#define DEBUG_TYPE "loop-rotate"
#include "llvm/Transforms/Scalar.h"
-#include "llvm/Function.h"
-#include "llvm/IntrinsicInst.h"
+#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/CodeMetrics.h"
-#include "llvm/Analysis/LoopPass.h"
#include "llvm/Analysis/InstructionSimplify.h"
+#include "llvm/Analysis/LoopPass.h"
#include "llvm/Analysis/ScalarEvolution.h"
+#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
-#include "llvm/Transforms/Utils/Local.h"
+#include "llvm/IR/CFG.h"
+#include "llvm/IR/Dominators.h"
+#include "llvm/IR/Function.h"
+#include "llvm/IR/IntrinsicInst.h"
+#include "llvm/Support/Debug.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
+#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/SSAUpdater.h"
#include "llvm/Transforms/Utils/ValueMapper.h"
-#include "llvm/Support/CFG.h"
-#include "llvm/Support/Debug.h"
-#include "llvm/ADT/Statistic.h"
using namespace llvm;
#define MAX_HEADER_SIZE 16
}
// LCSSA form makes instruction renaming easier.
- virtual void getAnalysisUsage(AnalysisUsage &AU) const {
- AU.addPreserved<DominatorTree>();
+ void getAnalysisUsage(AnalysisUsage &AU) const override {
+ AU.addPreserved<DominatorTreeWrapperPass>();
AU.addRequired<LoopInfo>();
AU.addPreserved<LoopInfo>();
AU.addRequiredID(LoopSimplifyID);
AU.addRequiredID(LCSSAID);
AU.addPreservedID(LCSSAID);
AU.addPreserved<ScalarEvolution>();
+ AU.addRequired<TargetTransformInfo>();
}
- bool runOnLoop(Loop *L, LPPassManager &LPM);
- void simplifyLoopLatch(Loop *L);
- bool rotateLoop(Loop *L);
+ bool runOnLoop(Loop *L, LPPassManager &LPM) override;
+ bool simplifyLoopLatch(Loop *L);
+ bool rotateLoop(Loop *L, bool SimplifiedLatch);
private:
LoopInfo *LI;
+ const TargetTransformInfo *TTI;
};
}
char LoopRotate::ID = 0;
INITIALIZE_PASS_BEGIN(LoopRotate, "loop-rotate", "Rotate Loops", false, false)
+INITIALIZE_AG_DEPENDENCY(TargetTransformInfo)
INITIALIZE_PASS_DEPENDENCY(LoopInfo)
INITIALIZE_PASS_DEPENDENCY(LoopSimplify)
INITIALIZE_PASS_DEPENDENCY(LCSSA)
/// Rotate Loop L as many times as possible. Return true if
/// the loop is rotated at least once.
bool LoopRotate::runOnLoop(Loop *L, LPPassManager &LPM) {
+ if (skipOptnoneFunction(L))
+ return false;
+
LI = &getAnalysis<LoopInfo>();
+ TTI = &getAnalysis<TargetTransformInfo>();
// Simplify the loop latch before attempting to rotate the header
// upward. Rotation may not be needed if the loop tail can be folded into the
// loop exit.
- simplifyLoopLatch(L);
+ bool SimplifiedLatch = simplifyLoopLatch(L);
// One loop can be rotated multiple times.
bool MadeChange = false;
- while (rotateLoop(L))
+ while (rotateLoop(L, SimplifiedLatch)) {
MadeChange = true;
-
+ SimplifiedLatch = false;
+ }
return MadeChange;
}
for (Value::use_iterator UI = OrigHeaderVal->use_begin(),
UE = OrigHeaderVal->use_end(); UI != UE; ) {
// Grab the use before incrementing the iterator.
- Use &U = UI.getUse();
+ Use &U = *UI;
// Increment the iterator before removing the use from the list.
++UI;
/// canonical form so downstream passes can handle it.
///
/// I don't believe this invalidates SCEV.
-void LoopRotate::simplifyLoopLatch(Loop *L) {
+bool LoopRotate::simplifyLoopLatch(Loop *L) {
BasicBlock *Latch = L->getLoopLatch();
if (!Latch || Latch->hasAddressTaken())
- return;
+ return false;
BranchInst *Jmp = dyn_cast<BranchInst>(Latch->getTerminator());
if (!Jmp || !Jmp->isUnconditional())
- return;
+ return false;
BasicBlock *LastExit = Latch->getSinglePredecessor();
if (!LastExit || !L->isLoopExiting(LastExit))
- return;
+ return false;
BranchInst *BI = dyn_cast<BranchInst>(LastExit->getTerminator());
if (!BI)
- return;
+ return false;
if (!shouldSpeculateInstrs(Latch->begin(), Jmp))
- return;
+ return false;
DEBUG(dbgs() << "Folding loop latch " << Latch->getName() << " into "
<< LastExit->getName() << "\n");
// Nuke the Latch block.
assert(Latch->empty() && "unable to evacuate Latch");
LI->removeBlock(Latch);
- if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>())
- DT->eraseNode(Latch);
+ if (DominatorTreeWrapperPass *DTWP =
+ getAnalysisIfAvailable<DominatorTreeWrapperPass>())
+ DTWP->getDomTree().eraseNode(Latch);
Latch->eraseFromParent();
+ return true;
}
/// Rotate loop LP. Return true if the loop is rotated.
-bool LoopRotate::rotateLoop(Loop *L) {
+///
+/// \param SimplifiedLatch is true if the latch was just folded into the final
+/// loop exit. In this case we may want to rotate even though the new latch is
+/// now an exiting branch. This rotation would have happened had the latch not
+/// been simplified. However, if SimplifiedLatch is false, then we avoid
+/// rotating loops in which the latch exits to avoid excessive or endless
+/// rotation. LoopRotate should be repeatable and converge to a canonical
+/// form. This property is satisfied because simplifying the loop latch can only
+/// happen once across multiple invocations of the LoopRotate pass.
+bool LoopRotate::rotateLoop(Loop *L, bool SimplifiedLatch) {
// If the loop has only one block then there is not much to rotate.
if (L->getBlocks().size() == 1)
return false;
// If the loop latch already contains a branch that leaves the loop then the
// loop is already rotated.
- if (OrigLatch == 0 || L->isLoopExiting(OrigLatch))
+ if (OrigLatch == 0)
return false;
- // Check size of original header and reject loop if it is very big.
+ // Rotate if either the loop latch does *not* exit the loop, or if the loop
+ // latch was just simplified.
+ if (L->isLoopExiting(OrigLatch) && !SimplifiedLatch)
+ return false;
+
+ // Check size of original header and reject loop if it is very big or we can't
+ // duplicate blocks inside it.
{
CodeMetrics Metrics;
- Metrics.analyzeBasicBlock(OrigHeader);
+ Metrics.analyzeBasicBlock(OrigHeader, *TTI);
+ if (Metrics.notDuplicatable) {
+ DEBUG(dbgs() << "LoopRotation: NOT rotating - contains non-duplicatable"
+ << " instructions: "; L->dump());
+ return false;
+ }
if (Metrics.NumInsts > MAX_HEADER_SIZE)
return false;
}
// The conditional branch can't be folded, handle the general case.
// Update DominatorTree to reflect the CFG change we just made. Then split
// edges as necessary to preserve LoopSimplify form.
- if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>()) {
+ if (DominatorTreeWrapperPass *DTWP =
+ getAnalysisIfAvailable<DominatorTreeWrapperPass>()) {
+ DominatorTree &DT = DTWP->getDomTree();
// Everything that was dominated by the old loop header is now dominated
// by the original loop preheader. Conceptually the header was merged
// into the preheader, even though we reuse the actual block as a new
// loop latch.
- DomTreeNode *OrigHeaderNode = DT->getNode(OrigHeader);
+ DomTreeNode *OrigHeaderNode = DT.getNode(OrigHeader);
SmallVector<DomTreeNode *, 8> HeaderChildren(OrigHeaderNode->begin(),
OrigHeaderNode->end());
- DomTreeNode *OrigPreheaderNode = DT->getNode(OrigPreheader);
+ DomTreeNode *OrigPreheaderNode = DT.getNode(OrigPreheader);
for (unsigned I = 0, E = HeaderChildren.size(); I != E; ++I)
- DT->changeImmediateDominator(HeaderChildren[I], OrigPreheaderNode);
+ DT.changeImmediateDominator(HeaderChildren[I], OrigPreheaderNode);
- assert(DT->getNode(Exit)->getIDom() == OrigPreheaderNode);
- assert(DT->getNode(NewHeader)->getIDom() == OrigPreheaderNode);
+ assert(DT.getNode(Exit)->getIDom() == OrigPreheaderNode);
+ assert(DT.getNode(NewHeader)->getIDom() == OrigPreheaderNode);
// Update OrigHeader to be dominated by the new header block.
- DT->changeImmediateDominator(OrigHeader, OrigLatch);
+ DT.changeImmediateDominator(OrigHeader, OrigLatch);
}
// Right now OrigPreHeader has two successors, NewHeader and ExitBlock, and
NewPH->setName(NewHeader->getName() + ".lr.ph");
// Preserve canonical loop form, which means that 'Exit' should have only
- // one predecessor.
- BasicBlock *ExitSplit = SplitCriticalEdge(L->getLoopLatch(), Exit, this);
- ExitSplit->moveBefore(Exit);
+ // one predecessor. Note that Exit could be an exit block for multiple
+ // nested loops, causing both of the edges to now be critical and need to
+ // be split.
+ SmallVector<BasicBlock *, 4> ExitPreds(pred_begin(Exit), pred_end(Exit));
+ bool SplitLatchEdge = false;
+ for (SmallVectorImpl<BasicBlock *>::iterator PI = ExitPreds.begin(),
+ PE = ExitPreds.end();
+ PI != PE; ++PI) {
+ // We only need to split loop exit edges.
+ Loop *PredLoop = LI->getLoopFor(*PI);
+ if (!PredLoop || PredLoop->contains(Exit))
+ continue;
+ SplitLatchEdge |= L->getLoopLatch() == *PI;
+ BasicBlock *ExitSplit = SplitCriticalEdge(*PI, Exit, this);
+ ExitSplit->moveBefore(Exit);
+ }
+ assert(SplitLatchEdge &&
+ "Despite splitting all preds, failed to split latch exit?");
} else {
// We can fold the conditional branch in the preheader, this makes things
// simpler. The first step is to remove the extra edge to the Exit block.
PHBI->eraseFromParent();
// With our CFG finalized, update DomTree if it is available.
- if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>()) {
+ if (DominatorTreeWrapperPass *DTWP =
+ getAnalysisIfAvailable<DominatorTreeWrapperPass>()) {
+ DominatorTree &DT = DTWP->getDomTree();
// Update OrigHeader to be dominated by the new header block.
- DT->changeImmediateDominator(NewHeader, OrigPreheader);
- DT->changeImmediateDominator(OrigHeader, OrigLatch);
+ DT.changeImmediateDominator(NewHeader, OrigPreheader);
+ DT.changeImmediateDominator(OrigHeader, OrigLatch);
// Brute force incremental dominator tree update. Call
// findNearestCommonDominator on all CFG predecessors of each child of the
// original header.
- DomTreeNode *OrigHeaderNode = DT->getNode(OrigHeader);
- SmallVector<DomTreeNode *, 8> WorkList(OrigHeaderNode->begin(),
- OrigHeaderNode->end());
- while (!WorkList.empty()) {
- DomTreeNode *Node = WorkList.pop_back_val();
- BasicBlock *BB = Node->getBlock();
- BasicBlock *NearestDom = 0;
- for (pred_iterator PI = pred_begin(BB), PE = pred_end(BB); PI != PE;
- ++PI) {
- BasicBlock *Pred = *PI;
-
- // We have to process predecessors of a node before we touch the
- // actual node. If one of the predecessors is in our worklist, put it
- // and the currently processed node on the worklist and go processing
- // the predecessor.
- SmallVectorImpl<DomTreeNode *>::iterator I =
- std::find(WorkList.begin(), WorkList.end(), DT->getNode(Pred));
- if (I != WorkList.end()) {
- WorkList.push_back(Node);
- std::swap(*I, WorkList.back());
- // The predecessor is now at the end of the worklist.
- NearestDom = 0;
- break;
+ DomTreeNode *OrigHeaderNode = DT.getNode(OrigHeader);
+ SmallVector<DomTreeNode *, 8> HeaderChildren(OrigHeaderNode->begin(),
+ OrigHeaderNode->end());
+ bool Changed;
+ do {
+ Changed = false;
+ for (unsigned I = 0, E = HeaderChildren.size(); I != E; ++I) {
+ DomTreeNode *Node = HeaderChildren[I];
+ BasicBlock *BB = Node->getBlock();
+
+ pred_iterator PI = pred_begin(BB);
+ BasicBlock *NearestDom = *PI;
+ for (pred_iterator PE = pred_end(BB); PI != PE; ++PI)
+ NearestDom = DT.findNearestCommonDominator(NearestDom, *PI);
+
+ // Remember if this changes the DomTree.
+ if (Node->getIDom()->getBlock() != NearestDom) {
+ DT.changeImmediateDominator(BB, NearestDom);
+ Changed = true;
}
-
- // On the first iteration start with Pred, on the other iterations we
- // narrow it down to the nearest common dominator.
- if (!NearestDom)
- NearestDom = Pred;
- else
- NearestDom = DT->findNearestCommonDominator(NearestDom, Pred);
}
- if (NearestDom)
- DT->changeImmediateDominator(BB, NearestDom);
- }
+ // If the dominator changed, this may have an effect on other
+ // predecessors, continue until we reach a fixpoint.
+ } while (Changed);
}
}
++NumRotated;
return true;
}
-
projects / oota-llvm.git / blobdiff