructions in SeqStartBB outside of that BB we don't need the extractor logic which probably does a lot more than we need. If you agree, maybe something like this:

for (Instruction &I : *SeqStartBB) {
  SmallPtrSet<Instruction *, 4> OutsideUsers;
  for (User &Usr : I.users()) {
    Instruction &UsrI = *cast<Instruction>(Usr);
    if (UsrI.getParent() != SeqStartBB)
      OutsideUsers.insert(&UsrI);
  }
  if (OutsideUsers.empty())
    continue;
  // Do the stuff as below.
}

899

Early exit:

if (!isa<Call>)
  return true;

902

if (!CalledFunction) return false;

908

const auto &RFI :

951

Nit: pre-increment everywhere above: ++It

llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h

public:
   OpenMPIRBuilder(Module &M) : M(M), Builder(M.getContext()) {}
 
   /// Initialize the internal state, this will put structures types and
   /// potentially other helpers into the underlying module. Must be called
   /// before any other method and only once!
   void initialize();
 
   /// Finalize the underlying module, e.g., by outlining regions.
-  void finalize();
+  /// \param AllowExtractorSinking Flag to include sinking instructions,
+  ///                              emitted by CodeExtractor, in the
+  ///                              outlined region. Default is false.
+  void finalize(bool AllowExtractorSinking = false);
 
   /// Add attributes known for \p FnID to \p Fn.
   void addAttributes(omp::RuntimeFunction FnID, Function &Fn);
 
   /// Type used throughout for insertion points.
   using InsertPointTy = IRBuilder<>::InsertPoint;
 
   /// Callback type for variable finalization (think destructors).

llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

Function *OpenMPIRBuilder::getOrCreateRuntimeFunctionPtr(RuntimeFunction FnID) {
   FunctionCallee RTLFn = getOrCreateRuntimeFunction(M, FnID);
   auto *Fn = dyn_cast<llvm::Function>(RTLFn.getCallee());
   assert(Fn && "Failed to create OpenMP runtime function pointer");
   return Fn;
 }
 
 void OpenMPIRBuilder::initialize() { initializeTypes(M); }
 
-void OpenMPIRBuilder::finalize() {
+void OpenMPIRBuilder::finalize(bool AllowExtractorSinking) {
   SmallPtrSet<BasicBlock *, 32> ParallelRegionBlockSet;
   SmallVector<BasicBlock *, 32> Blocks;
   for (OutlineInfo &OI : OutlineInfos) {
     ParallelRegionBlockSet.clear();
     Blocks.clear();
     OI.collectBlocks(ParallelRegionBlockSet, Blocks);
 
     Function *OuterFn = OI.EntryBB->getParent();
for (OutlineInfo &OI : OutlineInfos) {
     M.getFunctionList().insertAfter(OuterFn->getIterator(), OutlinedFn);
 
     // Remove the artificial entry introduced by the extractor right away, we
     // made our own entry block after all.
     {
       BasicBlock &ArtificialEntry = OutlinedFn->getEntryBlock();
       assert(ArtificialEntry.getUniqueSuccessor() == OI.EntryBB);
       assert(OI.EntryBB->getUniquePredecessor() == &ArtificialEntry);
+      if (AllowExtractorSinking) {
+        // Move instructions from the to-be-deleted ArtificialEntry to the entry
+        // basic block of the parallel region. CodeExtractor may have sunk
+        // allocas/bitcasts for values that are solely used in the outlined
+        // region and do not escape.
+        assert(!ArtificialEntry.empty() &&
+               "Expected instructions to sink in the outlined region");
+        for (BasicBlock::iterator It = ArtificialEntry.begin(),
+                                  End = ArtificialEntry.end();
+             It != End;) {
+          Instruction &I = *It;
+          It++;
+
+          if (I.isTerminator())
+            continue;
+
+          I.moveBefore(*OI.EntryBB, OI.EntryBB->getFirstInsertionPt());
+        }
+      }
       OI.EntryBB->moveBefore(&ArtificialEntry);
       ArtificialEntry.eraseFromParent();
     }
     assert(&OutlinedFn->getEntryBlock() == OI.EntryBB);
     assert(OutlinedFn && OutlinedFn->getNumUses() == 1);
 
     // Run a user callback, e.g. to add attributes.
     if (OI.PostOutlineCB)

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

 #include "llvm/Frontend/OpenMP/OMPConstants.h"
 #include "llvm/Frontend/OpenMP/OMPIRBuilder.h"
 #include "llvm/InitializePasses.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Transforms/IPO.h"
 #include "llvm/Transforms/IPO/Attributor.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
 #include "llvm/Transforms/Utils/CallGraphUpdater.h"
+#include "llvm/Transforms/Utils/CodeExtractor.h"
 
 using namespace llvm;
 using namespace omp;
 
 #define DEBUG_TYPE "openmp-opt"
 
 static cl::opt<bool> DisableOpenMPOptimizations(
     "openmp-opt-disable", cl::ZeroOrMore,
for (Use &U : RFI.Declaration->uses()) {
       } else {
         RFI.getOrCreateUseVector(nullptr).push_back(&U);
         ++NumUses;
       }
     }
     return NumUses;
   }
 
-  // Helper function to recollect uses of all runtime functions.
-  void recollectUses() {
-    for (int Idx = 0; Idx < RFIs.size(); ++Idx) {
-      auto &RFI = RFIs[static_cast<RuntimeFunction>(Idx)];
-      RFI.clearUsesMap();
-      collectUses(RFI, /*CollectStats*/ false);
-    }
+  // Helper function to recollect uses of a runtime function.
+  void recollectUsesForFunction(RuntimeFunction RTF) {
+    auto &RFI = RFIs[RTF];
+
+  // Helper function to recollect uses of all runtime functions.
+  void recollectUses() {
+    for (int Idx = 0; Idx < RFIs.size(); ++Idx)
+      recollectUsesForFunction(static_cast<RuntimeFunction>(Idx));
   }
 
   /// Helper to initialize all runtime function information for those defined
   /// in OpenMPKinds.def.
   void initializeRuntimeFunctions() {
     Module &M = *((*ModuleSlice.begin())->getParent());
 
     // Helper macros for handling __VA_ARGS__ in OMP_RTL
bool mergeParallelRegions() {
 
     // Check if there are any __kmpc_fork_call calls to merge.
     OMPInformationCache::RuntimeFunctionInfo &RFI =
         OMPInfoCache.RFIs[OMPRTL___kmpc_fork_call];
 
     if (!RFI.Declaration)
       return false;
 
-    // Check if there any __kmpc_push_proc_bind calls for explicit affinities.
-    OMPInformationCache::RuntimeFunctionInfo &ProcBindRFI =
-        OMPInfoCache.RFIs[OMPRTL___kmpc_push_proc_bind];
-
-    // Defensively abort if explicit affinities are set.
-    // TODO: Track ICV proc_bind to merge when mergable regions have the same
-    // affinity.
-    if (ProcBindRFI.Declaration)
-      return false;
+    // Unmergable calls that prevent merging a parallel region.
+    OMPInformationCache::RuntimeFunctionInfo UnmergableCallsInfo[] = {
+        OMPInfoCache.RFIs[OMPRTL___kmpc_push_proc_bind],
+        OMPInfoCache.RFIs[OMPRTL___kmpc_push_num_threads],
+    };
 
     bool Changed = false;
     LoopInfo *LI = nullptr;
     DominatorTree *DT = nullptr;
 
     SmallDenseMap<BasicBlock *, SmallPtrSet<Instruction *, 4>> BB2PRMap;
 
     BasicBlock *StartBB = nullptr, *EndBB = nullptr;
     auto BodyGenCB = [&](InsertPointTy AllocaIP, InsertPointTy CodeGenIP,
                          BasicBlock &ContinuationIP) {
       BasicBlock *CGStartBB = CodeGenIP.getBlock();
       BasicBlock *CGEndBB =
           SplitBlock(CGStartBB, &*CodeGenIP.getPoint(), DT, LI);
       assert(StartBB != nullptr && "StartBB should not be null");
       CGStartBB->getTerminator()->setSuccessor(0, StartBB);
       assert(EndBB != nullptr && "EndBB should not be null");
       EndBB->getTerminator()->setSuccessor(0, CGEndBB);
     };
 
     auto PrivCB = [&](InsertPointTy AllocaIP, InsertPointTy CodeGenIP, Value &,
                       Value &Inner, Value *&ReplacementValue) -> InsertPointTy {
       ReplacementValue = &Inner;
       return CodeGenIP;
     };
 
     auto FiniCB = [&](InsertPointTy CodeGenIP) {};
 
+    /// Create a sequential execution region within a merged parallel region,
+    /// encapsulated in a master construct with a barrier for synchronization.
+    auto CreateSequentialRegion = [&](Function *OuterFn,
+                                      BasicBlock *OuterPredBB,
+                                      Instruction *SeqStartI,
+                                      Instruction *SeqEndI) {
+      // Isolate the instructions of the sequential region to a separate
+      // block.
+      BasicBlock *ParentBB = SeqStartI->getParent();
+      BasicBlock *SeqEndBB =
+          SplitBlock(ParentBB, SeqEndI->getNextNode(), DT, LI);
+      BasicBlock *SeqAfterBB =
+          SplitBlock(SeqEndBB, &*SeqEndBB->getFirstInsertionPt(), DT, LI);
+      BasicBlock *SeqStartBB =
+          SplitBlock(ParentBB, SeqStartI, DT, LI, nullptr, "seq.par.merged");
+
+      assert(ParentBB->getUniqueSuccessor() == SeqStartBB &&
+             "Expected a different CFG");
+      const DebugLoc DL = ParentBB->getTerminator()->getDebugLoc();
+      ParentBB->getTerminator()->eraseFromParent();
+
+      auto BodyGenCB = [&](InsertPointTy AllocaIP, InsertPointTy CodeGenIP,
+                           BasicBlock &ContinuationIP) {
+        BasicBlock *CGStartBB = CodeGenIP.getBlock();
+        BasicBlock *CGEndBB =
+            SplitBlock(CGStartBB, &*CodeGenIP.getPoint(), DT, LI);
+        assert(SeqStartBB != nullptr && "SeqStartBB should not be null");
+        CGStartBB->getTerminator()->setSuccessor(0, SeqStartBB);
+        assert(SeqEndBB != nullptr && "SeqEndBB should not be null");
+        SeqEndBB->getTerminator()->setSuccessor(0, CGEndBB);
+      };
+      auto FiniCB = [&](InsertPointTy CodeGenIP) {};
+
+      // Find outputs from the sequential region to outside users and
+      // broadcast their values to them.
+      for (Instruction &I : *SeqStartBB) {
+        SmallPtrSet<Instruction *, 4> OutsideUsers;
+        for (User *Usr : I.users()) {
+          Instruction &UsrI = *cast<Instruction>(Usr);
+          // Ignore outputs to LT intrinsics, code extraction for the merged
+          // parallel region will fix them.
+          if (UsrI.isLifetimeStartOrEnd())
+            continue;
+
+          if (UsrI.getParent() != SeqStartBB)
+            OutsideUsers.insert(&UsrI);
+        }
+
+        if (OutsideUsers.empty())
+          continue;
+
+        // Emit an alloca in the outer region to store the broadcasted
+        // value.
+        const DataLayout &DL = M.getDataLayout();
+        AllocaInst *AllocaI = new AllocaInst(
+            I.getType(), DL.getAllocaAddrSpace(), nullptr,
+            I.getName() + ".seq.output.alloc", &OuterFn->front().front());
+
+        // Emit a store instruction in the sequential BB to update the
+        // value.
+        new StoreInst(&I, AllocaI, SeqStartBB->getTerminator());
+
+        // Emit a load instruction and replace the use of the output value
+        // with it.
+        for (Instruction *UsrI : OutsideUsers) {
+          LoadInst *LoadI = new LoadInst(I.getType(), AllocaI,
+                                         I.getName() + ".seq.output.load", UsrI);
+          UsrI->replaceUsesOfWith(&I, LoadI);
+        }
+      }
+
+      OpenMPIRBuilder::LocationDescription Loc(
+          InsertPointTy(ParentBB, ParentBB->end()), DL);
+      InsertPointTy SeqAfterIP =
+          OMPInfoCache.OMPBuilder.createMaster(Loc, BodyGenCB, FiniCB);
+
+      OMPInfoCache.OMPBuilder.createBarrier(SeqAfterIP, OMPD_parallel);
+
+      BranchInst::Create(SeqAfterBB, SeqAfterIP.getBlock());
+
+      LLVM_DEBUG(dbgs() << TAG << "After sequential inlining " << *OuterFn
+                        << "\n");
+    };
+
     // Helper to merge the __kmpc_fork_call calls in MergableCIs. They are all
     // contained in BB and only separated by instructions that can be
     // redundantly executed in parallel. The block BB is split before the first
     // call (in MergableCIs) and after the last so the entire region we merge
     // into a single parallel region is contained in a single basic block
     // without any other instructions. We use the OpenMPIRBuilder to outline
     // that block and call the resulting function via __kmpc_fork_call.
     auto Merge = [&](SmallVectorImpl<CallInst *> &MergableCIs, BasicBlock *BB) {
auto Merge = [&](SmallVectorImpl<CallInst *> &MergableCIs, BasicBlock *BB) {
           SplitBlock(EndBB, &*EndBB->getFirstInsertionPt(), DT, LI);
       StartBB = SplitBlock(BB, MergableCIs.front(), DT, LI, nullptr,
                            "omp.par.merged");
 
       assert(BB->getUniqueSuccessor() == StartBB && "Expected a different CFG");
       const DebugLoc DL = BB->getTerminator()->getDebugLoc();
       BB->getTerminator()->eraseFromParent();
 
+      // Create sequential regions for sequential instructions that are
+      // in-between mergable parallel regions.
+      for (auto *It = MergableCIs.begin(), *End = MergableCIs.end() - 1;
+           It != End; ++It) {
+        Instruction *ForkCI = *It;
+        Instruction *NextForkCI = *(It + 1);
+
+        // Continue if there are not in-between instructions.
+        if (ForkCI->getNextNode() == NextForkCI)
+          continue;
+
+        CreateSequentialRegion(OriginalFn, BB, ForkCI->getNextNode(),
+                               NextForkCI->getPrevNode());
+      }
+
       OpenMPIRBuilder::LocationDescription Loc(InsertPointTy(BB, BB->end()),
                                                DL);
       IRBuilder<>::InsertPoint AllocaIP(
           &OriginalFn->getEntryBlock(),
           OriginalFn->getEntryBlock().getFirstInsertionPt());
       // Create the merged parallel region with default proc binding, to
       // avoid overriding binding settings, and without explicit cancellation.
       InsertPointTy AfterIP = OMPInfoCache.OMPBuilder.createParallel(
           Loc, AllocaIP, BodyGenCB, PrivCB, FiniCB, nullptr, nullptr,
           OMP_PROC_BIND_default, /* IsCancellable */ false);
       BranchInst::Create(AfterBB, AfterIP.getBlock());
 
       // Perform the actual outlining.
-      OMPInfoCache.OMPBuilder.finalize();
+      OMPInfoCache.OMPBuilder.finalize(/* AllowExtractorSinking */ true);
 
       Function *OutlinedFn = MergableCIs.front()->getCaller();
 
       // Replace the __kmpc_fork_call calls with direct calls to the outlined
       // callbacks.
       SmallVector<Value *, 8> Args;
       for (auto *CI : MergableCIs) {
         Value *Callee =
bool mergeParallelRegions() {
     for (auto &It : BB2PRMap) {
       auto &CIs = It.getSecond();
       if (CIs.size() < 2)
         continue;
 
       BasicBlock *BB = It.getFirst();
       SmallVector<CallInst *, 4> MergableCIs;
 
+      /// Returns true if the instruction is mergable, false otherwise.
+      /// A terminator instruction is unmergable by definition since merging
+      /// works within a BB. Instructions before the mergable region are
+      /// mergable if they are not calls to OpenMP runtime functions that may
+      /// set different execution parameters for subsequent parallel regions.
+      /// Instructions in-between parallel regions are mergable if they are not
+      /// calls to any non-intrinsic function since that may call a non-mergable
+      /// OpenMP runtime function.
+      auto IsMergable = [&](Instruction &I, bool IsBeforeMergableRegion) {
+        // We do not merge across BBs, hence return false (unmergable) if the
+        // instruction is a terminator.
+        if (I.isTerminator())
+          return false;
+
+        if (!isa<CallInst>(&I))
+          return true;
+
+        CallInst *CI = cast<CallInst>(&I);
+        if (IsBeforeMergableRegion) {
+          Function *CalledFunction = CI->getCalledFunction();
+          if (!CalledFunction)
+            return false;
+          // Return false (unmergable) if the call before the parallel
+          // region calls an explicit affinity (proc_bind) or number of
+          // threads (num_threads) compiler-generated function. Those settings
+          // may be incompatible with following parallel regions.
+          // TODO: ICV tracking to detect compatibility.
+          for (const auto &RFI : UnmergableCallsInfo) {
+            if (CalledFunction == RFI.Declaration)
+              return false;
+          }
+        } else {
+          // Return false (unmergable) if there is a call instruction
+          // in-between parallel regions when it is not an intrinsic. It
+          // may call an unmergable OpenMP runtime function in its callpath.
+          // TODO: Keep track of possible OpenMP calls in the callpath.
+          if (!isa<IntrinsicInst>(CI))
+            return false;
+        }
+
+        return true;
+      };
       // Find maximal number of parallel region CIs that are safe to merge.
-      for (Instruction &I : *BB) {
+      for (auto It = BB->begin(), End = BB->end(); It != End;) {
+        Instruction &I = *It;
+        ++It;
+
         if (CIs.count(&I)) {
           MergableCIs.push_back(cast<CallInst>(&I));
           continue;
         }
 
-        if (isSafeToSpeculativelyExecute(&I, &I, DT))
+        // Continue expanding if the instruction is mergable.
+        if (IsMergable(I, MergableCIs.empty()))
           continue;
 
+        // Forward the instruction iterator to skip the next parallel region
+        // since there is an unmergable instruction which can affect it.
+        for (; It != End; ++It) {
+          Instruction &SkipI = *It;
+          if (CIs.count(&SkipI)) {
+            LLVM_DEBUG(dbgs() << TAG << "Skip parallel region " << SkipI
+                              << " due to " << I << "\n");
+            ++It;
+            break;
+          }
+        }
+
+        // Store mergable regions found.
         if (MergableCIs.size() > 1) {
           MergableCIsVector.push_back(MergableCIs);
           LLVM_DEBUG(dbgs() << TAG << "Found " << MergableCIs.size()
                             << " parallel regions in block " << BB->getName()
                             << " of function " << BB->getParent()->getName()
                             << "\n";);
         }
 
         MergableCIs.clear();
       }
 
       if (!MergableCIsVector.empty()) {
         Changed = true;
 
         for (auto &MergableCIs : MergableCIsVector)
           Merge(MergableCIs, BB);
       }
     }
 
     if (Changed) {
-      // Update RFI info to set it up for later passes.
-      RFI.clearUsesMap();
-      OMPInfoCache.collectUses(RFI, /* CollectStats */ false);
-
-      // Collect uses for the emitted barrier call.
-      OMPInformationCache::RuntimeFunctionInfo &BarrierRFI =
-          OMPInfoCache.RFIs[OMPRTL___kmpc_barrier];
-      BarrierRFI.clearUsesMap();
-      OMPInfoCache.collectUses(BarrierRFI, /* CollectStats */ false);
+      /// Re-collect use for fork calls, emitted barrier calls, and
+      /// any emitted master/end_master calls.
+      OMPInfoCache.recollectUsesForFunction(OMPRTL___kmpc_fork_call);
+      OMPInfoCache.recollectUsesForFunction(OMPRTL___kmpc_barrier);
+      OMPInfoCache.recollectUsesForFunction(OMPRTL___kmpc_master);
+      OMPInfoCache.recollectUsesForFunction(OMPRTL___kmpc_end_master);
     }
 
     return Changed;
   }
 
   /// Try to delete parallel regions if possible.
   bool deleteParallelRegions() {
     const unsigned CallbackCalleeOperand = 2;
 
     OMPInformationCache::RuntimeFunctionInfo &RFI =
         OMPInfoCache.RFIs[OMPRTL___kmpc_fork_call];
 
     if (!RFI.Declaration)
       return false;

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 -openmpopt -openmp-opt-enable-merging  < %s -enable-new-pm=0 | FileCheck %s
-; RUN: opt -S -passes='attributor,cgscc(openmpopt)' -openmp-opt-enable-merging < %s | FileCheck %s
-
-target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
-
-%struct.ident_t = type { i32, i32, i32, i32, i8* }
-
-@0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
-@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
-
-;   void merge_all() {
-;       int a = 1;
-;   #pragma omp parallel
-;       {
-;           a = 2;
-;       }
-;   #pragma omp parallel
-;       {
-;           a = 3;
-;       }
-;   }
-;
-; Merge all parallel regions.
-define dso_local void @merge_all() local_unnamed_addr  {
-  %1 = alloca i32, align 4
-  %2 = bitcast i32* %1 to i8*
-  store i32 1, i32* %1, align 4
-  %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_all..omp_par to void (i32*, i32*, ...)*), i32* nonnull %1)
-  %4 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_all..omp_par.1 to void (i32*, i32*, ...)*), i32* nonnull %1)
+; RUN: opt -S -attributor -openmpopt -openmp-opt-enable-merging -enable-new-pm=0 < %s | FileCheck %s
+; RUN: opt -S -passes='attributor,cgscc(openmpopt)' -openmp-opt-enable-merging  < %s | FileCheck %s
+; #include <omp.h>
+; void foo();
+; void use(int);
+; void usef(float);
+; void merge(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void unmergable_proc_bind(int a) {
+; #pragma omp parallel proc_bind(close)
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void unmergable_num_threads(int a) {
+; #pragma omp parallel num_threads(a)
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void unmergable_seq_call(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     foo();
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void merge_seq(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     a = a + 1;
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     use(a);
+; }
+; void merge_seq_float(float f, float *p) {
+; #pragma omp parallel
+;     {
+;         use(f);
+;     }
+;     *p = f + 3.14f;
+; #pragma omp parallel
+;     {
+;         use(f);
+;     }
+; }
+; void merge_seq_firstprivate(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     a = a + 1;
+; #pragma omp parallel firstprivate(a)
+;     {
+;         use(a);
+;     }
+;     use(a);
+; }
+; void merge_seq_sink_lt(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     {
+;         int b = (int)&b;
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void merge_seq_par_use(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     int b = a + 1;
+; #pragma omp parallel
+;     {
+;         use(a);
+;         use(b);
+;     }
+; }
+; void merge_cancellable_regions(int cancel1, int cancel2)
+; {
+; #pragma omp parallel
+;     {
+;         if(cancel1) {
+; #pragma omp cancel parallel
+;         }
+;     }
+; #pragma omp parallel
+;     {
+;         if (cancel2) {
+; #pragma omp cancel parallel
+;         }
+;     }
+; }
+; void merge_cancellable_regions_seq(int cancel1, int cancel2)
+; {
+; #pragma omp parallel
+;     {
+;         if(cancel1) {
+; #pragma omp cancel parallel
+;         }
+;     }
+;     cancel2 = !cancel1;
+; #pragma omp parallel
+;     {
+;         if (cancel2) {
+; #pragma omp cancel parallel
+;         }
+;     }
+; }
+; void merge_3(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void merge_3_seq(int a, int b) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     b = a + 1;
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     b = b + a;
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     use(b);
+; }
+; void unmergable_3_seq_call(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     foo();
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     foo();
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void unmergable_3_proc_bind(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel proc_bind(close)
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void unmergable_3_num_threads(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel num_threads(a)
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+; void merge_2_unmergable_1(int a) {
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+;     foo();
+; #pragma omp parallel
+;     {
+;         use(a);
+;     }
+; }
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+
+%struct.ident_t = type { i32, i32, i32, i32, i8* }
+
+@0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
+@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
+
+define dso_local void @merge(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
   ret void
 }
 
-define internal void @merge_all..omp_par.1(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 3, i32* %2, align 4
+define internal void @.omp_outlined.(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-define internal void @merge_all..omp_par(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 2, i32* %2, align 4
+declare dso_local void @use(i32) local_unnamed_addr
+
+declare !callback !1 void @__kmpc_fork_call(%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) local_unnamed_addr
+
+define internal void @.omp_outlined..1(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
+define dso_local void @unmergable_proc_bind(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
+  store i32 %a, i32* %a.addr, align 4
+  call void @__kmpc_push_proc_bind(%struct.ident_t* nonnull @1, i32 %0, i32 3)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..2(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
 
 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr
 
-declare !callback !1 void @__kmpc_fork_call(%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) local_unnamed_addr
+declare void @__kmpc_push_proc_bind(%struct.ident_t*, i32, i32) local_unnamed_addr
 
-;   void merge_none() {
-;       int a = 1;
-;   #pragma omp parallel
-;       {
-;           a = 2;
-;       }
-;       a = 3;
-;   #pragma omp parallel
-;       {
-;           a = 4;
-;       }
-;   }
-;
-; Does not merge parallel regions, in-between store
-; instruction is unsafe to execute in parallel.
-define dso_local void @merge_none() local_unnamed_addr  {
-  %1 = alloca i32, align 4
-  %2 = bitcast i32* %1 to i8*
-  store i32 1, i32* %1, align 4
-  %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_none..omp_par to void (i32*, i32*, ...)*), i32* nonnull %1)
-  store i32 3, i32* %1, align 4
-  %4 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_none..omp_par.2 to void (i32*, i32*, ...)*), i32* nonnull %1)
+define internal void @.omp_outlined..3(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-define internal void @merge_none..omp_par.2(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 4, i32* %2, align 4
+define dso_local void @unmergable_num_threads(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
+  store i32 %a, i32* %a.addr, align 4
+  call void @__kmpc_push_num_threads(%struct.ident_t* nonnull @1, i32 %0, i32 %a)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..4 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..5 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
   ret void
 }
 
-define internal void @merge_none..omp_par(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 2, i32* %2, align 4
+define internal void @.omp_outlined..4(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-;   void merge_some() {
-;       int a = 1;
-;   #pragma omp parallel
-;       {
-;           a = 2;
-;       }
-;       a = 3;
-;   #pragma omp parallel
-;       {
-;           a = 4;
-;       }
-;   #pragma omp parallel
-;       {
-;           a = 5;
-;       }
-;   }
-;
-; Do not merge first parallel region, due to the
-; unsafe store, but merge the two next parallel
-; regions.
-define dso_local void @merge_some() local_unnamed_addr  {
-  %1 = alloca i32, align 4
-  %2 = bitcast i32* %1 to i8*
-  store i32 1, i32* %1, align 4
-  %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_some..omp_par to void (i32*, i32*, ...)*), i32* nonnull %1)
-  store i32 3, i32* %1, align 4
-  %4 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_some..omp_par.3 to void (i32*, i32*, ...)*), i32* nonnull %1)
-  %5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_some..omp_par.4 to void (i32*, i32*, ...)*), i32* nonnull %1)
+declare void @__kmpc_push_num_threads(%struct.ident_t*, i32, i32) local_unnamed_addr
+
+define internal void @.omp_outlined..5(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-define internal void @merge_some..omp_par.4(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 5, i32* %2, align 4
+define dso_local void @unmergable_seq_call(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..6 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (...) @foo()
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..7 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
   ret void
 }
 
-define internal void @merge_some..omp_par.3(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 4, i32* %2, align 4
+define internal void @.omp_outlined..6(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-define internal void @merge_some..omp_par(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture %2)  {
-  store i32 2, i32* %2, align 4
+declare dso_local void @foo(...) local_unnamed_addr
+
+define internal void @.omp_outlined..7(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-;   void merge_cancellable_regions(int cancel1, int cancel2)
-;   {
-;   #pragma omp parallel
-;       {
-;           if(cancel1) {
-;   #pragma omp cancel parallel
-;           }
-;       }
-;   #pragma omp parallel
-;       {
-;           if (cancel2) {
-;   #pragma omp cancel parallel
-;           }
-;       }
-;   }
-;
-; Merge correctly cancellable regions.
-define dso_local void @merge_cancellable_regions(i32 %0, i32 %1) local_unnamed_addr  {
-  %3 = alloca i32, align 4
-  %4 = alloca i32, align 4
-  store i32 %0, i32* %3, align 4
-  store i32 %1, i32* %4, align 4
-  %5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_cancellable_regions..omp_par to void (i32*, i32*, ...)*), i32* nonnull %3)
-  %6 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_cancellable_regions..omp_par.5 to void (i32*, i32*, ...)*), i32* nonnull %4)
+define dso_local void @merge_seq(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %0 = load i32, i32* %a.addr, align 4
+  %add = add nsw i32 %0, 1
+  store i32 %add, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..9 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %1 = load i32, i32* %a.addr, align 4
+  call void @use(i32 %1)
+  ret void
+}
+
+define internal void @.omp_outlined..8(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..9(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @merge_seq_float(float %f, float* nocapture %p) local_unnamed_addr  {
+entry:
+  %f.addr = alloca float, align 4
+  store float %f, float* %f.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*)* @.omp_outlined..10 to void (i32*, i32*, ...)*), float* nonnull %f.addr)
+  %0 = load float, float* %f.addr, align 4
+  %add = fadd float %0, 0x40091EB860000000
+  store float %add, float* %p, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, float*)* @.omp_outlined..11 to void (i32*, i32*, ...)*), float* nonnull %f.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..10(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., float* nocapture nonnull readonly align 4 dereferenceable(4) %f)  {
+entry:
+  %0 = load float, float* %f, align 4
+  %conv = fptosi float %0 to i32
+  call void @use(i32 %conv)
+  ret void
+}
+
+define internal void @.omp_outlined..11(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., float* nocapture nonnull readonly align 4 dereferenceable(4) %f)  {
+entry:
+  %0 = load float, float* %f, align 4
+  %conv = fptosi float %0 to i32
+  call void @use(i32 %conv)
+  ret void
+}
+
+define dso_local void @merge_seq_firstprivate(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %0 = load i32, i32* %a.addr, align 4
+  %add = add nsw i32 %0, 1
+  store i32 %add, i32* %a.addr, align 4
+  %a.casted.sroa.0.0.insert.ext = zext i32 %add to i64
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64)* @.omp_outlined..13 to void (i32*, i32*, ...)*), i64 %a.casted.sroa.0.0.insert.ext)
+  %1 = load i32, i32* %a.addr, align 4
+  call void @use(i32 %1)
+  ret void
+}
+
+define internal void @.omp_outlined..12(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..13(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i64 %a)  {
+entry:
+  %a.addr.sroa.0.0.extract.trunc = trunc i64 %a to i32
+  call void @use(i32 %a.addr.sroa.0.0.extract.trunc)
+  ret void
+}
+
+define dso_local void @merge_seq_sink_lt(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %b = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %0 = bitcast i32* %b to i8*
+  call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %0)
+  %1 = ptrtoint i32* %b to i64
+  %2 = trunc i64 %1 to i32
+  store i32 %2, i32* %b, align 4
+  call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %0)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..15 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..14(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
 }
 
-define internal void @merge_cancellable_regions..omp_par.5(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture readonly %2)  {
-  %4 = load i32, i32* %2, align 4
-  %5 = icmp eq i32 %4, 0
-  br i1 %5, label %6, label %7
+declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture)
+
+declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
 
-6:                                                ; preds = %3
+define internal void @.omp_outlined..15(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
   ret void
+}
 
-7:                                                ; preds = %3
-  %8 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  %9 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %8, i32 1)
+define dso_local void @merge_seq_par_use(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %b = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..16 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %0 = bitcast i32* %b to i8*
+  call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %0)
+  %1 = load i32, i32* %a.addr, align 4
+  %add = add nsw i32 %1, 1
+  store i32 %add, i32* %b, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32*)* @.omp_outlined..17 to void (i32*, i32*, ...)*), i32* nonnull %a.addr, i32* nonnull %b)
+  call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %0)
   ret void
 }
 
-define internal void @merge_cancellable_regions..omp_par(i32* noalias nocapture readnone %0, i32* noalias nocapture readnone %1, i32* nocapture readonly %2)  {
-  %4 = load i32, i32* %2, align 4
-  %5 = icmp eq i32 %4, 0
-  br i1 %5, label %6, label %7
+define internal void @.omp_outlined..16(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
 
-6:                                                ; preds = %3
+define internal void @.omp_outlined..17(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a, i32* nocapture nonnull readonly align 4 dereferenceable(4) %b)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  %1 = load i32, i32* %b, align 4
+  call void @use(i32 %1)
   ret void
+}
 
-7:                                                ; preds = %3
-  %8 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
-  %9 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %8, i32 1)
+define dso_local void @merge_cancellable_regions(i32 %cancel1, i32 %cancel2) local_unnamed_addr  {
+entry:
+  %cancel1.addr = alloca i32, align 4
+  %cancel2.addr = alloca i32, align 4
+  store i32 %cancel1, i32* %cancel1.addr, align 4
+  store i32 %cancel2, i32* %cancel2.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..18 to void (i32*, i32*, ...)*), i32* nonnull %cancel1.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..19 to void (i32*, i32*, ...)*), i32* nonnull %cancel2.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..18(i32* noalias nocapture readonly %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %cancel1)  {
+entry:
+  %0 = load i32, i32* %cancel1, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  br i1 %tobool.not, label %if.end, label %if.then
+
+if.then:                                          ; preds = %entry
+  %1 = load i32, i32* %.global_tid., align 4
+  %2 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %1, i32 1)
+  ret void
+
+if.end:                                           ; preds = %entry
   ret void
 }
 
 declare i32 @__kmpc_cancel(%struct.ident_t*, i32, i32) local_unnamed_addr
 
+define internal void @.omp_outlined..19(i32* noalias nocapture readonly %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %cancel2)  {
+entry:
+  %0 = load i32, i32* %cancel2, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  br i1 %tobool.not, label %if.end, label %if.then
+
+if.then:                                          ; preds = %entry
+  %1 = load i32, i32* %.global_tid., align 4
+  %2 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %1, i32 1)
+  ret void
+
+if.end:                                           ; preds = %entry
+  ret void
+}
+
+define dso_local void @merge_cancellable_regions_seq(i32 %cancel1, i32 %cancel2) local_unnamed_addr  {
+entry:
+  %cancel1.addr = alloca i32, align 4
+  %cancel2.addr = alloca i32, align 4
+  store i32 %cancel1, i32* %cancel1.addr, align 4
+  store i32 %cancel2, i32* %cancel2.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..20 to void (i32*, i32*, ...)*), i32* nonnull %cancel1.addr)
+  %0 = load i32, i32* %cancel1.addr, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  %lnot.ext = zext i1 %tobool.not to i32
+  store i32 %lnot.ext, i32* %cancel2.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..21 to void (i32*, i32*, ...)*), i32* nonnull %cancel2.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..20(i32* noalias nocapture readonly %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %cancel1)  {
+entry:
+  %0 = load i32, i32* %cancel1, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  br i1 %tobool.not, label %if.end, label %if.then
+
+if.then:                                          ; preds = %entry
+  %1 = load i32, i32* %.global_tid., align 4
+  %2 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %1, i32 1)
+  ret void
+
+if.end:                                           ; preds = %entry
+  ret void
+}
+
+define internal void @.omp_outlined..21(i32* noalias nocapture readonly %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %cancel2)  {
+entry:
+  %0 = load i32, i32* %cancel2, align 4
+  %tobool.not = icmp eq i32 %0, 0
+  br i1 %tobool.not, label %if.end, label %if.then
+
+if.then:                                          ; preds = %entry
+  %1 = load i32, i32* %.global_tid., align 4
+  %2 = call i32 @__kmpc_cancel(%struct.ident_t* nonnull @1, i32 %1, i32 1)
+  ret void
+
+if.end:                                           ; preds = %entry
+  ret void
+}
+
+define dso_local void @merge_3(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..22 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..23 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..24 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..22(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..23(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..24(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @merge_3_seq(i32 %a, i32 %b) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..25 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %0 = load i32, i32* %a.addr, align 4
+  %add = add nsw i32 %0, 1
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..26 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %1 = load i32, i32* %a.addr, align 4
+  %add1 = add nsw i32 %add, %1
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..27 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void @use(i32 %add1)
+  ret void
+}
+
+define internal void @.omp_outlined..25(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..26(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..27(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @unmergable_3_seq_call(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..28 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (...) @foo()
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..29 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (...) @foo()
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..30 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..28(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..29(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..30(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @unmergable_3_proc_bind(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..31 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void @__kmpc_push_proc_bind(%struct.ident_t* nonnull @1, i32 %0, i32 3)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..32 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..33 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..31(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..32(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..33(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @unmergable_3_num_threads(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..34 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  %1 = load i32, i32* %a.addr, align 4
+  call void @__kmpc_push_num_threads(%struct.ident_t* nonnull @1, i32 %0, i32 %1)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..35 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..36 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..34(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..35(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..36(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define dso_local void @merge_2_unmergable_1(i32 %a) local_unnamed_addr  {
+entry:
+  %a.addr = alloca i32, align 4
+  store i32 %a, i32* %a.addr, align 4
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..37 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..38 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  call void (...) @foo()
+  call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @.omp_outlined..39 to void (i32*, i32*, ...)*), i32* nonnull %a.addr)
+  ret void
+}
+
+define internal void @.omp_outlined..37(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..38(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
+define internal void @.omp_outlined..39(i32* noalias nocapture readnone %.global_tid., i32* noalias nocapture readnone %.bound_tid., i32* nocapture nonnull readonly align 4 dereferenceable(4) %a)  {
+entry:
+  %0 = load i32, i32* %a, align 4
+  call void @use(i32 %0)
+  ret void
+}
+
 
 !llvm.module.flags = !{!0}
 
 !0 = !{i32 1, !"wchar_size", i32 4}
 !1 = !{!2}
 !2 = !{i64 2, i64 -1, i64 -1, i1 true}
-; CHECK-LABEL: define {{[^@]+}}@merge_all() local_unnamed_addr {
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1:@.*]])
-; CHECK-NEXT:    [[TMP2:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    store i32 1, i32* [[TMP2]], align 4
-; CHECK-NEXT:    [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* [[GLOB1]])
+; 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_all..omp_par.2 to void (i32*, i32*, ...)*), i32* [[TMP2]])
+; 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 [[DOTSPLIT_SPLIT:%.*]]
-; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
 ;
 ;
-; CHECK-LABEL: define {{[^@]+}}@merge_all..omp_par.2
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[TMP0:%.*]]) [[ATTR0:#.*]] {
+; 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:    [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
-; CHECK-NEXT:    store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], 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 @merge_all..omp_par(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP0]])
+; 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 @merge_all..omp_par.1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP0]])
-; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
-; CHECK:       .split:
+; 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:    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 {{[^@]+}}@merge_all..omp_par.1
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1:#.*]] {
-; CHECK-NEXT:    store i32 3, i32* [[TMP2]], align 4
+; 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-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-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 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:    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-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_all..omp_par
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 2, i32* [[TMP2]], align 4
+; 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-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_none() local_unnamed_addr {
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
-; CHECK-NEXT:    [[TMP2:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    store i32 1, i32* [[TMP2]], 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*)* @merge_none..omp_par to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2]])
-; CHECK-NEXT:    store i32 3, i32* [[TMP2]], 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*)* @merge_none..omp_par.2 to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2]])
+; 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 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:    ret void
 ;
 ;
-; CHECK-LABEL: define {{[^@]+}}@merge_none..omp_par.2
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 4, i32* [[TMP2]], align 4
+; 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-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_none..omp_par
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 2, i32* [[TMP2]], align 4
+; 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-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_some() local_unnamed_addr {
-; CHECK-NEXT:    [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
-; CHECK-NEXT:    [[TMP2:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    store i32 1, i32* [[TMP2]], 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*)* @merge_some..omp_par to void (i32*, i32*, ...)*), i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2]])
-; CHECK-NEXT:    store i32 3, i32* [[TMP2]], align 4
+; 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 (...) @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:    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-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-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 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 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 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-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-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 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*)* @merge_some..omp_par.5 to void (i32*, i32*, ...)*), i32* [[TMP2]])
+; 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 [[DOTSPLIT_SPLIT:%.*]]
-; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
 ;
 ;
-; CHECK-LABEL: define {{[^@]+}}@merge_some..omp_par.5
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[TMP0:%.*]]) [[ATTR0]] {
+; 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:    [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
-; CHECK-NEXT:    store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], 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 @merge_some..omp_par.3(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP0]])
+; 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:    call void @__kmpc_barrier(%struct.ident_t* [[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
-; CHECK-NEXT:    call void @merge_some..omp_par.4(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP0]])
-; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
-; CHECK:       .split:
+; 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 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 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-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 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 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_some..omp_par.4
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 5, i32* [[TMP2]], align 4
+; 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_some..omp_par.3
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 4, i32* [[TMP2]], align 4
+; 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_some..omp_par
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture nofree noundef nonnull writeonly align 4 dereferenceable(4) [[TMP2:%.*]]) [[ATTR1]] {
-; CHECK-NEXT:    store i32 2, i32* [[TMP2]], align 4
+; 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 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 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 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-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 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 noundef nonnull readonly align 4 dereferenceable(4) [[A_ADDR]], i32* nocapture 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 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-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 [[TMP0:%.*]], i32 [[TMP1:%.*]]) local_unnamed_addr {
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull align 8 dereferenceable(24) [[GLOB1]])
-; CHECK-NEXT:    [[TMP4:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    [[TMP5:%.*]] = alloca i32, align 4
-; CHECK-NEXT:    store i32 [[TMP0]], i32* [[TMP4]], align 4
-; CHECK-NEXT:    store i32 [[TMP1]], i32* [[TMP5]], align 4
+; 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.6 to void (i32*, i32*, ...)*), i32* [[TMP4]], i32* [[TMP5]])
+; 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 [[DOTSPLIT_SPLIT:%.*]]
-; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[ENTRY_SPLIT_SPLIT:%.*]]
+; CHECK:       entry.split.split:
 ; CHECK-NEXT:    ret void
 ;
 ;
-; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par.6
-; CHECK-SAME: (i32* noalias [[TID_ADDR:%.*]], i32* noalias [[ZERO_ADDR:%.*]], i32* [[TMP0:%.*]], i32* [[TMP1:%.*]]) [[ATTR0]] {
+; 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:    [[TMP2:%.*]] = load i32, i32* [[TID_ADDR]], align 4
-; CHECK-NEXT:    store i32 [[TMP2]], i32* [[TID_ADDR_LOCAL]], 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 @merge_cancellable_regions..omp_par(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[TMP0]])
+; 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 @merge_cancellable_regions..omp_par.5(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[TMP1]])
-; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
-; CHECK:       .split:
+; 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:    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 {{[^@]+}}@merge_cancellable_regions..omp_par.5
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[TMP2:%.*]]) {
-; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
-; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[TMP4]], 0
-; CHECK-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
-; CHECK:       6:
+; 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-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:       7:
-; CHECK-NEXT:    [[TMP8:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull [[GLOB1]])
-; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP8]], i32 noundef 1)
+; CHECK:       if.end:
 ; CHECK-NEXT:    ret void
 ;
 ;
-; CHECK-LABEL: define {{[^@]+}}@merge_cancellable_regions..omp_par
-; CHECK-SAME: (i32* noalias nocapture nofree readnone [[TMP0:%.*]], i32* noalias nocapture nofree readnone [[TMP1:%.*]], i32* nocapture noundef nonnull readonly align 4 dereferenceable(4) [[TMP2:%.*]]) {
-; CHECK-NEXT:    [[TMP4:%.*]] = load i32, i32* [[TMP2]], align 4
-; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[TMP4]], 0
-; CHECK-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP7:%.*]]
-; CHECK:       6:
-; CHECK-NEXT:    ret void
-; CHECK:       7:
-; CHECK-NEXT:    [[TMP8:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull [[GLOB1]])
-; CHECK-NEXT:    [[TMP9:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* noundef nonnull [[GLOB1]], i32 [[TMP8]], i32 noundef 1)
+; 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-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 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 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 readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture 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 readonly [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], i32* nocapture 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 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:    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-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-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-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 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 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 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 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-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-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 (...) @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 (...) @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:    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-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-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-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 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:    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-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-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-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 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:    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-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-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-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 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 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:    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-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-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-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* [[A]], align 4
+; CHECK-NEXT:    call void @use(i32 [[TMP0]])
 ; CHECK-NEXT:    ret void
 ;