C++ DominatorTree类代码示例

/// Return true if the specified block dominates at least
/// one of the blocks in the specified list.
static bool
blockDominatesAnExit(BasicBlock *BB,
                     DominatorTree &DT,
                     const SmallVectorImpl<BasicBlock *> &ExitBlocks) {
    DomTreeNode *DomNode = DT.getNode(BB);
    return any_of(ExitBlocks, [&](BasicBlock *EB) {
        return DT.dominates(DomNode, DT.getNode(EB));
    });
}

void DominatorTree::verifyAnalysis() const {
  if (!VerifyDomInfo) return;

  Function &F = *getRoot()->getParent();

  DominatorTree OtherDT;
  OtherDT.getBase().recalculate(F);
  assert(!compare(OtherDT) && "Invalid DominatorTree info!");
}

static void clearDomtree(Function *F, DominatorTree &DT) {
  DomTreeNode *N = DT.getNode(&F->getEntryBlock());
  std::vector<BasicBlock *> Nodes;
  for (po_iterator<DomTreeNode *> I = po_begin(N), E = po_end(N); I != E; ++I)
    Nodes.push_back(I->getBlock());

  for (std::vector<BasicBlock *>::iterator I = Nodes.begin(), E = Nodes.end();
       I != E; ++I)
    DT.eraseNode(*I);
}

void DominatorTree::verifyAnalysis() const {
  if (!VerifyDomInfo) return;

  Function &F = *getRoot()->getParent();

  DominatorTree OtherDT;
  OtherDT.getBase().recalculate(F);
  if (compare(OtherDT)) {
    errs() << "DominatorTree is not up to date!\nComputed:\n";
    print(errs());
    errs() << "\nActual:\n";
    OtherDT.print(errs());
    abort();
  }
}

bool polly::isHoistableLoad(LoadInst *LInst, Region &R, LoopInfo &LI,
                            ScalarEvolution &SE, const DominatorTree &DT) {
  Loop *L = LI.getLoopFor(LInst->getParent());
  auto *Ptr = LInst->getPointerOperand();
  const SCEV *PtrSCEV = SE.getSCEVAtScope(Ptr, L);
  while (L && R.contains(L)) {
    if (!SE.isLoopInvariant(PtrSCEV, L))
      return false;
    L = L->getParentLoop();
  }

  for (auto *User : Ptr->users()) {
    auto *UserI = dyn_cast<Instruction>(User);
    if (!UserI || !R.contains(UserI))
      continue;
    if (!UserI->mayWriteToMemory())
      continue;

    auto &BB = *UserI->getParent();
    bool DominatesAllPredecessors = true;
    for (auto Pred : predecessors(R.getExit()))
      if (R.contains(Pred) && !DT.dominates(&BB, Pred))
        DominatesAllPredecessors = false;

    if (!DominatesAllPredecessors)
      continue;

    return false;
  }

  return true;
}

/// isLCSSAForm - Return true if the Loop is in LCSSA form
bool Loop::isLCSSAForm(DominatorTree &DT) const {
  // Sort the blocks vector so that we can use binary search to do quick
  // lookups.
  SmallPtrSet<BasicBlock*, 16> LoopBBs(block_begin(), block_end());

  for (block_iterator BI = block_begin(), E = block_end(); BI != E; ++BI) {
    BasicBlock *BB = *BI;
    for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E;++I)
      for (Value::use_iterator UI = I->use_begin(), E = I->use_end(); UI != E;
           ++UI) {
        User *U = *UI;
        BasicBlock *UserBB = cast<Instruction>(U)->getParent();
        if (PHINode *P = dyn_cast<PHINode>(U))
          UserBB = P->getIncomingBlock(UI);

        // Check the current block, as a fast-path, before checking whether
        // the use is anywhere in the loop.  Most values are used in the same
        // block they are defined in.  Also, blocks not reachable from the
        // entry are special; uses in them don't need to go through PHIs.
        if (UserBB != BB &&
            !LoopBBs.count(UserBB) &&
            DT.isReachableFromEntry(UserBB))
          return false;
      }
  }

  return true;
}

// Check that 'BB' doesn't have any uses outside of the 'L'
static bool isBlockInLCSSAForm(const Loop &L, const BasicBlock &BB,
                               DominatorTree &DT) {
  for (const Instruction &I : BB) {
    // Tokens can't be used in PHI nodes and live-out tokens prevent loop
    // optimizations, so for the purposes of considered LCSSA form, we
    // can ignore them.
    if (I.getType()->isTokenTy())
      continue;

    for (const Use &U : I.uses()) {
      const Instruction *UI = cast<Instruction>(U.getUser());
      const BasicBlock *UserBB = UI->getParent();
      if (const PHINode *P = dyn_cast<PHINode>(UI))
        UserBB = P->getIncomingBlock(U);

      // Check the current block, as a fast-path, before checking whether
      // the use is anywhere in the loop.  Most values are used in the same
      // block they are defined in.  Also, blocks not reachable from the
      // entry are special; uses in them don't need to go through PHIs.
      if (UserBB != &BB && !L.contains(UserBB) &&
          DT.isReachableFromEntry(UserBB))
        return false;
    }
  }
  return true;
}

Function *PartialInlinerImpl::FunctionCloner::doFunctionOutlining() {
  // Returns true if the block is to be partial inlined into the caller
  // (i.e. not to be extracted to the out of line function)
  auto ToBeInlined = [&, this](BasicBlock *BB) {
    return BB == ClonedOI->ReturnBlock ||
           (std::find(ClonedOI->Entries.begin(), ClonedOI->Entries.end(), BB) !=
            ClonedOI->Entries.end());
  };

  // Gather up the blocks that we're going to extract.
  std::vector<BasicBlock *> ToExtract;
  ToExtract.push_back(ClonedOI->NonReturnBlock);
  OutlinedRegionCost +=
      PartialInlinerImpl::computeBBInlineCost(ClonedOI->NonReturnBlock);
  for (BasicBlock &BB : *ClonedFunc)
    if (!ToBeInlined(&BB) && &BB != ClonedOI->NonReturnBlock) {
      ToExtract.push_back(&BB);
      // FIXME: the code extractor may hoist/sink more code
      // into the outlined function which may make the outlining
      // overhead (the difference of the outlined function cost
      // and OutliningRegionCost) look larger.
      OutlinedRegionCost += computeBBInlineCost(&BB);
    }

  // The CodeExtractor needs a dominator tree.
  DominatorTree DT;
  DT.recalculate(*ClonedFunc);

  // Manually calculate a BlockFrequencyInfo and BranchProbabilityInfo.
  LoopInfo LI(DT);
  BranchProbabilityInfo BPI(*ClonedFunc, LI);
  ClonedFuncBFI.reset(new BlockFrequencyInfo(*ClonedFunc, BPI, LI));

  // Extract the body of the if.
  OutlinedFunc = CodeExtractor(ToExtract, &DT, /*AggregateArgs*/ false,
                               ClonedFuncBFI.get(), &BPI)
                     .extractCodeRegion();

  if (OutlinedFunc) {
    OutliningCallBB = PartialInlinerImpl::getOneCallSiteTo(OutlinedFunc)
        .getInstruction()
        ->getParent();
    assert(OutliningCallBB->getParent() == ClonedFunc);
  }

  return OutlinedFunc;
}

/// SinkInstruction - Determine whether it is safe to sink the specified machine
/// instruction out of its current block into a successor.
static bool SinkInstruction(Instruction *Inst,
                            SmallPtrSetImpl<Instruction *> &Stores,
                            DominatorTree &DT, LoopInfo &LI, AAResults &AA) {

  // Don't sink static alloca instructions.  CodeGen assumes allocas outside the
  // entry block are dynamically sized stack objects.
  if (AllocaInst *AI = dyn_cast<AllocaInst>(Inst))
    if (AI->isStaticAlloca())
      return false;

  // Check if it's safe to move the instruction.
  if (!isSafeToMove(Inst, AA, Stores))
    return false;

  // FIXME: This should include support for sinking instructions within the
  // block they are currently in to shorten the live ranges.  We often get
  // instructions sunk into the top of a large block, but it would be better to
  // also sink them down before their first use in the block.  This xform has to
  // be careful not to *increase* register pressure though, e.g. sinking
  // "x = y + z" down if it kills y and z would increase the live ranges of y
  // and z and only shrink the live range of x.

  // SuccToSinkTo - This is the successor to sink this instruction to, once we
  // decide.
  BasicBlock *SuccToSinkTo = nullptr;

  // Instructions can only be sunk if all their uses are in blocks
  // dominated by one of the successors.
  // Look at all the dominated blocks and see if we can sink it in one.
  DomTreeNode *DTN = DT.getNode(Inst->getParent());
  for (DomTreeNode::iterator I = DTN->begin(), E = DTN->end();
      I != E && SuccToSinkTo == nullptr; ++I) {
    BasicBlock *Candidate = (*I)->getBlock();
    // A node always immediate-dominates its children on the dominator
    // tree.
    if (IsAcceptableTarget(Inst, Candidate, DT, LI))
      SuccToSinkTo = Candidate;
  }

  // If no suitable postdominator was found, look at all the successors and
  // decide which one we should sink to, if any.
  for (succ_iterator I = succ_begin(Inst->getParent()),
      E = succ_end(Inst->getParent()); I != E && !SuccToSinkTo; ++I) {
    if (IsAcceptableTarget(Inst, *I, DT, LI))
      SuccToSinkTo = *I;
  }

  // If we couldn't find a block to sink to, ignore this instruction.
  if (!SuccToSinkTo)
    return false;

  LLVM_DEBUG(dbgs() << "Sink" << *Inst << " (";
             Inst->getParent()->printAsOperand(dbgs(), false); dbgs() << " -> ";
             SuccToSinkTo->printAsOperand(dbgs(), false); dbgs() << ")\n");

  // Move the instruction.
  Inst->moveBefore(&*SuccToSinkTo->getFirstInsertionPt());
  return true;
}

bool SanitizerCoverageModule::runOnFunction(Function &F) {
  if (F.empty()) return false;
  if (F.getName().find(".module_ctor") != std::string::npos)
    return false;  // Should not instrument sanitizer init functions.
  // Don't instrument functions using SEH for now. Splitting basic blocks like
  // we do for coverage breaks WinEHPrepare.
  // FIXME: Remove this when SEH no longer uses landingpad pattern matching.
  if (F.hasPersonalityFn() &&
      isAsynchronousEHPersonality(classifyEHPersonality(F.getPersonalityFn())))
    return false;
  if (Options.CoverageType >= SanitizerCoverageOptions::SCK_Edge)
    SplitAllCriticalEdges(F);
  SmallVector<Instruction*, 8> IndirCalls;
  SmallVector<BasicBlock *, 16> BlocksToInstrument;
  SmallVector<Instruction*, 8> CmpTraceTargets;
  SmallVector<Instruction*, 8> SwitchTraceTargets;

  DominatorTree DT;
  DT.recalculate(F);
  for (auto &BB : F) {
    if (shouldInstrumentBlock(&BB, &DT))
      BlocksToInstrument.push_back(&BB);
    for (auto &Inst : BB) {
      if (Options.IndirectCalls) {
        CallSite CS(&Inst);
        if (CS && !CS.getCalledFunction())
          IndirCalls.push_back(&Inst);
      }
      if (Options.TraceCmp) {
        if (isa<ICmpInst>(&Inst))
          CmpTraceTargets.push_back(&Inst);
        if (isa<SwitchInst>(&Inst))
          SwitchTraceTargets.push_back(&Inst);
      }
    }
  }

  InjectCoverage(F, BlocksToInstrument);
  InjectCoverageForIndirectCalls(F, IndirCalls);
  InjectTraceForCmp(F, CmpTraceTargets);
  InjectTraceForSwitch(F, SwitchTraceTargets);
  return true;
}

OptimizationRemarkEmitter::OptimizationRemarkEmitter(const Function *F)
    : F(F), BFI(nullptr) {
  if (!F->getContext().getDiagnosticHotnessRequested())
    return;

  // First create a dominator tree.
  DominatorTree DT;
  DT.recalculate(*const_cast<Function *>(F));

  // Generate LoopInfo from it.
  LoopInfo LI;
  LI.analyze(DT);

  // Then compute BranchProbabilityInfo.
  BranchProbabilityInfo BPI;
  BPI.calculate(*F, LI);

  // Finally compute BFI.
  OwnedBFI = llvm::make_unique<BlockFrequencyInfo>(*F, BPI, LI);
  BFI = OwnedBFI.get();
}

/// Returns true if this loop is known to contain a call safepoint which
/// must unconditionally execute on any iteration of the loop which returns
/// to the loop header via an edge from Pred.  Returns a conservative correct
/// answer; i.e. false is always valid.
static bool containsUnconditionalCallSafepoint(Loop *L, BasicBlock *Header,
                                               BasicBlock *Pred,
                                               DominatorTree &DT,
                                               const TargetLibraryInfo &TLI) {
  // In general, we're looking for any cut of the graph which ensures
  // there's a call safepoint along every edge between Header and Pred.
  // For the moment, we look only for the 'cuts' that consist of a single call
  // instruction in a block which is dominated by the Header and dominates the
  // loop latch (Pred) block.  Somewhat surprisingly, walking the entire chain
  // of such dominating blocks gets substantially more occurrences than just
  // checking the Pred and Header blocks themselves.  This may be due to the
  // density of loop exit conditions caused by range and null checks.
  // TODO: structure this as an analysis pass, cache the result for subloops,
  // avoid dom tree recalculations
  assert(DT.dominates(Header, Pred) && "loop latch not dominated by header?");

  BasicBlock *Current = Pred;
  while (true) {
    for (Instruction &I : *Current) {
      if (auto CS = CallSite(&I))
        // Note: Technically, needing a safepoint isn't quite the right
        // condition here.  We should instead be checking if the target method
        // has an
        // unconditional poll. In practice, this is only a theoretical concern
        // since we don't have any methods with conditional-only safepoint
        // polls.
        if (needsStatepoint(CS, TLI))
          return true;
    }

    if (Current == Header)
      break;
    Current = DT.getNode(Current)->getIDom()->getBlock();
  }

  return false;
}

/// Return a set of basic blocks to insert sinked instructions.
///
/// The returned set of basic blocks (BBsToSinkInto) should satisfy:
///
/// * Inside the loop \p L
/// * For each UseBB in \p UseBBs, there is at least one BB in BBsToSinkInto
///   that domintates the UseBB
/// * Has minimum total frequency that is no greater than preheader frequency
///
/// The purpose of the function is to find the optimal sinking points to
/// minimize execution cost, which is defined as "sum of frequency of
/// BBsToSinkInto".
/// As a result, the returned BBsToSinkInto needs to have minimum total
/// frequency.
/// Additionally, if the total frequency of BBsToSinkInto exceeds preheader
/// frequency, the optimal solution is not sinking (return empty set).
///
/// \p ColdLoopBBs is used to help find the optimal sinking locations.
/// It stores a list of BBs that is:
///
/// * Inside the loop \p L
/// * Has a frequency no larger than the loop's preheader
/// * Sorted by BB frequency
///
/// The complexity of the function is O(UseBBs.size() * ColdLoopBBs.size()).
/// To avoid expensive computation, we cap the maximum UseBBs.size() in its
/// caller.
static SmallPtrSet<BasicBlock *, 2>
findBBsToSinkInto(const Loop &L, const SmallPtrSetImpl<BasicBlock *> &UseBBs,
                  const SmallVectorImpl<BasicBlock *> &ColdLoopBBs,
                  DominatorTree &DT, BlockFrequencyInfo &BFI) {
  SmallPtrSet<BasicBlock *, 2> BBsToSinkInto;
  if (UseBBs.size() == 0)
    return BBsToSinkInto;

  BBsToSinkInto.insert(UseBBs.begin(), UseBBs.end());
  SmallPtrSet<BasicBlock *, 2> BBsDominatedByColdestBB;

  // For every iteration:
  //   * Pick the ColdestBB from ColdLoopBBs
  //   * Find the set BBsDominatedByColdestBB that satisfy:
  //     - BBsDominatedByColdestBB is a subset of BBsToSinkInto
  //     - Every BB in BBsDominatedByColdestBB is dominated by ColdestBB
  //   * If Freq(ColdestBB) < Freq(BBsDominatedByColdestBB), remove
  //     BBsDominatedByColdestBB from BBsToSinkInto, add ColdestBB to
  //     BBsToSinkInto
  for (BasicBlock *ColdestBB : ColdLoopBBs) {
    BBsDominatedByColdestBB.clear();
    for (BasicBlock *SinkedBB : BBsToSinkInto)
      if (DT.dominates(ColdestBB, SinkedBB))
        BBsDominatedByColdestBB.insert(SinkedBB);
    if (BBsDominatedByColdestBB.size() == 0)
      continue;
    if (adjustedSumFreq(BBsDominatedByColdestBB, BFI) >
        BFI.getBlockFreq(ColdestBB)) {
      for (BasicBlock *DominatedBB : BBsDominatedByColdestBB) {
        BBsToSinkInto.erase(DominatedBB);
      }
      BBsToSinkInto.insert(ColdestBB);
    }
  }

  // If the total frequency of BBsToSinkInto is larger than preheader frequency,
  // do not sink.
  if (adjustedSumFreq(BBsToSinkInto, BFI) >
      BFI.getBlockFreq(L.getLoopPreheader()))
    BBsToSinkInto.clear();
  return BBsToSinkInto;
}

/// IsAcceptableTarget - Return true if it is possible to sink the instruction
/// in the specified basic block.
static bool IsAcceptableTarget(Instruction *Inst, BasicBlock *SuccToSinkTo,
                               DominatorTree &DT, LoopInfo &LI) {
  assert(Inst && "Instruction to be sunk is null");
  assert(SuccToSinkTo && "Candidate sink target is null");

  // It is not possible to sink an instruction into its own block.  This can
  // happen with loops.
  if (Inst->getParent() == SuccToSinkTo)
    return false;

  // It's never legal to sink an instruction into a block which terminates in an
  // EH-pad.
  if (SuccToSinkTo->getTerminator()->isExceptionalTerminator())
    return false;

  // If the block has multiple predecessors, this would introduce computation
  // on different code paths.  We could split the critical edge, but for now we
  // just punt.
  // FIXME: Split critical edges if not backedges.
  if (SuccToSinkTo->getUniquePredecessor() != Inst->getParent()) {
    // We cannot sink a load across a critical edge - there may be stores in
    // other code paths.
    if (Inst->mayReadFromMemory())
      return false;

    // We don't want to sink across a critical edge if we don't dominate the
    // successor. We could be introducing calculations to new code paths.
    if (!DT.dominates(Inst->getParent(), SuccToSinkTo))
      return false;

    // Don't sink instructions into a loop.
    Loop *succ = LI.getLoopFor(SuccToSinkTo);
    Loop *cur = LI.getLoopFor(Inst->getParent());
    if (succ != nullptr && succ != cur)
      return false;
  }

  // Finally, check that all the uses of the instruction are actually
  // dominated by the candidate
  return AllUsesDominatedByBlock(Inst, SuccToSinkTo, DT);
}

/// AllUsesDominatedByBlock - Return true if all uses of the specified value
/// occur in blocks dominated by the specified block.
static bool AllUsesDominatedByBlock(Instruction *Inst, BasicBlock *BB,
                                    DominatorTree &DT) {
  // Ignoring debug uses is necessary so debug info doesn't affect the code.
  // This may leave a referencing dbg_value in the original block, before
  // the definition of the vreg.  Dwarf generator handles this although the
  // user might not get the right info at runtime.
  for (Use &U : Inst->uses()) {
    // Determine the block of the use.
    Instruction *UseInst = cast<Instruction>(U.getUser());
    BasicBlock *UseBlock = UseInst->getParent();
    if (PHINode *PN = dyn_cast<PHINode>(UseInst)) {
      // PHI nodes use the operand in the predecessor block, not the block with
      // the PHI.
      unsigned Num = PHINode::getIncomingValueNumForOperand(U.getOperandNo());
      UseBlock = PN->getIncomingBlock(Num);
    }
    // Check that it dominates.
    if (!DT.dominates(BB, UseBlock))
      return false;
  }
  return true;
}