C++ BasicBlock::begin方法代码示例

/// ProcessLoop - Walk the loop structure in depth first order, ensuring that
/// all loops have preheaders.
///
bool LoopSimplify::ProcessLoop(Loop *L, LPPassManager &LPM) {
  bool Changed = false;
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 (Loop::block_iterator BB = L->block_begin(), E = L->block_end();
       BB != E; ++BB) {
    if (*BB == L->getHeader()) continue;

    SmallPtrSet<BasicBlock*, 4> BadPreds;
    for (pred_iterator PI = pred_begin(*BB),
         PE = pred_end(*BB); PI != PE; ++PI) {
      BasicBlock *P = *PI;
      if (!L->contains(P))
        BadPreds.insert(P);
    }

    // Delete each unique out-of-loop (and thus dead) predecessor.
    for (SmallPtrSet<BasicBlock*, 4>::iterator I = BadPreds.begin(),
         E = BadPreds.end(); I != E; ++I) {

      DEBUG(dbgs() << "LoopSimplify: Deleting edge from dead predecessor "
                   << (*I)->getName() << "\n");

      // Inform each successor of each dead pred.
      for (succ_iterator SI = succ_begin(*I), SE = succ_end(*I); SI != SE; ++SI)
        (*SI)->removePredecessor(*I);
      // Zap the dead pred's terminator and replace it with unreachable.
      TerminatorInst *TI = (*I)->getTerminator();
       TI->replaceAllUsesWith(UndefValue::get(TI->getType()));
      (*I)->getTerminator()->eraseFromParent();
      new UnreachableInst((*I)->getContext(), *I);
      Changed = true;
    }
  }

  // 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 (SmallVectorImpl<BasicBlock *>::iterator I = ExitingBlocks.begin(),
       E = ExitingBlocks.end(); I != E; ++I)
    if (BranchInst *BI = dyn_cast<BranchInst>((*I)->getTerminator()))
      if (BI->isConditional()) {
        if (UndefValue *Cond = dyn_cast<UndefValue>(BI->getCondition())) {

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

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

          // This may make the loop analyzable, force SCEV recomputation.
          if (SE)
            SE->forgetLoop(L);

          Changed = true;
        }
      }

  // Does the loop already have a preheader?  If so, don't insert one.
  BasicBlock *Preheader = L->getLoopPreheader();
  if (!Preheader) {
    Preheader = InsertPreheaderForLoop(L);
    if (Preheader) {
      ++NumInserted;
      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.
  SmallVector<BasicBlock*, 8> ExitBlocks;
  L->getExitBlocks(ExitBlocks);

  SmallSetVector<BasicBlock *, 8> ExitBlockSet(ExitBlocks.begin(),
                                               ExitBlocks.end());
  for (SmallSetVector<BasicBlock *, 8>::iterator I = ExitBlockSet.begin(),
         E = ExitBlockSet.end(); I != E; ++I) {
    BasicBlock *ExitBlock = *I;
    for (pred_iterator PI = pred_begin(ExitBlock), PE = pred_end(ExitBlock);
         PI != PE; ++PI)
      // Must be exactly this loop: no subloops, parent loops, or non-loop preds
      // allowed.
      if (!L->contains(*PI)) {
        if (RewriteLoopExitBlock(L, ExitBlock)) {
          ++NumInserted;
          Changed = true;
        }
        break;
      }
//.........这里部分代码省略.........

/// handleEndBlock - Remove dead stores to stack-allocated locations in the
/// function end block.  Ex:
/// %A = alloca i32
/// ...
/// store i32 1, i32* %A
/// ret void
bool DSE::handleEndBlock(BasicBlock &BB) {
  bool MadeChange = false;

  // Keep track of all of the stack objects that are dead at the end of the
  // function.
  SmallSetVector<Value*, 16> DeadStackObjects;

  // Find all of the alloca'd pointers in the entry block.
  BasicBlock *Entry = BB.getParent()->begin();
  for (BasicBlock::iterator I = Entry->begin(), E = Entry->end(); I != E; ++I) {
    if (isa<AllocaInst>(I))
      DeadStackObjects.insert(I);

    // Okay, so these are dead heap objects, but if the pointer never escapes
    // then it's leaked by this function anyways.
    else if (isAllocLikeFn(I, TLI) && !PointerMayBeCaptured(I, true, true))
      DeadStackObjects.insert(I);
  }

  // Treat byval or inalloca arguments the same, stores to them are dead at the
  // end of the function.
  for (Function::arg_iterator AI = BB.getParent()->arg_begin(),
       AE = BB.getParent()->arg_end(); AI != AE; ++AI)
    if (AI->hasByValOrInAllocaAttr())
      DeadStackObjects.insert(AI);

  const DataLayout &DL = BB.getModule()->getDataLayout();

  // Scan the basic block backwards
  for (BasicBlock::iterator BBI = BB.end(); BBI != BB.begin(); ){
    --BBI;

    // If we find a store, check to see if it points into a dead stack value.
    if (hasMemoryWrite(BBI, TLI) && isRemovable(BBI)) {
      // See through pointer-to-pointer bitcasts
      SmallVector<Value *, 4> Pointers;
      GetUnderlyingObjects(getStoredPointerOperand(BBI), Pointers, DL);

      // Stores to stack values are valid candidates for removal.
      bool AllDead = true;
      for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(),
           E = Pointers.end(); I != E; ++I)
        if (!DeadStackObjects.count(*I)) {
          AllDead = false;
          break;
        }

      if (AllDead) {
        Instruction *Dead = BBI++;

        DEBUG(dbgs() << "DSE: Dead Store at End of Block:\n  DEAD: "
                     << *Dead << "\n  Objects: ";
              for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(),
                   E = Pointers.end(); I != E; ++I) {
                dbgs() << **I;
                if (std::next(I) != E)
                  dbgs() << ", ";
              }
              dbgs() << '\n');

        // DCE instructions only used to calculate that store.
        DeleteDeadInstruction(Dead, *MD, TLI, &DeadStackObjects);
        ++NumFastStores;
        MadeChange = true;
        continue;
      }
    }

    // Remove any dead non-memory-mutating instructions.
    if (isInstructionTriviallyDead(BBI, TLI)) {
      Instruction *Inst = BBI++;
      DeleteDeadInstruction(Inst, *MD, TLI, &DeadStackObjects);
      ++NumFastOther;
      MadeChange = true;
      continue;
    }

    if (isa<AllocaInst>(BBI)) {
      // Remove allocas from the list of dead stack objects; there can't be
      // any references before the definition.
      DeadStackObjects.remove(BBI);
      continue;
    }

    if (auto CS = CallSite(BBI)) {
      // Remove allocation function calls from the list of dead stack objects; 
      // there can't be any references before the definition.
      if (isAllocLikeFn(BBI, TLI))
        DeadStackObjects.remove(BBI);

      // If this call does not access memory, it can't be loading any of our
      // pointers.
      if (AA->doesNotAccessMemory(CS))
        continue;
//.........这里部分代码省略.........

/// RewriteLoopExitValues - Check to see if this loop has a computable
/// loop-invariant execution count.  If so, this means that we can compute the
/// final value of any expressions that are recurrent in the loop, and
/// substitute the exit values from the loop into any instructions outside of
/// the loop that use the final values of the current expressions.
///
/// This is mostly redundant with the regular IndVarSimplify activities that
/// happen later, except that it's more powerful in some cases, because it's
/// able to brute-force evaluate arbitrary instructions as long as they have
/// constant operands at the beginning of the loop.
void IndVarSimplify::RewriteLoopExitValues(Loop *L,
                                           SCEVExpander &Rewriter) {
  // Verify the input to the pass in already in LCSSA form.
  assert(L->isLCSSAForm());

  SmallVector<BasicBlock*, 8> ExitBlocks;
  L->getUniqueExitBlocks(ExitBlocks);

  // Find all values that are computed inside the loop, but used outside of it.
  // Because of LCSSA, these values will only occur in LCSSA PHI Nodes.  Scan
  // the exit blocks of the loop to find them.
  for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i) {
    BasicBlock *ExitBB = ExitBlocks[i];

    // If there are no PHI nodes in this exit block, then no values defined
    // inside the loop are used on this path, skip it.
    PHINode *PN = dyn_cast<PHINode>(ExitBB->begin());
    if (!PN) continue;

    unsigned NumPreds = PN->getNumIncomingValues();

    // Iterate over all of the PHI nodes.
    BasicBlock::iterator BBI = ExitBB->begin();
    while ((PN = dyn_cast<PHINode>(BBI++))) {
      if (PN->use_empty())
        continue; // dead use, don't replace it

      // SCEV only supports integer expressions for now.
      if (!PN->getType()->isIntegerTy() && !PN->getType()->isPointerTy())
        continue;

      // It's necessary to tell ScalarEvolution about this explicitly so that
      // it can walk the def-use list and forget all SCEVs, as it may not be
      // watching the PHI itself. Once the new exit value is in place, there
      // may not be a def-use connection between the loop and every instruction
      // which got a SCEVAddRecExpr for that loop.
      SE->forgetValue(PN);

      // Iterate over all of the values in all the PHI nodes.
      for (unsigned i = 0; i != NumPreds; ++i) {
        // If the value being merged in is not integer or is not defined
        // in the loop, skip it.
        Value *InVal = PN->getIncomingValue(i);
        if (!isa<Instruction>(InVal))
          continue;

        // If this pred is for a subloop, not L itself, skip it.
        if (LI->getLoopFor(PN->getIncomingBlock(i)) != L)
          continue; // The Block is in a subloop, skip it.

        // Check that InVal is defined in the loop.
        Instruction *Inst = cast<Instruction>(InVal);
        if (!L->contains(Inst))
          continue;

        // Okay, this instruction has a user outside of the current loop
        // and varies predictably *inside* the loop.  Evaluate the value it
        // contains when the loop exits, if possible.
        const SCEV *ExitValue = SE->getSCEVAtScope(Inst, L->getParentLoop());
        if (!ExitValue->isLoopInvariant(L))
          continue;

        Changed = true;
        ++NumReplaced;

        Value *ExitVal = Rewriter.expandCodeFor(ExitValue, PN->getType(), Inst);

        DEBUG(dbgs() << "INDVARS: RLEV: AfterLoopVal = " << *ExitVal << '\n'
                     << "  LoopVal = " << *Inst << "\n");

        PN->setIncomingValue(i, ExitVal);

        // If this instruction is dead now, delete it.
        RecursivelyDeleteTriviallyDeadInstructions(Inst);

        if (NumPreds == 1) {
          // Completely replace a single-pred PHI. This is safe, because the
          // NewVal won't be variant in the loop, so we don't need an LCSSA phi
          // node anymore.
          PN->replaceAllUsesWith(ExitVal);
          RecursivelyDeleteTriviallyDeadInstructions(PN);
        }
      }
      if (NumPreds != 1) {
        // Clone the PHI and delete the original one. This lets IVUsers and
        // any other maps purge the original user from their records.
        PHINode *NewPN = cast<PHINode>(PN->clone());
        NewPN->takeName(PN);
        NewPN->insertBefore(PN);
        PN->replaceAllUsesWith(NewPN);
//.........这里部分代码省略.........

InstructionInfoTable::InstructionInfoTable(Module *m) 
  : dummyString(""), dummyInfo(0, dummyString, -1, 0, 0), idCounter(0) {
  unsigned id = 0;
  std::map<const Instruction*, unsigned> lineTable;
  buildInstructionToLineMap(m, lineTable);

  // Sort the list of functions, in order to get consistent IDs
  std::vector<Function*> functions;

  for (Module::iterator fnIt = m->begin(), fn_ie = m->end();
         fnIt != fn_ie; ++fnIt) {
    functions.push_back(&(*fnIt));
  }

  std::sort(functions.begin(), functions.end(), ltfunc());

  for (std::vector<Function*>::iterator fnIt = functions.begin();
      fnIt != functions.end(); fnIt++) {
    Function *f = *fnIt;
    const std::string *initialFile = &dummyString;
    unsigned initialLine = 0;
    int initialID = -1;

    // It may be better to look for the closest stoppoint to the entry
    // following the CFG, but it is not clear that it ever matters in
    // practice.
    for (inst_iterator it = inst_begin(f), ie = inst_end(f);
         it != ie; ++it)
      if (getInstructionDebugInfo(&*it, initialFile, initialID, initialLine))
        break;
    
    typedef std::map<BasicBlock*, std::pair<std::pair<const std::string*, int>, unsigned> >
      sourceinfo_ty;
    sourceinfo_ty sourceInfo;
    for (llvm::Function::iterator bbIt = f->begin(), bbie = f->end();
         bbIt != bbie; ++bbIt) {
      std::pair<sourceinfo_ty::iterator, bool>
        res = sourceInfo.insert(std::make_pair(bbIt,
                                               std::make_pair(
                                                   std::make_pair(initialFile, initialID),
                                                              initialLine)));
      if (!res.second)
        continue;

      std::vector<BasicBlock*> worklist;
      worklist.push_back(bbIt);

      do {
        BasicBlock *bb = worklist.back();
        worklist.pop_back();

        sourceinfo_ty::iterator si = sourceInfo.find(bb);
        assert(si != sourceInfo.end());
        const std::string *file = si->second.first.first;
        unsigned line = si->second.second;
        int fileID = si->second.first.second;
        
        for (BasicBlock::iterator it = bb->begin(), ie = bb->end();
             it != ie; ++it) {
          Instruction *instr = it;
          unsigned assemblyLine = 0;
          std::map<const Instruction*, unsigned>::const_iterator ltit = 
            lineTable.find(instr);
          if (ltit!=lineTable.end())
            assemblyLine = ltit->second;
          getInstructionDebugInfo(instr, file, fileID, line);
          infos.insert(std::make_pair(instr,
                                      InstructionInfo(id++,
                                                      *file,
                                                      fileID,
                                                      line,
                                                      assemblyLine)));        
        }
        
        for (succ_iterator it = succ_begin(bb), ie = succ_end(bb); 
             it != ie; ++it) {
          if (sourceInfo.insert(std::make_pair(*it,
                                               std::make_pair(std::make_pair(file, fileID), line))).second)
            worklist.push_back(*it);
        }
      } while (!worklist.empty());
    }
  }
}

void RegionExtractor::findInputsOutputs(ValueSet &Inputs,
                                      ValueSet &Outputs) const {
  for (SetVector<BasicBlock *>::const_iterator I = Blocks.begin(),
                                               E = Blocks.end();
       I != E; ++I) {
    BasicBlock *BB = *I;

    // If a used value is defined outside the region, it's an input.  If an
    // instruction is used outside the region, it's an output.
    for (BasicBlock::iterator II = BB->begin(), IE = BB->end();
         II != IE; ++II) {
      for (User::op_iterator OI = II->op_begin(), OE = II->op_end();
           OI != OE; ++OI)
        if (definedInCaller(Blocks, *OI))
          Inputs.insert(*OI);
#if LLVM_VERSION_MINOR == 5
      for (User *U : II->users())
        if (!definedInRegion(Blocks, U)) {
#else
      for (Value::use_iterator UI = II->use_begin(), UE = II->use_end();
           UI != UE; ++UI)
        if (!definedInRegion(Blocks, *UI)) {
#endif
          Outputs.insert(II);
          break;
        }
    }
  }
}

/// severSplitPHINodes - If a PHI node has multiple inputs from outside of the
/// region, we need to split the entry block of the region so that the PHI node
/// is easier to deal with.
void RegionExtractor::severSplitPHINodes(BasicBlock *&Header) {
  unsigned NumPredsFromRegion = 0;
  unsigned NumPredsOutsideRegion = 0;

  if (Header != &Header->getParent()->getEntryBlock()) {
    PHINode *PN = dyn_cast<PHINode>(Header->begin());
    if (!PN) return; // No PHI nodes.

    // If the header node contains any PHI nodes, check to see if there is more
    // than one entry from outside the region.  If so, we need to sever the
    // header block into two.
    for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
      if (Blocks.count(PN->getIncomingBlock(i)))
        ++NumPredsFromRegion;
      else
        ++NumPredsOutsideRegion;

    // If there is one (or fewer) predecessor from outside the region, we don't
    // need to do anything special.
    if (NumPredsOutsideRegion <= 1) return;
  }

  // Otherwise, we need to split the header block into two pieces: one
  // containing PHI nodes merging values from outside of the region, and a
  // second that contains all of the code for the block and merges back any
  // incoming values from inside of the region.
  BasicBlock::iterator AfterPHIs = Header->getFirstNonPHI();
  BasicBlock *NewBB = Header->splitBasicBlock(AfterPHIs,
                                              Header->getName()+".ce");

  // We only want to code extract the second block now, and it becomes the new
  // header of the region.
  BasicBlock *OldPred = Header;
  Blocks.remove(OldPred);
  Blocks.insert(NewBB);
  Header = NewBB;

  // Okay, update dominator sets. The blocks that dominate the new one are the
  // blocks that dominate TIBB plus the new block itself.
  if (DT)
    DT->splitBlock(NewBB);

  // Okay, now we need to adjust the PHI nodes and any branches from within the
  // region to go to the new header block instead of the old header block.
  if (NumPredsFromRegion) {
    PHINode *PN = cast<PHINode>(OldPred->begin());
    // Loop over all of the predecessors of OldPred that are in the region,
    // changing them to branch to NewBB instead.
    for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
      if (Blocks.count(PN->getIncomingBlock(i))) {
        TerminatorInst *TI = PN->getIncomingBlock(i)->getTerminator();
        TI->replaceUsesOfWith(OldPred, NewBB);
      }

    // Okay, everything within the region is now branching to the right block, we
    // just have to update the PHI nodes now, inserting PHI nodes into NewBB.
    for (AfterPHIs = OldPred->begin(); isa<PHINode>(AfterPHIs); ++AfterPHIs) {
      PHINode *PN = cast<PHINode>(AfterPHIs);
      // Create a new PHI node in the new region, which has an incoming value
      // from OldPred of PN.
      PHINode *NewPN = PHINode::Create(PN->getType(), 1 + NumPredsFromRegion,
                                       PN->getName()+".ce", NewBB->begin());
      NewPN->addIncoming(PN, OldPred);

      // Loop over all of the incoming value in PN, moving them to NewPN if they
      // are from the extracted region.
      for (unsigned i = 0; i != PN->getNumIncomingValues(); ++i) {
//.........这里部分代码省略.........

/// IsTrivialUnswitchCondition - Check to see if this unswitch condition is
/// trivial: that is, that the condition controls whether or not the loop does
/// anything at all.  If this is a trivial condition, unswitching produces no
/// code duplications (equivalently, it produces a simpler loop and a new empty
/// loop, which gets deleted).
///
/// If this is a trivial condition, return true, otherwise return false.  When
/// returning true, this sets Cond and Val to the condition that controls the
/// trivial condition: when Cond dynamically equals Val, the loop is known to
/// exit.  Finally, this sets LoopExit to the BB that the loop exits to when
/// Cond == Val.
///
bool LoopUnswitch::IsTrivialUnswitchCondition(Value *Cond, Constant **Val,
                                       BasicBlock **LoopExit) {
  BasicBlock *Header = currentLoop->getHeader();
  TerminatorInst *HeaderTerm = Header->getTerminator();
  LLVMContext &Context = Header->getContext();

  BasicBlock *LoopExitBB = 0;
  if (BranchInst *BI = dyn_cast<BranchInst>(HeaderTerm)) {
    // If the header block doesn't end with a conditional branch on Cond, we
    // can't handle it.
    if (!BI->isConditional() || BI->getCondition() != Cond)
      return false;

    // Check to see if a successor of the branch is guaranteed to
    // exit through a unique exit block without having any
    // side-effects.  If so, determine the value of Cond that causes it to do
    // this.
    if ((LoopExitBB = isTrivialLoopExitBlock(currentLoop,
                                             BI->getSuccessor(0)))) {
      if (Val) *Val = ConstantInt::getTrue(Context);
    } else if ((LoopExitBB = isTrivialLoopExitBlock(currentLoop,
                                                    BI->getSuccessor(1)))) {
      if (Val) *Val = ConstantInt::getFalse(Context);
    }
  } else if (SwitchInst *SI = dyn_cast<SwitchInst>(HeaderTerm)) {
    // If this isn't a switch on Cond, we can't handle it.
    if (SI->getCondition() != Cond) return false;

    // Check to see if a successor of the switch is guaranteed to go to the
    // latch block or exit through a one exit block without having any
    // side-effects.  If so, determine the value of Cond that causes it to do
    // this.
    // Note that we can't trivially unswitch on the default case or
    // on already unswitched cases.
    for (SwitchInst::CaseIt i = SI->case_begin(), e = SI->case_end();
         i != e; ++i) {
      BasicBlock *LoopExitCandidate;
      if ((LoopExitCandidate = isTrivialLoopExitBlock(currentLoop,
                                               i.getCaseSuccessor()))) {
        // Okay, we found a trivial case, remember the value that is trivial.
        ConstantInt *CaseVal = i.getCaseValue();

        // Check that it was not unswitched before, since already unswitched
        // trivial vals are looks trivial too.
        if (BranchesInfo.isUnswitched(SI, CaseVal))
          continue;
        LoopExitBB = LoopExitCandidate;
        if (Val) *Val = CaseVal;
        break;
      }
    }
  }

  // If we didn't find a single unique LoopExit block, or if the loop exit block
  // contains phi nodes, this isn't trivial.
  if (!LoopExitBB || isa<PHINode>(LoopExitBB->begin()))
    return false;   // Can't handle this.

  if (LoopExit) *LoopExit = LoopExitBB;

  // We already know that nothing uses any scalar values defined inside of this
  // loop.  As such, we just have to check to see if this loop will execute any
  // side-effecting instructions (e.g. stores, calls, volatile loads) in the
  // part of the loop that the code *would* execute.  We already checked the
  // tail, check the header now.
  for (BasicBlock::iterator I = Header->begin(), E = Header->end(); I != E; ++I)
    if (I->mayHaveSideEffects())
      return false;
  return true;
}

// RewriteLoopBodyWithConditionConstant - We know either that the value LIC has
// the value specified by Val in the specified loop, or we know it does NOT have
// that value.  Rewrite any uses of LIC or of properties correlated to it.
void LoopUnswitch::RewriteLoopBodyWithConditionConstant(Loop *L, Value *LIC,
                                                        Constant *Val,
                                                        bool IsEqual) {
  assert(!isa<Constant>(LIC) && "Why are we unswitching on a constant?");

  // FIXME: Support correlated properties, like:
  //  for (...)
  //    if (li1 < li2)
  //      ...
  //    if (li1 > li2)
  //      ...

  // FOLD boolean conditions (X|LIC), (X&LIC).  Fold conditional branches,
  // selects, switches.
  std::vector<Instruction*> Worklist;
  LLVMContext &Context = Val->getContext();

  // If we know that LIC == Val, or that LIC == NotVal, just replace uses of LIC
  // in the loop with the appropriate one directly.
  if (IsEqual || (isa<ConstantInt>(Val) &&
      Val->getType()->isIntegerTy(1))) {
    Value *Replacement;
    if (IsEqual)
      Replacement = Val;
    else
      Replacement = ConstantInt::get(Type::getInt1Ty(Val->getContext()),
                                     !cast<ConstantInt>(Val)->getZExtValue());

    for (User *U : LIC->users()) {
      Instruction *UI = dyn_cast<Instruction>(U);
      if (!UI || !L->contains(UI))
        continue;
      Worklist.push_back(UI);
    }

    for (std::vector<Instruction*>::iterator UI = Worklist.begin(),
         UE = Worklist.end(); UI != UE; ++UI)
      (*UI)->replaceUsesOfWith(LIC, Replacement);

    SimplifyCode(Worklist, L);
    return;
  }

  // Otherwise, we don't know the precise value of LIC, but we do know that it
  // is certainly NOT "Val".  As such, simplify any uses in the loop that we
  // can.  This case occurs when we unswitch switch statements.
  for (User *U : LIC->users()) {
    Instruction *UI = dyn_cast<Instruction>(U);
    if (!UI || !L->contains(UI))
      continue;

    Worklist.push_back(UI);

    // TODO: We could do other simplifications, for example, turning
    // 'icmp eq LIC, Val' -> false.

    // If we know that LIC is not Val, use this info to simplify code.
    SwitchInst *SI = dyn_cast<SwitchInst>(UI);
    if (SI == 0 || !isa<ConstantInt>(Val)) continue;

    SwitchInst::CaseIt DeadCase = SI->findCaseValue(cast<ConstantInt>(Val));
    // Default case is live for multiple values.
    if (DeadCase == SI->case_default()) continue;

    // Found a dead case value.  Don't remove PHI nodes in the
    // successor if they become single-entry, those PHI nodes may
    // be in the Users list.

    BasicBlock *Switch = SI->getParent();
    BasicBlock *SISucc = DeadCase.getCaseSuccessor();
    BasicBlock *Latch = L->getLoopLatch();

    BranchesInfo.setUnswitched(SI, Val);

    if (!SI->findCaseDest(SISucc)) continue;  // Edge is critical.
    // If the DeadCase successor dominates the loop latch, then the
    // transformation isn't safe since it will delete the sole predecessor edge
    // to the latch.
    if (Latch && DT->dominates(SISucc, Latch))
      continue;

    // FIXME: This is a hack.  We need to keep the successor around
    // and hooked up so as to preserve the loop structure, because
    // trying to update it is complicated.  So instead we preserve the
    // loop structure and put the block on a dead code path.
    SplitEdge(Switch, SISucc, this);
    // Compute the successors instead of relying on the return value
    // of SplitEdge, since it may have split the switch successor
    // after PHI nodes.
    BasicBlock *NewSISucc = DeadCase.getCaseSuccessor();
    BasicBlock *OldSISucc = *succ_begin(NewSISucc);
    // Create an "unreachable" destination.
    BasicBlock *Abort = BasicBlock::Create(Context, "us-unreachable",
                                           Switch->getParent(),
                                           OldSISucc);
    new UnreachableInst(Context, Abort);
    // Force the new case destination to branch to the "unreachable"
//.........这里部分代码省略.........

/// Attempt to merge an objc_release with a store, load, and objc_retain to form
/// an objc_storeStrong. This can be a little tricky because the instructions
/// don't always appear in order, and there may be unrelated intervening
/// instructions.
void ObjCARCContract::ContractRelease(Instruction *Release,
                                      inst_iterator &Iter) {
  LoadInst *Load = dyn_cast<LoadInst>(GetObjCArg(Release));
  if (!Load || !Load->isSimple()) return;

  // For now, require everything to be in one basic block.
  BasicBlock *BB = Release->getParent();
  if (Load->getParent() != BB) return;

  // Walk down to find the store and the release, which may be in either order.
  BasicBlock::iterator I = Load, End = BB->end();
  ++I;
  AliasAnalysis::Location Loc = AA->getLocation(Load);
  StoreInst *Store = 0;
  bool SawRelease = false;
  for (; !Store || !SawRelease; ++I) {
    if (I == End)
      return;

    Instruction *Inst = I;
    if (Inst == Release) {
      SawRelease = true;
      continue;
    }

    InstructionClass Class = GetBasicInstructionClass(Inst);

    // Unrelated retains are harmless.
    if (IsRetain(Class))
      continue;

    if (Store) {
      // The store is the point where we're going to put the objc_storeStrong,
      // so make sure there are no uses after it.
      if (CanUse(Inst, Load, PA, Class))
        return;
    } else if (AA->getModRefInfo(Inst, Loc) & AliasAnalysis::Mod) {
      // We are moving the load down to the store, so check for anything
      // else which writes to the memory between the load and the store.
      Store = dyn_cast<StoreInst>(Inst);
      if (!Store || !Store->isSimple()) return;
      if (Store->getPointerOperand() != Loc.Ptr) return;
    }
  }

  Value *New = StripPointerCastsAndObjCCalls(Store->getValueOperand());

  // Walk up to find the retain.
  I = Store;
  BasicBlock::iterator Begin = BB->begin();
  while (I != Begin && GetBasicInstructionClass(I) != IC_Retain)
    --I;
  Instruction *Retain = I;
  if (GetBasicInstructionClass(Retain) != IC_Retain) return;
  if (GetObjCArg(Retain) != New) return;

  Changed = true;
  ++NumStoreStrongs;

  LLVMContext &C = Release->getContext();
  Type *I8X = PointerType::getUnqual(Type::getInt8Ty(C));
  Type *I8XX = PointerType::getUnqual(I8X);

  Value *Args[] = { Load->getPointerOperand(), New };
  if (Args[0]->getType() != I8XX)
    Args[0] = new BitCastInst(Args[0], I8XX, "", Store);
  if (Args[1]->getType() != I8X)
    Args[1] = new BitCastInst(Args[1], I8X, "", Store);
  Constant *Decl = EP.get(ARCRuntimeEntryPoints::EPT_StoreStrong);
  CallInst *StoreStrong = CallInst::Create(Decl, Args, "", Store);
  StoreStrong->setDoesNotThrow();
  StoreStrong->setDebugLoc(Store->getDebugLoc());

  // We can't set the tail flag yet, because we haven't yet determined
  // whether there are any escaping allocas. Remember this call, so that
  // we can set the tail flag once we know it's safe.
  StoreStrongCalls.insert(StoreStrong);

  if (&*Iter == Store) ++Iter;
  Store->eraseFromParent();
  Release->eraseFromParent();
  EraseInstruction(Retain);
  if (Load->use_empty())
    Load->eraseFromParent();
}

bool ObjCARCContract::runOnFunction(Function &F) {
  if (!EnableARCOpts)
    return false;

  // If nothing in the Module uses ARC, don't do anything.
  if (!Run)
    return false;

  Changed = false;
  AA = &getAnalysis<AliasAnalysis>();
  DT = &getAnalysis<DominatorTree>();

  PA.setAA(&getAnalysis<AliasAnalysis>());

  // Track whether it's ok to mark objc_storeStrong calls with the "tail"
  // keyword. Be conservative if the function has variadic arguments.
  // It seems that functions which "return twice" are also unsafe for the
  // "tail" argument, because they are setjmp, which could need to
  // return to an earlier stack state.
  bool TailOkForStoreStrongs = !F.isVarArg() &&
                               !F.callsFunctionThatReturnsTwice();

  // For ObjC library calls which return their argument, replace uses of the
  // argument with uses of the call return value, if it dominates the use. This
  // reduces register pressure.
  SmallPtrSet<Instruction *, 4> DependingInstructions;
  SmallPtrSet<const BasicBlock *, 4> Visited;
  for (inst_iterator I = inst_begin(&F), E = inst_end(&F); I != E; ) {
    Instruction *Inst = &*I++;

    DEBUG(dbgs() << "ObjCARCContract: Visiting: " << *Inst << "\n");

    // Only these library routines return their argument. In particular,
    // objc_retainBlock does not necessarily return its argument.
    InstructionClass Class = GetBasicInstructionClass(Inst);
    switch (Class) {
    case IC_FusedRetainAutorelease:
    case IC_FusedRetainAutoreleaseRV:
      break;
    case IC_Autorelease:
    case IC_AutoreleaseRV:
      if (ContractAutorelease(F, Inst, Class, DependingInstructions, Visited))
        continue;
      break;
    case IC_Retain:
      // Attempt to convert retains to retainrvs if they are next to function
      // calls.
      if (!OptimizeRetainCall(F, Inst))
        break;
      // If we succeed in our optimization, fall through.
      // FALLTHROUGH
    case IC_RetainRV: {
      // If we're compiling for a target which needs a special inline-asm
      // marker to do the retainAutoreleasedReturnValue optimization,
      // insert it now.
      if (!RetainRVMarker)
        break;
      BasicBlock::iterator BBI = Inst;
      BasicBlock *InstParent = Inst->getParent();

      // Step up to see if the call immediately precedes the RetainRV call.
      // If it's an invoke, we have to cross a block boundary. And we have
      // to carefully dodge no-op instructions.
      do {
        if (&*BBI == InstParent->begin()) {
          BasicBlock *Pred = InstParent->getSinglePredecessor();
          if (!Pred)
            goto decline_rv_optimization;
          BBI = Pred->getTerminator();
          break;
        }
        --BBI;
      } while (IsNoopInstruction(BBI));

      if (&*BBI == GetObjCArg(Inst)) {
        DEBUG(dbgs() << "ObjCARCContract: Adding inline asm marker for "
                        "retainAutoreleasedReturnValue optimization.\n");
        Changed = true;
        InlineAsm *IA =
          InlineAsm::get(FunctionType::get(Type::getVoidTy(Inst->getContext()),
                                           /*isVarArg=*/false),
                         RetainRVMarker->getString(),
                         /*Constraints=*/"", /*hasSideEffects=*/true);
        CallInst::Create(IA, "", Inst);
      }
    decline_rv_optimization:
      break;
    }
    case IC_InitWeak: {
      // objc_initWeak(p, null) => *p = null
      CallInst *CI = cast<CallInst>(Inst);
      if (IsNullOrUndef(CI->getArgOperand(1))) {
        Value *Null =
          ConstantPointerNull::get(cast<PointerType>(CI->getType()));
        Changed = true;
        new StoreInst(Null, CI->getArgOperand(0), CI);

        DEBUG(dbgs() << "OBJCARCContract: Old = " << *CI << "\n"
                     << "                 New = " << *Null << "\n");

//.........这里部分代码省略.........

bool llvm::expandAtomicRMWToCmpXchg(AtomicRMWInst *AI,
                                    CreateCmpXchgInstFun CreateCmpXchg) {
  assert(AI);

  AtomicOrdering MemOpOrder =
      AI->getOrdering() == Unordered ? Monotonic : AI->getOrdering();
  Value *Addr = AI->getPointerOperand();
  BasicBlock *BB = AI->getParent();
  Function *F = BB->getParent();
  LLVMContext &Ctx = F->getContext();

  // Given: atomicrmw some_op iN* %addr, iN %incr ordering
  //
  // The standard expansion we produce is:
  //     [...]
  //     %init_loaded = load atomic iN* %addr
  //     br label %loop
  // loop:
  //     %loaded = phi iN [ %init_loaded, %entry ], [ %new_loaded, %loop ]
  //     %new = some_op iN %loaded, %incr
  //     %pair = cmpxchg iN* %addr, iN %loaded, iN %new
  //     %new_loaded = extractvalue { iN, i1 } %pair, 0
  //     %success = extractvalue { iN, i1 } %pair, 1
  //     br i1 %success, label %atomicrmw.end, label %loop
  // atomicrmw.end:
  //     [...]
  BasicBlock *ExitBB = BB->splitBasicBlock(AI, "atomicrmw.end");
  BasicBlock *LoopBB = BasicBlock::Create(Ctx, "atomicrmw.start", F, ExitBB);

  // This grabs the DebugLoc from AI.
  IRBuilder<> Builder(AI);

  // The split call above "helpfully" added a branch at the end of BB (to the
  // wrong place), but we want a load. It's easiest to just remove
  // the branch entirely.
  std::prev(BB->end())->eraseFromParent();
  Builder.SetInsertPoint(BB);
  LoadInst *InitLoaded = Builder.CreateLoad(Addr);
  // Atomics require at least natural alignment.
  InitLoaded->setAlignment(AI->getType()->getPrimitiveSizeInBits() / 8);
  Builder.CreateBr(LoopBB);

  // Start the main loop block now that we've taken care of the preliminaries.
  Builder.SetInsertPoint(LoopBB);
  PHINode *Loaded = Builder.CreatePHI(AI->getType(), 2, "loaded");
  Loaded->addIncoming(InitLoaded, BB);

  Value *NewVal =
      performAtomicOp(AI->getOperation(), Builder, Loaded, AI->getValOperand());

  Value *NewLoaded = nullptr;
  Value *Success = nullptr;

  CreateCmpXchg(Builder, Addr, Loaded, NewVal, MemOpOrder,
                Success, NewLoaded);
  assert(Success && NewLoaded);

  Loaded->addIncoming(NewLoaded, LoopBB);

  Builder.CreateCondBr(Success, ExitBB, LoopBB);

  Builder.SetInsertPoint(ExitBB, ExitBB->begin());

  AI->replaceAllUsesWith(NewLoaded);
  AI->eraseFromParent();

  return true;
}

    Value* ModuloSchedulerDriverPass::copyLoopBodyToHeader(Instruction* inst,
            Instruction* induction, BasicBlock* header, int offset){

        // Holds the body of the interesting loop
        BasicBlock *body = inst->getParent();

        assert(header && "Header is null");
        assert(header->getTerminator() && "Header has no terminator");

        // Maps the old instructions to the new Instructions
        DenseMap<const Value *, Value *>  ValueMap;
        // Do the actual clone
        stringstream iname;
        iname<<"___"<<offset<<"___";
        BasicBlock* newBB = CloneBasicBlock(body, ValueMap, iname.str().c_str());

        // Fixing the dependencies for each of the instructions in the cloned BB
        // They now depend on themselves rather on the old cloned BB.
        for (BasicBlock::iterator it = newBB->begin(); it != newBB->end(); ++it) {
            for (Instruction::op_iterator ops = (it)->op_begin(); ops != (it)->op_end(); ++ops) {
                if (ValueMap.end() != ValueMap.find(*ops)) {
                    //*ops = ValueMap[*ops];
                    it->replaceUsesOfWith(*ops, ValueMap[*ops]);
                }
            }
        }

        // Fixing the PHI nodes since they are no longer needed
        for (BasicBlock::iterator it = newBB->begin(); it != newBB->end(); ++it) {
            if (PHINode *phi = dyn_cast<PHINode>(it)) {
                // Taking the preheader entryfrom the PHI node

                Value* prevalue = phi->getIncomingValue(phi->getBasicBlockIndex(header));
                assert(prevalue && "no prevalue. Don't know what to do");

                // If we are handling a PHI node which is the induction index ? A[PHI(i,0)] ?
                // If so, turn it into A[i + offset]
                if (ValueMap[induction] == phi) {
                    Instruction *add = subscripts::incrementValue(prevalue, offset);
                    //add->insertBefore(phi); This is the same as next line (compiles on LLVM2.1)
                    phi->getParent()->getInstList().insert(phi, add);
                    phi->replaceAllUsesWith(add);
                }  else {
                    // eliminating the PHI node all together
                    // This is just a regular variable or constant. No need to increment
                    // the index.
                    phi->replaceAllUsesWith(prevalue);
                }
            } 
        }

        // Move all non PHI and non terminator instructions into the header.
        while (!newBB->getFirstNonPHI()->isTerminator()) {
            Instruction* inst = newBB->getFirstNonPHI();
            if (dyn_cast<StoreInst>(inst)) {
                inst->eraseFromParent();
            } else {
                inst->moveBefore(header->getTerminator());
            }
        }
        newBB->dropAllReferences();
        return ValueMap[inst];
    }

bool AtomicExpand::expandAtomicCmpXchg(AtomicCmpXchgInst *CI) {
  AtomicOrdering SuccessOrder = CI->getSuccessOrdering();
  AtomicOrdering FailureOrder = CI->getFailureOrdering();
  Value *Addr = CI->getPointerOperand();
  BasicBlock *BB = CI->getParent();
  Function *F = BB->getParent();
  LLVMContext &Ctx = F->getContext();
  // If getInsertFencesForAtomic() returns true, then the target does not want
  // to deal with memory orders, and emitLeading/TrailingFence should take care
  // of everything. Otherwise, emitLeading/TrailingFence are no-op and we
  // should preserve the ordering.
  AtomicOrdering MemOpOrder =
      TLI->getInsertFencesForAtomic() ? Monotonic : SuccessOrder;

  // Given: cmpxchg some_op iN* %addr, iN %desired, iN %new success_ord fail_ord
  //
  // The full expansion we produce is:
  //     [...]
  //     fence?
  // cmpxchg.start:
  //     %loaded = @load.linked(%addr)
  //     %should_store = icmp eq %loaded, %desired
  //     br i1 %should_store, label %cmpxchg.trystore,
  //                          label %cmpxchg.failure
  // cmpxchg.trystore:
  //     %stored = @store_conditional(%new, %addr)
  //     %success = icmp eq i32 %stored, 0
  //     br i1 %success, label %cmpxchg.success, label %loop/%cmpxchg.failure
  // cmpxchg.success:
  //     fence?
  //     br label %cmpxchg.end
  // cmpxchg.failure:
  //     fence?
  //     br label %cmpxchg.end
  // cmpxchg.end:
  //     %success = phi i1 [true, %cmpxchg.success], [false, %cmpxchg.failure]
  //     %restmp = insertvalue { iN, i1 } undef, iN %loaded, 0
  //     %res = insertvalue { iN, i1 } %restmp, i1 %success, 1
  //     [...]
  BasicBlock *ExitBB = BB->splitBasicBlock(CI, "cmpxchg.end");
  auto FailureBB = BasicBlock::Create(Ctx, "cmpxchg.failure", F, ExitBB);
  auto SuccessBB = BasicBlock::Create(Ctx, "cmpxchg.success", F, FailureBB);
  auto TryStoreBB = BasicBlock::Create(Ctx, "cmpxchg.trystore", F, SuccessBB);
  auto LoopBB = BasicBlock::Create(Ctx, "cmpxchg.start", F, TryStoreBB);

  // This grabs the DebugLoc from CI
  IRBuilder<> Builder(CI);

  // The split call above "helpfully" added a branch at the end of BB (to the
  // wrong place), but we might want a fence too. It's easiest to just remove
  // the branch entirely.
  std::prev(BB->end())->eraseFromParent();
  Builder.SetInsertPoint(BB);
  TLI->emitLeadingFence(Builder, SuccessOrder, /*IsStore=*/true,
                        /*IsLoad=*/true);
  Builder.CreateBr(LoopBB);

  // Start the main loop block now that we've taken care of the preliminaries.
  Builder.SetInsertPoint(LoopBB);
  Value *Loaded = TLI->emitLoadLinked(Builder, Addr, MemOpOrder);
  Value *ShouldStore =
      Builder.CreateICmpEQ(Loaded, CI->getCompareOperand(), "should_store");

  // If the cmpxchg doesn't actually need any ordering when it fails, we can
  // jump straight past that fence instruction (if it exists).
  Builder.CreateCondBr(ShouldStore, TryStoreBB, FailureBB);

  Builder.SetInsertPoint(TryStoreBB);
  Value *StoreSuccess = TLI->emitStoreConditional(
      Builder, CI->getNewValOperand(), Addr, MemOpOrder);
  StoreSuccess = Builder.CreateICmpEQ(
      StoreSuccess, ConstantInt::get(Type::getInt32Ty(Ctx), 0), "success");
  Builder.CreateCondBr(StoreSuccess, SuccessBB,
                       CI->isWeak() ? FailureBB : LoopBB);

  // Make sure later instructions don't get reordered with a fence if necessary.
  Builder.SetInsertPoint(SuccessBB);
  TLI->emitTrailingFence(Builder, SuccessOrder, /*IsStore=*/true,
                         /*IsLoad=*/true);
  Builder.CreateBr(ExitBB);

  Builder.SetInsertPoint(FailureBB);
  TLI->emitTrailingFence(Builder, FailureOrder, /*IsStore=*/true,
                         /*IsLoad=*/true);
  Builder.CreateBr(ExitBB);

  // Finally, we have control-flow based knowledge of whether the cmpxchg
  // succeeded or not. We expose this to later passes by converting any
  // subsequent "icmp eq/ne %loaded, %oldval" into a use of an appropriate PHI.

  // Setup the builder so we can create any PHIs we need.
  Builder.SetInsertPoint(ExitBB, ExitBB->begin());
  PHINode *Success = Builder.CreatePHI(Type::getInt1Ty(Ctx), 2);
  Success->addIncoming(ConstantInt::getTrue(Ctx), SuccessBB);
  Success->addIncoming(ConstantInt::getFalse(Ctx), FailureBB);

  // Look for any users of the cmpxchg that are just comparing the loaded value
  // against the desired one, and replace them with the CFG-derived version.
  SmallVector<ExtractValueInst *, 2> PrunedInsts;
  for (auto User : CI->users()) {
//.........这里部分代码省略.........

/// SplitCriticalEdge - If this edge is a critical edge, insert a new node to
/// split the critical edge.  This will update DominatorTree information if it
/// is available, thus calling this pass will not invalidate either of them.
/// This returns the new block if the edge was split, null otherwise.
///
/// If MergeIdenticalEdges is true (not the default), *all* edges from TI to the
/// specified successor will be merged into the same critical edge block.
/// This is most commonly interesting with switch instructions, which may
/// have many edges to any one destination.  This ensures that all edges to that
/// dest go to one block instead of each going to a different block, but isn't
/// the standard definition of a "critical edge".
///
/// It is invalid to call this function on a critical edge that starts at an
/// IndirectBrInst.  Splitting these edges will almost always create an invalid
/// program because the address of the new block won't be the one that is jumped
/// to.
///
BasicBlock *llvm::SplitCriticalEdge(TerminatorInst *TI, unsigned SuccNum,
                                    Pass *P, bool MergeIdenticalEdges,
                                    bool DontDeleteUselessPhis,
                                    bool SplitLandingPads) {
  if (!isCriticalEdge(TI, SuccNum, MergeIdenticalEdges)) return 0;

  assert(!isa<IndirectBrInst>(TI) &&
         "Cannot split critical edge from IndirectBrInst");

  BasicBlock *TIBB = TI->getParent();
  BasicBlock *DestBB = TI->getSuccessor(SuccNum);

  // Splitting the critical edge to a landing pad block is non-trivial. Don't do
  // it in this generic function.
  if (DestBB->isLandingPad()) return 0;

  // Create a new basic block, linking it into the CFG.
  BasicBlock *NewBB = BasicBlock::Create(TI->getContext(),
                      TIBB->getName() + "." + DestBB->getName() + "_crit_edge");
  // Create our unconditional branch.
  BranchInst *NewBI = BranchInst::Create(DestBB, NewBB);
  NewBI->setDebugLoc(TI->getDebugLoc());

  // Branch to the new block, breaking the edge.
  TI->setSuccessor(SuccNum, NewBB);

  // Insert the block into the function... right after the block TI lives in.
  Function &F = *TIBB->getParent();
  Function::iterator FBBI = TIBB;
  F.getBasicBlockList().insert(++FBBI, NewBB);

  // If there are any PHI nodes in DestBB, we need to update them so that they
  // merge incoming values from NewBB instead of from TIBB.
  {
    unsigned BBIdx = 0;
    for (BasicBlock::iterator I = DestBB->begin(); isa<PHINode>(I); ++I) {
      // We no longer enter through TIBB, now we come in through NewBB.
      // Revector exactly one entry in the PHI node that used to come from
      // TIBB to come from NewBB.
      PHINode *PN = cast<PHINode>(I);

      // Reuse the previous value of BBIdx if it lines up.  In cases where we
      // have multiple phi nodes with *lots* of predecessors, this is a speed
      // win because we don't have to scan the PHI looking for TIBB.  This
      // happens because the BB list of PHI nodes are usually in the same
      // order.
      if (PN->getIncomingBlock(BBIdx) != TIBB)
        BBIdx = PN->getBasicBlockIndex(TIBB);
      PN->setIncomingBlock(BBIdx, NewBB);
    }
  }

  // If there are any other edges from TIBB to DestBB, update those to go
  // through the split block, making those edges non-critical as well (and
  // reducing the number of phi entries in the DestBB if relevant).
  if (MergeIdenticalEdges) {
    for (unsigned i = SuccNum+1, e = TI->getNumSuccessors(); i != e; ++i) {
      if (TI->getSuccessor(i) != DestBB) continue;

      // Remove an entry for TIBB from DestBB phi nodes.
      DestBB->removePredecessor(TIBB, DontDeleteUselessPhis);

      // We found another edge to DestBB, go to NewBB instead.
      TI->setSuccessor(i, NewBB);
    }
  }



  // If we don't have a pass object, we can't update anything...
  if (P == 0) return NewBB;

  DominatorTreeWrapperPass *DTWP =
      P->getAnalysisIfAvailable<DominatorTreeWrapperPass>();
  DominatorTree *DT = DTWP ? &DTWP->getDomTree() : 0;
  LoopInfo *LI = P->getAnalysisIfAvailable<LoopInfo>();

  // If we have nothing to update, just return.
  if (DT == 0 && LI == 0)
    return NewBB;

  // Now update analysis information.  Since the only predecessor of NewBB is
  // the TIBB, TIBB clearly dominates NewBB.  TIBB usually doesn't dominate
//.........这里部分代码省略.........

/// InsertUniqueBackedgeBlock - 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.
///
BasicBlock *
LoopSimplify::InsertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader) {
  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 0;

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

  // Figure out which basic blocks contain back-edges to the loop header.
  std::vector<BasicBlock*> BackedgeBlocks;
  for (pred_iterator I = pred_begin(Header), E = pred_end(Header); I != E; ++I){
    BasicBlock *P = *I;

    // Indirectbr edges cannot be split, so we must fail if we find one.
    if (isa<IndirectBrInst>(P->getTerminator()))
      return 0;

    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);

  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(); ++InsertPos;
  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);
    if (AA) AA->copyValue(PN, NewPN);

    // 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 = 0;
    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 == 0)
            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);
      if (AA) AA->deleteValue(NewPN);
      BEBlock->getInstList().erase(NewPN);
    }
  }

  // Now that all of the PHI nodes have been inserted and adjusted, modify the
  // backedge blocks to just to the BEBlock instead of the header.
  for (unsigned i = 0, e = BackedgeBlocks.size(); i != e; ++i) {
    TerminatorInst *TI = BackedgeBlocks[i]->getTerminator();
//.........这里部分代码省略.........

template<class T, class Callback> void postCommitOptimiseBlocks(T itstart, T itend, Callback& CB, Function::iterator& firstFailedBlock) {

  // Zap any instructions we've created that are trivially dead.
  // TODO: improve DIE to catch more cases like this before synthesis, or adopt
  // on-demand synthesis to similar effect.

  std::vector<Instruction*> Del;

  for(T it = itstart; it != itend; ++it) {

    BasicBlock* BB = it;
    for(BasicBlock::iterator II = BB->begin(), IE = BB->end(); II != IE; ++II) {

      if(isInstructionTriviallyDead(II, GlobalTLI))
	Del.push_back(II);

    }

  }

  for(std::vector<Instruction*>::iterator Delit = Del.begin(), Delend = Del.end(); Delit != Delend; ++Delit)
    DeleteDeadInstruction(*Delit);

  Del.clear();

  // Now coalesce any long chains of BBs.

  std::vector<std::vector<BasicBlock*> > Chains;

  DenseSet<BasicBlock*> Seen;

  for(T it = itstart; it != itend; ++it) {

    BasicBlock* BB = it;
    if(!Seen.insert(BB).second)
      continue;

    // Find chain start:
    while(BasicBlock* PrevBB = getChainPrev(BB))
      BB = PrevBB;

    Seen.insert(BB);

    Chains.resize(Chains.size() + 1);
    std::vector<BasicBlock*>& Chain = Chains.back();
    Chain.push_back(BB);

    while(BasicBlock* NextBB = getChainNext(BB)) {
      Chain.push_back(NextBB);
      Seen.insert(NextBB);
      BB = NextBB;
    }

    if(Chain.size() == 1)
      Chains.pop_back();
    else {

      /*
      errs() << "Chain: ";

      for(std::vector<BasicBlock*>::iterator it = Chain.begin(), 
	    itend = Chain.end(); it != itend; ++it) {

	if(it != Chain.begin())
	  errs() << ", ";
	
	errs() << (*it)->getName();

      }
      
      errs() << "\n";
      */

    }

  }

  // Merge each chain found

  for(std::vector<std::vector<BasicBlock*> >::iterator chainit = Chains.begin(),
	itend = Chains.end(); chainit != itend; ++chainit) {

    std::vector<BasicBlock*>& Chain = *chainit;

    BasicBlock* Start = Chain[0];
    CB.willReplace(Start);

    for(unsigned i = 1, ilim = Chain.size(); i != ilim; ++i) {

      if(i != 0 && (i % 10000 == 0)) 
	errs() << ".";

      BasicBlock* PBB = Chain.back()->getSinglePredecessor();

      // First failed block goes away; next one takes its place.
      if(Function::iterator(PBB) == firstFailedBlock)
	++firstFailedBlock;

      MergeBasicBlockIntoOnlyPred(Chain.back());

//.........这里部分代码省略.........