Transforms/Utils/LoopSimplify.cpp

//===- LoopSimplify.cpp - Loop Canonicalization Pass ----------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This pass performs several transformations to transform natural loops into a
// simpler form, which makes subsequent analyses and transformations simpler and
// more effective.
//
// Loop pre-header insertion guarantees that there is a single, non-critical
// entry edge from outside of the loop to the loop header.  This simplifies a
// number of analyses and transformations, such as LICM.
//
// Loop exit-block insertion guarantees that all exit blocks from the loop
// (blocks which are outside of the loop that have predecessors inside of the
// loop) only have predecessors from inside of the loop (and are thus dominated
// by the loop header).  This simplifies transformations such as store-sinking
// that are built into LICM.
//
// This pass also guarantees that loops will have exactly one backedge.
//
// Indirectbr instructions introduce several complications. If the loop
// contains or is entered by an indirectbr instruction, it may not be possible
// to transform the loop and make these guarantees. Client code should check
// that these conditions are true before relying on them.
//
// Similar complications arise from callbr instructions, particularly in
// asm-goto where blockaddress expressions are used.
//
// Note that the simplifycfg pass will clean up blocks which are split out but
// end up being unnecessary, so usage of this pass should not pessimize
// generated code.
//
// This pass obviously modifies the CFG, but updates loop information and
// dominator information.
//
//===----------------------------------------------------------------------===//

#include "llvm/Transforms/Utils/LoopSimplify.h"
#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/SetOperations.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AssumptionCache.h"
#include "llvm/Analysis/BasicAliasAnalysis.h"
#include "llvm/Analysis/BranchProbabilityInfo.h"
#include "llvm/Analysis/DependenceAnalysis.h"
#include "llvm/Analysis/GlobalsModRef.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/MemorySSA.h"
#include "llvm/Analysis/MemorySSAUpdater.h"
#include "llvm/Analysis/ScalarEvolution.h"
#include "llvm/Analysis/ScalarEvolutionAliasAnalysis.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/Type.h"
#include "llvm/InitializePasses.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Transforms/Utils.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/LoopUtils.h"
using namespace llvm;

#define DEBUG_TYPE "loop-simplify"

STATISTIC(NumNested  , "Number of nested loops split out");

// If the block isn't already, move the new block to right after some 'outside
// block' block.  This prevents the preheader from being placed inside the loop
// body, e.g. when the loop hasn't been rotated.
static void placeSplitBlockCarefully(BasicBlock *NewBB,
                                     SmallVectorImpl<BasicBlock *> &SplitPreds,
                                     Loop *L) {
  // Check to see if NewBB is already well placed.
  Function::iterator BBI = --NewBB->getIterator();
  for (unsigned i = 0, e = SplitPreds.size(); i != e; ++i) {
    if (&*BBI == SplitPreds[i])
      return;
  }

  // If it isn't already after an outside block, move it after one.  This is
  // always good as it makes the uncond branch from the outside block into a
  // fall-through.

  // Figure out *which* outside block to put this after.  Prefer an outside
  // block that neighbors a BB actually in the loop.
  BasicBlock *FoundBB = nullptr;
  for (unsigned i = 0, e = SplitPreds.size(); i != e; ++i) {
    Function::iterator BBI = SplitPreds[i]->getIterator();
    if (++BBI != NewBB->getParent()->end() && L->contains(&*BBI)) {
      FoundBB = SplitPreds[i];
      break;
    }
  }

  // If our heuristic for a *good* bb to place this after doesn't find
  // anything, just pick something.  It's likely better than leaving it within
  // the loop.
  if (!FoundBB)
    FoundBB = SplitPreds[0];
  NewBB->moveAfter(FoundBB);
}

/// InsertPreheaderForLoop - Once we discover that a loop doesn't have a
/// preheader, this method is called to insert one.  This method has two phases:
/// preheader insertion and analysis updating.
///
BasicBlock *llvm::InsertPreheaderForLoop(Loop *L, DominatorTree *DT,
                                         LoopInfo *LI, MemorySSAUpdater *MSSAU,
                                         bool PreserveLCSSA) {
  BasicBlock *Header = L->getHeader();

  // Compute the set of predecessors of the loop that are not in the loop.
  SmallVector<BasicBlock*, 8> OutsideBlocks;
  for (BasicBlock *P : predecessors(Header)) {
    if (!L->contains(P)) {         // Coming in from outside the loop?
      // If the loop is branched to from an indirect terminator, we won't
      // be able to fully transform the loop, because it prohibits
      // edge splitting.
      if (P->getTerminator()->isIndirectTerminator())
        return nullptr;

      // Keep track of it.
      OutsideBlocks.push_back(P);
    }
  }

  // Split out the loop pre-header.
  BasicBlock *PreheaderBB;
  PreheaderBB = SplitBlockPredecessors(Header, OutsideBlocks, ".preheader", DT,
                                       LI, MSSAU, PreserveLCSSA);
  if (!PreheaderBB)
    return nullptr;

  LLVM_DEBUG(dbgs() << "LoopSimplify: Creating pre-header "
                    << PreheaderBB->getName() << "\n");

  // Make sure that NewBB is put someplace intelligent, which doesn't mess up
  // code layout too horribly.
  placeSplitBlockCarefully(PreheaderBB, OutsideBlocks, L);

  return PreheaderBB;
}

/// Add the specified block, and all of its predecessors, to the specified set,
/// if it's not already in there.  Stop predecessor traversal when we reach
/// StopBlock.
static void addBlockAndPredsToSet(BasicBlock *InputBB, BasicBlock *StopBlock,
                                  SmallPtrSetImpl<BasicBlock *> &Blocks) {
  SmallVector<BasicBlock *, 8> Worklist;
  Worklist.push_back(InputBB);
  do {
    BasicBlock *BB = Worklist.pop_back_val();
    if (Blocks.insert(BB).second && BB != StopBlock)
      // If BB is not already processed and it is not a stop block then
      // insert its predecessor in the work list
      append_range(Worklist, predecessors(BB));
  } while (!Worklist.empty());
}

/// The first part of loop-nestification is to find a PHI node that tells
/// us how to partition the loops.
static PHINode *findPHIToPartitionLoops(Loop *L, DominatorTree *DT,
                                        AssumptionCache *AC) {
  const DataLayout &DL = L->getHeader()->getModule()->getDataLayout();
  for (BasicBlock::iterator I = L->getHeader()->begin(); isa<PHINode>(I); ) {
    PHINode *PN = cast<PHINode>(I);
    ++I;
    if (Value *V = SimplifyInstruction(PN, {DL, nullptr, DT, AC})) {
      // This is a degenerate PHI already, don't modify it!
      PN->replaceAllUsesWith(V);
      PN->eraseFromParent();
      continue;
    }

    // Scan this PHI node looking for a use of the PHI node by itself.
    for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
      if (PN->getIncomingValue(i) == PN &&
          L->contains(PN->getIncomingBlock(i)))
        // We found something tasty to remove.
        return PN;
  }
  return nullptr;
}

/// If this loop has multiple backedges, try to pull one of them out into
/// a nested loop.
///
/// This is important for code that looks like
/// this:
///
///  Loop:
///     ...
///     br cond, Loop, Next
///     ...
///     br cond2, Loop, Out
///
/// To identify this common case, we look at the PHI nodes in the header of the
/// loop.  PHI nodes with unchanging values on one backedge correspond to values
/// that change in the "outer" loop, but not in the "inner" loop.
///
/// If we are able to separate out a loop, return the new outer loop that was
/// created.
///
static Loop *separateNestedLoop(Loop *L, BasicBlock *Preheader,
                                DominatorTree *DT, LoopInfo *LI,
                                ScalarEvolution *SE, bool PreserveLCSSA,
                                AssumptionCache *AC, MemorySSAUpdater *MSSAU) {
  // Don't try to separate loops without a preheader.
  if (!Preheader)
    return nullptr;

  // Treat the presence of convergent functions conservatively. The
  // transformation is invalid if calls to certain convergent
  // functions (like an AMDGPU barrier) get included in the resulting
  // inner loop. But blocks meant for the inner loop will be
  // identified later at a point where it's too late to abort the
  // transformation. Also, the convergent attribute is not really
  // sufficient to express the semantics of functions that are
  // affected by this transformation. So we choose to back off if such
  // a function call is present until a better alternative becomes
  // available. This is similar to the conservative treatment of
  // convergent function calls in GVNHoist and JumpThreading.
  for (auto BB : L->blocks()) {
    for (auto &II : *BB) {
      if (auto CI = dyn_cast<CallBase>(&II)) {
        if (CI->isConvergent()) {
          return nullptr;
        }
      }
    }
  }

  // The header is not a landing pad; preheader insertion should ensure this.
  BasicBlock *Header = L->getHeader();
  assert(!Header->isEHPad() && "Can't insert backedge to EH pad");

  PHINode *PN = findPHIToPartitionLoops(L, DT, AC);
  if (!PN) return nullptr;  // No known way to partition.

  // Pull out all predecessors that have varying values in the loop.  This
  // handles the case when a PHI node has multiple instances of itself as
  // arguments.
  SmallVector<BasicBlock*, 8> OuterLoopPreds;
  for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
    if (PN->getIncomingValue(i) != PN ||
        !L->contains(PN->getIncomingBlock(i))) {
      // We can't split indirect control flow edges.
      if (PN->getIncomingBlock(i)->getTerminator()->isIndirectTerminator())
        return nullptr;
      OuterLoopPreds.push_back(PN->getIncomingBlock(i));
    }
  }
  LLVM_DEBUG(dbgs() << "LoopSimplify: Splitting out a new outer loop\n");

  // If ScalarEvolution is around and knows anything about values in
  // this loop, tell it to forget them, because we're about to
  // substantially change it.
  if (SE)
    SE->forgetLoop(L);

  BasicBlock *NewBB = SplitBlockPredecessors(Header, OuterLoopPreds, ".outer",
                                             DT, LI, MSSAU, PreserveLCSSA);

  // Make sure that NewBB is put someplace intelligent, which doesn't mess up
  // code layout too horribly.
  placeSplitBlockCarefully(NewBB, OuterLoopPreds, L);

  // Create the new outer loop.
  Loop *NewOuter = LI->AllocateLoop();

  // Change the parent loop to use the outer loop as its child now.
  if (Loop *Parent = L->getParentLoop())
    Parent->replaceChildLoopWith(L, NewOuter);
  else
    LI->changeTopLevelLoop(L, NewOuter);

  // L is now a subloop of our outer loop.
  NewOuter->addChildLoop(L);

  for (BasicBlock *BB : L->blocks())
    NewOuter->addBlockEntry(BB);

  // Now reset the header in L, which had been moved by
  // SplitBlockPredecessors for the outer loop.
  L->moveToHeader(Header);

  // Determine which blocks should stay in L and which should be moved out to
  // the Outer loop now.
  SmallPtrSet<BasicBlock *, 4> BlocksInL;
  for (BasicBlock *P : predecessors(Header)) {
    if (DT->dominates(Header, P))
      addBlockAndPredsToSet(P, Header, BlocksInL);
  }

  // Scan all of the loop children of L, moving them to OuterLoop if they are
  // not part of the inner loop.
  const std::vector<Loop*> &SubLoops = L->getSubLoops();
  for (size_t I = 0; I != SubLoops.size(); )
    if (BlocksInL.count(SubLoops[I]->getHeader()))
      ++I;   // Loop remains in L
    else
      NewOuter->addChildLoop(L->removeChildLoop(SubLoops.begin() + I));

  SmallVector<BasicBlock *, 8> OuterLoopBlocks;
  OuterLoopBlocks.push_back(NewBB);
  // Now that we know which blocks are in L and which need to be moved to
  // OuterLoop, move any blocks that need it.
  for (unsigned i = 0; i != L->getBlocks().size(); ++i) {
    BasicBlock *BB = L->getBlocks()[i];
    if (!BlocksInL.count(BB)) {
      // Move this block to the parent, updating the exit blocks sets
      L->removeBlockFromLoop(BB);
      if ((*LI)[BB] == L) {
        LI->changeLoopFor(BB, NewOuter);
        OuterLoopBlocks.push_back(BB);
      }
      --i;
    }
  }

  // Split edges to exit blocks from the inner loop, if they emerged in the
  // process of separating the outer one.
  formDedicatedExitBlocks(L, DT, LI, MSSAU, PreserveLCSSA);

  if (PreserveLCSSA) {
    // Fix LCSSA form for L. Some values, which previously were only used inside
    // L, can now be used in NewOuter loop. We need to insert phi-nodes for them
    // in corresponding exit blocks.
    // We don't need to form LCSSA recursively, because there cannot be uses
    // inside a newly created loop of defs from inner loops as those would
    // already be a use of an LCSSA phi node.
    formLCSSA(*L, *DT, LI, SE);

    assert(NewOuter->isRecursivelyLCSSAForm(*DT, *LI) &&
           "LCSSA is broken after separating nested loops!");
  }

  return NewOuter;
}

/// This method is called when the specified loop has more than one
/// backedge in it.
///
/// If this occurs, revector all of these backedges to target a new basic block
/// and have that block branch to the loop header.  This ensures that loops
/// have exactly one backedge.
static BasicBlock *insertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader,
                                             DominatorTree *DT, LoopInfo *LI,
                                             MemorySSAUpdater *MSSAU) {
  assert(L->getNumBackEdges() > 1 && "Must have > 1 backedge!");

  // Get information about the loop
  BasicBlock *Header = L->getHeader();
  Function *F = Header->getParent();

  // Unique backedge insertion currently depends on having a preheader.
  if (!Preheader)
    return nullptr;

  // The header is not an EH pad; preheader insertion should ensure this.
  assert(!Header->isEHPad() && "Can't insert backedge to EH pad");

  // Figure out which basic blocks contain back-edges to the loop header.
  std::vector<BasicBlock*> BackedgeBlocks;
  for (BasicBlock *P : predecessors(Header)) {
    // Indirect edges cannot be split, so we must fail if we find one.
    if (P->getTerminator()->isIndirectTerminator())
      return nullptr;

    if (P != Preheader) BackedgeBlocks.push_back(P);
  }

  // Create and insert the new backedge block...
  BasicBlock *BEBlock = BasicBlock::Create(Header->getContext(),
                                           Header->getName() + ".backedge", F);
  BranchInst *BETerminator = BranchInst::Create(Header, BEBlock);
  BETerminator->setDebugLoc(Header->getFirstNonPHI()->getDebugLoc());

  LLVM_DEBUG(dbgs() << "LoopSimplify: Inserting unique backedge block "
                    << BEBlock->getName() << "\n");

  // Move the new backedge block to right after the last backedge block.
  Function::iterator InsertPos = ++BackedgeBlocks.back()->getIterator();
  F->getBasicBlockList().splice(InsertPos, F->getBasicBlockList(), BEBlock);

  // Now that the block has been inserted into the function, create PHI nodes in
  // the backedge block which correspond to any PHI nodes in the header block.
  for (BasicBlock::iterator I = Header->begin(); isa<PHINode>(I); ++I) {
    PHINode *PN = cast<PHINode>(I);
    PHINode *NewPN = PHINode::Create(PN->getType(), BackedgeBlocks.size(),
                                     PN->getName()+".be", BETerminator);

    // Loop over the PHI node, moving all entries except the one for the
    // preheader over to the new PHI node.
    unsigned PreheaderIdx = ~0U;
    bool HasUniqueIncomingValue = true;
    Value *UniqueValue = nullptr;
    for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
      BasicBlock *IBB = PN->getIncomingBlock(i);
      Value *IV = PN->getIncomingValue(i);
      if (IBB == Preheader) {
        PreheaderIdx = i;
      } else {
        NewPN->addIncoming(IV, IBB);
        if (HasUniqueIncomingValue) {
          if (!UniqueValue)
            UniqueValue = IV;
          else if (UniqueValue != IV)
            HasUniqueIncomingValue = false;
        }
      }
    }

    // Delete all of the incoming values from the old PN except the preheader's
    assert(PreheaderIdx != ~0U && "PHI has no preheader entry??");
    if (PreheaderIdx != 0) {
      PN->setIncomingValue(0, PN->getIncomingValue(PreheaderIdx));
      PN->setIncomingBlock(0, PN->getIncomingBlock(PreheaderIdx));
    }
    // Nuke all entries except the zero'th.
    for (unsigned i = 0, e = PN->getNumIncomingValues()-1; i != e; ++i)
      PN->removeIncomingValue(e-i, false);

    // Finally, add the newly constructed PHI node as the entry for the BEBlock.
    PN->addIncoming(NewPN, BEBlock);

    // As an optimization, if all incoming values in the new PhiNode (which is a
    // subset of the incoming values of the old PHI node) have the same value,
    // eliminate the PHI Node.
    if (HasUniqueIncomingValue) {
      NewPN->replaceAllUsesWith(UniqueValue);
      BEBlock->getInstList().erase(NewPN);
    }
  }

  // Now that all of the PHI nodes have been inserted and adjusted, modify the
  // backedge blocks to jump to the BEBlock instead of the header.
  // If one of the backedges has llvm.loop metadata attached, we remove
  // it from the backedge and add it to BEBlock.
  unsigned LoopMDKind = BEBlock->getContext().getMDKindID("llvm.loop");
  MDNode *LoopMD = nullptr;
  for (unsigned i = 0, e = BackedgeBlocks.size(); i != e; ++i) {
    Instruction *TI = BackedgeBlocks[i]->getTerminator();
    if (!LoopMD)
      LoopMD = TI->getMetadata(LoopMDKind);
    TI->setMetadata(LoopMDKind, nullptr);
    TI->replaceSuccessorWith(Header, BEBlock);
  }
  BEBlock->getTerminator()->setMetadata(LoopMDKind, LoopMD);

  //===--- Update all analyses which we must preserve now -----------------===//

  // Update Loop Information - we know that this block is now in the current
  // loop and all parent loops.
  L->addBasicBlockToLoop(BEBlock, *LI);

  // Update dominator information
  DT->splitBlock(BEBlock);

  if (MSSAU)
    MSSAU->updatePhisWhenInsertingUniqueBackedgeBlock(Header, Preheader,
                                                      BEBlock);

  return BEBlock;
}

/// Simplify one loop and queue further loops for simplification.
static bool simplifyOneLoop(Loop *L, SmallVectorImpl<Loop *> &Worklist,
                            DominatorTree *DT, LoopInfo *LI,
                            ScalarEvolution *SE, AssumptionCache *AC,
                            MemorySSAUpdater *MSSAU, bool PreserveLCSSA) {
  bool Changed = false;
  if (MSSAU && VerifyMemorySSA)
    MSSAU->getMemorySSA()->verifyMemorySSA();

ReprocessLoop:

  // Check to see that no blocks (other than the header) in this loop have
  // predecessors that are not in the loop.  This is not valid for natural
  // loops, but can occur if the blocks are unreachable.  Since they are
  // unreachable we can just shamelessly delete those CFG edges!
  for (BasicBlock *BB : L->blocks()) {
    if (BB == L->getHeader())
      continue;

    SmallPtrSet<BasicBlock*, 4> BadPreds;
    for (BasicBlock *P : predecessors(BB))
      if (!L->contains(P))
        BadPreds.insert(P);

    // Delete each unique out-of-loop (and thus dead) predecessor.
    for (BasicBlock *P : BadPreds) {

      LLVM_DEBUG(dbgs() << "LoopSimplify: Deleting edge from dead predecessor "
                        << P->getName() << "\n");

      // Zap the dead pred's terminator and replace it with unreachable.
      Instruction *TI = P->getTerminator();
      changeToUnreachable(TI, PreserveLCSSA,
                          /*DTU=*/nullptr, MSSAU);
      Changed = true;
    }
  }

  if (MSSAU && VerifyMemorySSA)
    MSSAU->getMemorySSA()->verifyMemorySSA();

  // If there are exiting blocks with branches on undef, resolve the undef in
  // the direction which will exit the loop. This will help simplify loop
  // trip count computations.
  SmallVector<BasicBlock*, 8> ExitingBlocks;
  L->getExitingBlocks(ExitingBlocks);
  for (BasicBlock *ExitingBlock : ExitingBlocks)
    if (BranchInst *BI = dyn_cast<BranchInst>(ExitingBlock->getTerminator()))
      if (BI->isConditional()) {
        if (UndefValue *Cond = dyn_cast<UndefValue>(BI->getCondition())) {

          LLVM_DEBUG(dbgs()
                     << "LoopSimplify: Resolving \"br i1 undef\" to exit in "
                     << ExitingBlock->getName() << "\n");

          BI->setCondition(ConstantInt::get(Cond->getType(),
                                            !L->contains(BI->getSuccessor(0))));

          Changed = true;
        }
      }

  // Does the loop already have a preheader?  If so, don't insert one.
  BasicBlock *Preheader = L->getLoopPreheader();
  if (!Preheader) {
    Preheader = InsertPreheaderForLoop(L, DT, LI, MSSAU, PreserveLCSSA);
    if (Preheader)
      Changed = true;
  }

  // Next, check to make sure that all exit nodes of the loop only have
  // predecessors that are inside of the loop.  This check guarantees that the
  // loop preheader/header will dominate the exit blocks.  If the exit block has
  // predecessors from outside of the loop, split the edge now.
  if (formDedicatedExitBlocks(L, DT, LI, MSSAU, PreserveLCSSA))
    Changed = true;

  if (MSSAU && VerifyMemorySSA)
    MSSAU->getMemorySSA()->verifyMemorySSA();

  // If the header has more than two predecessors at this point (from the
  // preheader and from multiple backedges), we must adjust the loop.
  BasicBlock *LoopLatch = L->getLoopLatch();
  if (!LoopLatch) {
    // If this is really a nested loop, rip it out into a child loop.  Don't do
    // this for loops with a giant number of backedges, just factor them into a
    // common backedge instead.
    if (L->getNumBackEdges() < 8) {
      if (Loop *OuterL = separateNestedLoop(L, Preheader, DT, LI, SE,
                                            PreserveLCSSA, AC, MSSAU)) {
        ++NumNested;
        // Enqueue the outer loop as it should be processed next in our
        // depth-first nest walk.
        Worklist.push_back(OuterL);

        // This is a big restructuring change, reprocess the whole loop.
        Changed = true;
        // GCC doesn't tail recursion eliminate this.
        // FIXME: It isn't clear we can't rely on LLVM to TRE this.
        goto ReprocessLoop;
      }
    }

    // If we either couldn't, or didn't want to, identify nesting of the loops,
    // insert a new block that all backedges target, then make it jump to the
    // loop header.
    LoopLatch = insertUniqueBackedgeBlock(L, Preheader, DT, LI, MSSAU);
    if (LoopLatch)
      Changed = true;
  }

  if (MSSAU && VerifyMemorySSA)
    MSSAU->getMemorySSA()->verifyMemorySSA();

  const DataLayout &DL = L->getHeader()->getModule()->getDataLayout();

  // Scan over the PHI nodes in the loop header.  Since they now have only two
  // incoming values (the loop is canonicalized), we may have simplified the PHI
  // down to 'X = phi [X, Y]', which should be replaced with 'Y'.
  PHINode *PN;
  for (BasicBlock::iterator I = L->getHeader()->begin();
       (PN = dyn_cast<PHINode>(I++)); )
    if (Value *V = SimplifyInstruction(PN, {DL, nullptr, DT, AC})) {
      if (SE) SE->forgetValue(PN);
      if (!PreserveLCSSA || LI->replacementPreservesLCSSAForm(PN, V)) {
        PN->replaceAllUsesWith(V);
        PN->eraseFromParent();
        Changed = true;
      }
    }

  // If this loop has multiple exits and the exits all go to the same
  // block, attempt to merge the exits. This helps several passes, such
  // as LoopRotation, which do not support loops with multiple exits.
  // SimplifyCFG also does this (and this code uses the same utility
  // function), however this code is loop-aware, where SimplifyCFG is
  // not. That gives it the advantage of being able to hoist
  // loop-invariant instructions out of the way to open up more
  // opportunities, and the disadvantage of having the responsibility
  // to preserve dominator information.
  auto HasUniqueExitBlock = [&]() {
    BasicBlock *UniqueExit = nullptr;
    for (auto *ExitingBB : ExitingBlocks)
      for (auto *SuccBB : successors(ExitingBB)) {
        if (L->contains(SuccBB))
          continue;

        if (!UniqueExit)
          UniqueExit = SuccBB;
        else if (UniqueExit != SuccBB)
          return false;
      }

    return true;
  };
  if (HasUniqueExitBlock()) {
    for (unsigned i = 0, e = ExitingBlocks.size(); i != e; ++i) {
      BasicBlock *ExitingBlock = ExitingBlocks[i];
      if (!ExitingBlock->getSinglePredecessor()) continue;
      BranchInst *BI = dyn_cast<BranchInst>(ExitingBlock->getTerminator());
      if (!BI || !BI->isConditional()) continue;
      CmpInst *CI = dyn_cast<CmpInst>(BI->getCondition());
      if (!CI || CI->getParent() != ExitingBlock) continue;

      // Attempt to hoist out all instructions except for the
      // comparison and the branch.
      bool AllInvariant = true;
      bool AnyInvariant = false;
      for (auto I = ExitingBlock->instructionsWithoutDebug().begin(); &*I != BI; ) {
        Instruction *Inst = &*I++;
        if (Inst == CI)
          continue;
        if (!L->makeLoopInvariant(
                Inst, AnyInvariant,
                Preheader ? Preheader->getTerminator() : nullptr, MSSAU)) {
          AllInvariant = false;
          break;
        }
      }
      if (AnyInvariant) {
        Changed = true;
        // The loop disposition of all SCEV expressions that depend on any
        // hoisted values have also changed.
        if (SE)
          SE->forgetLoopDispositions(L);
      }
      if (!AllInvariant) continue;

      // The block has now been cleared of all instructions except for
      // a comparison and a conditional branch. SimplifyCFG may be able
      // to fold it now.
      if (!FoldBranchToCommonDest(BI, /*DTU=*/nullptr, MSSAU))
        continue;

      // Success. The block is now dead, so remove it from the loop,
      // update the dominator tree and delete it.
      LLVM_DEBUG(dbgs() << "LoopSimplify: Eliminating exiting block "
                        << ExitingBlock->getName() << "\n");

      assert(pred_empty(ExitingBlock));
      Changed = true;
      LI->removeBlock(ExitingBlock);

      DomTreeNode *Node = DT->getNode(ExitingBlock);
      while (!Node->isLeaf()) {
        DomTreeNode *Child = Node->back();
        DT->changeImmediateDominator(Child, Node->getIDom());
      }
      DT->eraseNode(ExitingBlock);
      if (MSSAU) {
        SmallSetVector<BasicBlock *, 8> ExitBlockSet;
        ExitBlockSet.insert(ExitingBlock);
        MSSAU->removeBlocks(ExitBlockSet);
      }

      BI->getSuccessor(0)->removePredecessor(
          ExitingBlock, /* KeepOneInputPHIs */ PreserveLCSSA);
      BI->getSuccessor(1)->removePredecessor(
          ExitingBlock, /* KeepOneInputPHIs */ PreserveLCSSA);
      ExitingBlock->eraseFromParent();
    }
  }

  // Changing exit conditions for blocks may affect exit counts of this loop and
  // any of its paretns, so we must invalidate the entire subtree if we've made
  // any changes.
  if (Changed && SE)
    SE->forgetTopmostLoop(L);

  if (MSSAU && VerifyMemorySSA)
    MSSAU->getMemorySSA()->verifyMemorySSA();

  return Changed;
}

bool llvm::simplifyLoop(Loop *L, DominatorTree *DT, LoopInfo *LI,
                        ScalarEvolution *SE, AssumptionCache *AC,
                        MemorySSAUpdater *MSSAU, bool PreserveLCSSA) {
  bool Changed = false;

#ifndef NDEBUG
  // If we're asked to preserve LCSSA, the loop nest needs to start in LCSSA
  // form.
  if (PreserveLCSSA) {
    assert(DT && "DT not available.");
    assert(LI && "LI not available.");
    assert(L->isRecursivelyLCSSAForm(*DT, *LI) &&
           "Requested to preserve LCSSA, but it's already broken.");
  }
#endif

  // Worklist maintains our depth-first queue of loops in this nest to process.
  SmallVector<Loop *, 4> Worklist;
  Worklist.push_back(L);

  // Walk the worklist from front to back, pushing newly found sub loops onto
  // the back. This will let us process loops from back to front in depth-first
  // order. We can use this simple process because loops form a tree.
  for (unsigned Idx = 0; Idx != Worklist.size(); ++Idx) {
    Loop *L2 = Worklist[Idx];
    Worklist.append(L2->begin(), L2->end());
  }

  while (!Worklist.empty())
    Changed |= simplifyOneLoop(Worklist.pop_back_val(), Worklist, DT, LI, SE,
                               AC, MSSAU, PreserveLCSSA);

  return Changed;
}

namespace {
  struct LoopSimplify : public FunctionPass {
    static char ID; // Pass identification, replacement for typeid
    LoopSimplify() : FunctionPass(ID) {
      initializeLoopSimplifyPass(*PassRegistry::getPassRegistry());
    }

    bool runOnFunction(Function &F) override;

    void getAnalysisUsage(AnalysisUsage &AU) const override {
      AU.addRequired<AssumptionCacheTracker>();

      // We need loop information to identify the loops...
      AU.addRequired<DominatorTreeWrapperPass>();
      AU.addPreserved<DominatorTreeWrapperPass>();

      AU.addRequired<LoopInfoWrapperPass>();
      AU.addPreserved<LoopInfoWrapperPass>();

      AU.addPreserved<BasicAAWrapperPass>();
      AU.addPreserved<AAResultsWrapperPass>();
      AU.addPreserved<GlobalsAAWrapperPass>();
      AU.addPreserved<ScalarEvolutionWrapperPass>();
      AU.addPreserved<SCEVAAWrapperPass>();
      AU.addPreservedID(LCSSAID);
      AU.addPreserved<DependenceAnalysisWrapperPass>();
      AU.addPreservedID(BreakCriticalEdgesID);  // No critical edges added.
      AU.addPreserved<BranchProbabilityInfoWrapperPass>();
      AU.addPreserved<MemorySSAWrapperPass>();
    }

    /// verifyAnalysis() - Verify LoopSimplifyForm's guarantees.
    void verifyAnalysis() const override;
  };
}

char LoopSimplify::ID = 0;
INITIALIZE_PASS_BEGIN(LoopSimplify, "loop-simplify",
                "Canonicalize natural loops", false, false)
INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass)
INITIALIZE_PASS_END(LoopSimplify, "loop-simplify",
                "Canonicalize natural loops", false, false)

// Publicly exposed interface to pass...
char &llvm::LoopSimplifyID = LoopSimplify::ID;
Pass *llvm::createLoopSimplifyPass() { return new LoopSimplify(); }

/// runOnFunction - Run down all loops in the CFG (recursively, but we could do
/// it in any convenient order) inserting preheaders...
///
bool LoopSimplify::runOnFunction(Function &F) {
  bool Changed = false;
  LoopInfo *LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
  DominatorTree *DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
  auto *SEWP = getAnalysisIfAvailable<ScalarEvolutionWrapperPass>();
  ScalarEvolution *SE = SEWP ? &SEWP->getSE() : nullptr;
  AssumptionCache *AC =
      &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
  MemorySSA *MSSA = nullptr;
  std::unique_ptr<MemorySSAUpdater> MSSAU;
  auto *MSSAAnalysis = getAnalysisIfAvailable<MemorySSAWrapperPass>();
  if (MSSAAnalysis) {
    MSSA = &MSSAAnalysis->getMSSA();
    MSSAU = std::make_unique<MemorySSAUpdater>(MSSA);
  }

  bool PreserveLCSSA = mustPreserveAnalysisID(LCSSAID);

  // Simplify each loop nest in the function.
  for (auto *L : *LI)
    Changed |= simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), PreserveLCSSA);

#ifndef NDEBUG
  if (PreserveLCSSA) {
    bool InLCSSA = all_of(
        *LI, [&](Loop *L) { return L->isRecursivelyLCSSAForm(*DT, *LI); });
    assert(InLCSSA && "LCSSA is broken after loop-simplify.");
  }
#endif
  return Changed;
}

PreservedAnalyses LoopSimplifyPass::run(Function &F,
                                        FunctionAnalysisManager &AM) {
  bool Changed = false;
  LoopInfo *LI = &AM.getResult<LoopAnalysis>(F);
  DominatorTree *DT = &AM.getResult<DominatorTreeAnalysis>(F);
  ScalarEvolution *SE = AM.getCachedResult<ScalarEvolutionAnalysis>(F);
  AssumptionCache *AC = &AM.getResult<AssumptionAnalysis>(F);
  auto *MSSAAnalysis = AM.getCachedResult<MemorySSAAnalysis>(F);
  std::unique_ptr<MemorySSAUpdater> MSSAU;
  if (MSSAAnalysis) {
    auto *MSSA = &MSSAAnalysis->getMSSA();
    MSSAU = std::make_unique<MemorySSAUpdater>(MSSA);
  }


  // Note that we don't preserve LCSSA in the new PM, if you need it run LCSSA
  // after simplifying the loops. MemorySSA is preserved if it exists.
  for (auto *L : *LI)
    Changed |=
        simplifyLoop(L, DT, LI, SE, AC, MSSAU.get(), /*PreserveLCSSA*/ false);

  if (!Changed)
    return PreservedAnalyses::all();

  PreservedAnalyses PA;
  PA.preserve<DominatorTreeAnalysis>();
  PA.preserve<LoopAnalysis>();
  PA.preserve<ScalarEvolutionAnalysis>();
  PA.preserve<DependenceAnalysis>();
  if (MSSAAnalysis)
    PA.preserve<MemorySSAAnalysis>();
  // BPI maps conditional terminators to probabilities, LoopSimplify can insert
  // blocks, but it does so only by splitting existing blocks and edges. This
  // results in the interesting property that all new terminators inserted are
  // unconditional branches which do not appear in BPI. All deletions are
  // handled via ValueHandle callbacks w/in BPI.
  PA.preserve<BranchProbabilityAnalysis>();
  return PA;
}

// FIXME: Restore this code when we re-enable verification in verifyAnalysis
// below.
#if 0
static void verifyLoop(Loop *L) {
  // Verify subloops.
  for (Loop::iterator I = L->begin(), E = L->end(); I != E; ++I)
    verifyLoop(*I);

  // It used to be possible to just assert L->isLoopSimplifyForm(), however
  // with the introduction of indirectbr, there are now cases where it's
  // not possible to transform a loop as necessary. We can at least check
  // that there is an indirectbr near any time there's trouble.

  // Indirectbr can interfere with preheader and unique backedge insertion.
  if (!L->getLoopPreheader() || !L->getLoopLatch()) {
    bool HasIndBrPred = false;
    for (BasicBlock *Pred : predecessors(L->getHeader()))
      if (isa<IndirectBrInst>(Pred->getTerminator())) {
        HasIndBrPred = true;
        break;
      }
    assert(HasIndBrPred &&
           "LoopSimplify has no excuse for missing loop header info!");
    (void)HasIndBrPred;
  }

  // Indirectbr can interfere with exit block canonicalization.
  if (!L->hasDedicatedExits()) {
    bool HasIndBrExiting = false;
    SmallVector<BasicBlock*, 8> ExitingBlocks;
    L->getExitingBlocks(ExitingBlocks);
    for (unsigned i = 0, e = ExitingBlocks.size(); i != e; ++i) {
      if (isa<IndirectBrInst>((ExitingBlocks[i])->getTerminator())) {
        HasIndBrExiting = true;
        break;
      }
    }

    assert(HasIndBrExiting &&
           "LoopSimplify has no excuse for missing exit block info!");
    (void)HasIndBrExiting;
  }
}
#endif

void LoopSimplify::verifyAnalysis() const {
  // FIXME: This routine is being called mid-way through the loop pass manager
  // as loop passes destroy this analysis. That's actually fine, but we have no
  // way of expressing that here. Once all of the passes that destroy this are
  // hoisted out of the loop pass manager we can add back verification here.
#if 0
  for (LoopInfo::iterator I = LI->begin(), E = LI->end(); I != E; ++I)
    verifyLoop(*I);
#endif
}