#define DEBUG_TYPE "loop-rotate"
#include "llvm/Transforms/Scalar.h"
#include "llvm/Function.h"
+#include "llvm/IntrinsicInst.h"
#include "llvm/Analysis/CodeMetrics.h"
#include "llvm/Analysis/LoopPass.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/ScalarEvolution.h"
+#include "llvm/Analysis/ValueTracking.h"
#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include "llvm/Transforms/Utils/SSAUpdater.h"
initializeLoopRotatePass(*PassRegistry::getPassRegistry());
}
- // Rotate Loop L as many times as possible. Return true if
- // loop is rotated at least once.
- bool runOnLoop(Loop *L, LPPassManager &LPM);
-
// LCSSA form makes instruction renaming easier.
virtual void getAnalysisUsage(AnalysisUsage &AU) const {
AU.addPreserved<DominatorTree>();
AU.addPreserved<ScalarEvolution>();
}
- // Helper functions
-
- /// Do actual work
+ bool runOnLoop(Loop *L, LPPassManager &LPM);
+ void simplifyLoopLatch(Loop *L);
bool rotateLoop(Loop *L);
-
- /// After loop rotation, loop pre-header has multiple sucessors.
- /// Insert one forwarding basic block to ensure that loop pre-header
- /// has only one successor.
- void preserveCanonicalLoopForm(Loop *L, BasicBlock *OrigHeader,
- BasicBlock *OrigPreHeader,
- BasicBlock *OrigLatch, BasicBlock *NewHeader,
- BasicBlock *Exit);
private:
LoopInfo *LI;
};
}
-
+
char LoopRotate::ID = 0;
INITIALIZE_PASS_BEGIN(LoopRotate, "loop-rotate", "Rotate Loops", false, false)
INITIALIZE_PASS_DEPENDENCY(LoopInfo)
bool LoopRotate::runOnLoop(Loop *L, LPPassManager &LPM) {
LI = &getAnalysis<LoopInfo>();
+ // 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);
+
// One loop can be rotated multiple times.
bool MadeChange = false;
while (rotateLoop(L))
return MadeChange;
}
+/// RewriteUsesOfClonedInstructions - We just cloned the instructions from the
+/// old header into the preheader. If there were uses of the values produced by
+/// these instruction that were outside of the loop, we have to insert PHI nodes
+/// to merge the two values. Do this now.
+static void RewriteUsesOfClonedInstructions(BasicBlock *OrigHeader,
+ BasicBlock *OrigPreheader,
+ ValueToValueMapTy &ValueMap) {
+ // Remove PHI node entries that are no longer live.
+ BasicBlock::iterator I, E = OrigHeader->end();
+ for (I = OrigHeader->begin(); PHINode *PN = dyn_cast<PHINode>(I); ++I)
+ PN->removeIncomingValue(PN->getBasicBlockIndex(OrigPreheader));
+
+ // Now fix up users of the instructions in OrigHeader, inserting PHI nodes
+ // as necessary.
+ SSAUpdater SSA;
+ for (I = OrigHeader->begin(); I != E; ++I) {
+ Value *OrigHeaderVal = I;
+
+ // If there are no uses of the value (e.g. because it returns void), there
+ // is nothing to rewrite.
+ if (OrigHeaderVal->use_empty())
+ continue;
+
+ Value *OrigPreHeaderVal = ValueMap[OrigHeaderVal];
+
+ // The value now exits in two versions: the initial value in the preheader
+ // and the loop "next" value in the original header.
+ SSA.Initialize(OrigHeaderVal->getType(), OrigHeaderVal->getName());
+ SSA.AddAvailableValue(OrigHeader, OrigHeaderVal);
+ SSA.AddAvailableValue(OrigPreheader, OrigPreHeaderVal);
+
+ // Visit each use of the OrigHeader instruction.
+ 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();
+
+ // Increment the iterator before removing the use from the list.
+ ++UI;
+
+ // SSAUpdater can't handle a non-PHI use in the same block as an
+ // earlier def. We can easily handle those cases manually.
+ Instruction *UserInst = cast<Instruction>(U.getUser());
+ if (!isa<PHINode>(UserInst)) {
+ BasicBlock *UserBB = UserInst->getParent();
+
+ // The original users in the OrigHeader are already using the
+ // original definitions.
+ if (UserBB == OrigHeader)
+ continue;
+
+ // Users in the OrigPreHeader need to use the value to which the
+ // original definitions are mapped.
+ if (UserBB == OrigPreheader) {
+ U = OrigPreHeaderVal;
+ continue;
+ }
+ }
+
+ // Anything else can be handled by SSAUpdater.
+ SSA.RewriteUse(U);
+ }
+ }
+}
+
+/// Determine whether the instructions in this range my be safely and cheaply
+/// speculated. This is not an important enough situation to develop complex
+/// heuristics. We handle a single arithmetic instruction along with any type
+/// conversions.
+static bool shouldSpeculateInstrs(BasicBlock::iterator Begin,
+ BasicBlock::iterator End) {
+ bool seenIncrement = false;
+ for (BasicBlock::iterator I = Begin; I != End; ++I) {
+
+ if (!isSafeToSpeculativelyExecute(I))
+ return false;
+
+ if (isa<DbgInfoIntrinsic>(I))
+ continue;
+
+ switch (I->getOpcode()) {
+ default:
+ return false;
+ case Instruction::GetElementPtr:
+ // GEPs are cheap if all indices are constant.
+ if (!cast<GEPOperator>(I)->hasAllConstantIndices())
+ return false;
+ // fall-thru to increment case
+ case Instruction::Add:
+ case Instruction::Sub:
+ case Instruction::And:
+ case Instruction::Or:
+ case Instruction::Xor:
+ case Instruction::Shl:
+ case Instruction::LShr:
+ case Instruction::AShr:
+ if (seenIncrement)
+ return false;
+ seenIncrement = true;
+ break;
+ case Instruction::Trunc:
+ case Instruction::ZExt:
+ case Instruction::SExt:
+ // ignore type conversions
+ break;
+ }
+ }
+ return true;
+}
+
+/// Fold the loop tail into the loop exit by speculating the loop tail
+/// instructions. Typically, this is a single post-increment. In the case of a
+/// simple 2-block loop, hoisting the increment can be much better than
+/// duplicating the entire loop header. In the cast of loops with early exits,
+/// rotation will not work anyway, but simplifyLoopLatch will put the loop in
+/// canonical form so downstream passes can handle it.
+///
+/// I don't believe this invalidates SCEV.
+void LoopRotate::simplifyLoopLatch(Loop *L) {
+ BasicBlock *Latch = L->getLoopLatch();
+ if (!Latch || Latch->hasAddressTaken())
+ return;
+
+ BranchInst *Jmp = dyn_cast<BranchInst>(Latch->getTerminator());
+ if (!Jmp || !Jmp->isUnconditional())
+ return;
+
+ BasicBlock *LastExit = Latch->getSinglePredecessor();
+ if (!LastExit || !L->isLoopExiting(LastExit))
+ return;
+
+ BranchInst *BI = dyn_cast<BranchInst>(LastExit->getTerminator());
+ if (!BI)
+ return;
+
+ if (!shouldSpeculateInstrs(Latch->begin(), Jmp))
+ return;
+
+ DEBUG(dbgs() << "Folding loop latch " << Latch->getName() << " into "
+ << LastExit->getName() << "\n");
+
+ // Hoist the instructions from Latch into LastExit.
+ LastExit->getInstList().splice(BI, Latch->getInstList(), Latch->begin(), Jmp);
+
+ unsigned FallThruPath = BI->getSuccessor(0) == Latch ? 0 : 1;
+ BasicBlock *Header = Jmp->getSuccessor(0);
+ assert(Header == L->getHeader() && "expected a backward branch");
+
+ // Remove Latch from the CFG so that LastExit becomes the new Latch.
+ BI->setSuccessor(FallThruPath, Header);
+ Latch->replaceSuccessorsPhiUsesWith(LastExit);
+ Jmp->eraseFromParent();
+
+ // Nuke the Latch block.
+ assert(Latch->empty() && "unable to evacuate Latch");
+ LI->removeBlock(Latch);
+ if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>())
+ DT->eraseNode(Latch);
+ Latch->eraseFromParent();
+}
+
/// Rotate loop LP. Return true if the loop is rotated.
bool LoopRotate::rotateLoop(Loop *L) {
// If the loop has only one block then there is not much to rotate.
if (L->getBlocks().size() == 1)
return false;
-
+
BasicBlock *OrigHeader = L->getHeader();
-
+
BranchInst *BI = dyn_cast<BranchInst>(OrigHeader->getTerminator());
if (BI == 0 || BI->isUnconditional())
return false;
-
+
// If the loop header is not one of the loop exiting blocks then
// either this loop is already rotated or it is not
// suitable for loop rotation transformations.
if (!L->isLoopExiting(OrigHeader))
return false;
- // Updating PHInodes in loops with multiple exits adds complexity.
+ // Updating PHInodes in loops with multiple exits adds complexity.
// Keep it simple, and restrict loop rotation to loops with one exit only.
// In future, lift this restriction and support for multiple exits if
// required.
}
// Now, this loop is suitable for rotation.
- BasicBlock *OrigPreHeader = L->getLoopPreheader();
+ BasicBlock *OrigPreheader = L->getLoopPreheader();
BasicBlock *OrigLatch = L->getLoopLatch();
- assert(OrigPreHeader && OrigLatch && "Loop not in canonical form?");
+
+ // If the loop could not be converted to canonical form, it must have an
+ // indirectbr in it, just give up.
+ if (OrigPreheader == 0 || OrigLatch == 0)
+ return false;
// Anything ScalarEvolution may know about this loop or the PHI nodes
// in its header will soon be invalidated.
if (L->contains(Exit))
std::swap(Exit, NewHeader);
assert(NewHeader && "Unable to determine new loop header");
- assert(L->contains(NewHeader) && !L->contains(Exit) &&
+ assert(L->contains(NewHeader) && !L->contains(Exit) &&
"Unable to determine loop header and exit blocks");
-
+
// This code assumes that the new header has exactly one predecessor.
// Remove any single-entry PHI nodes in it.
assert(NewHeader->getSinglePredecessor() &&
// For PHI nodes, the value available in OldPreHeader is just the
// incoming value from OldPreHeader.
for (; PHINode *PN = dyn_cast<PHINode>(I); ++I)
- ValueMap[PN] = PN->getIncomingValue(PN->getBasicBlockIndex(OrigPreHeader));
+ ValueMap[PN] = PN->getIncomingValueForBlock(OrigPreheader);
// For the rest of the instructions, either hoist to the OrigPreheader if
// possible or create a clone in the OldPreHeader if not.
- TerminatorInst *LoopEntryBranch = OrigPreHeader->getTerminator();
+ TerminatorInst *LoopEntryBranch = OrigPreheader->getTerminator();
while (I != E) {
Instruction *Inst = I++;
-
+
// If the instruction's operands are invariant and it doesn't read or write
// memory, then it is safe to hoist. Doing this doesn't change the order of
// execution in the preheader, but does prevent the instruction from
// memory (without proving that the loop doesn't write).
if (L->hasLoopInvariantOperands(Inst) &&
!Inst->mayReadFromMemory() && !Inst->mayWriteToMemory() &&
- !isa<TerminatorInst>(Inst)) {
+ !isa<TerminatorInst>(Inst) && !isa<DbgInfoIntrinsic>(Inst) &&
+ !isa<AllocaInst>(Inst)) {
Inst->moveBefore(LoopEntryBranch);
continue;
}
-
+
// Otherwise, create a duplicate of the instruction.
Instruction *C = Inst->clone();
-
+
// Eagerly remap the operands of the instruction.
RemapInstruction(C, ValueMap,
RF_NoModuleLevelChanges|RF_IgnoreMissingEntries);
-
+
// With the operands remapped, see if the instruction constant folds or is
// otherwise simplifyable. This commonly occurs because the entry from PHI
// nodes allows icmps and other instructions to fold.
for (unsigned i = 0, e = TI->getNumSuccessors(); i != e; ++i)
for (BasicBlock::iterator BI = TI->getSuccessor(i)->begin();
PHINode *PN = dyn_cast<PHINode>(BI); ++BI)
- PN->addIncoming(PN->getIncomingValueForBlock(OrigHeader), OrigPreHeader);
+ PN->addIncoming(PN->getIncomingValueForBlock(OrigHeader), OrigPreheader);
// Now that OrigPreHeader has a clone of OrigHeader's terminator, remove
// OrigPreHeader's old terminator (the original branch into the loop), and
// remove the corresponding incoming values from the PHI nodes in OrigHeader.
LoopEntryBranch->eraseFromParent();
- for (I = OrigHeader->begin(); PHINode *PN = dyn_cast<PHINode>(I); ++I)
- PN->removeIncomingValue(PN->getBasicBlockIndex(OrigPreHeader));
-
- // Now fix up users of the instructions in OrigHeader, inserting PHI nodes
- // as necessary.
- SSAUpdater SSA;
- for (I = OrigHeader->begin(); I != E; ++I) {
- Value *OrigHeaderVal = I;
- Value *OrigPreHeaderVal = ValueMap[OrigHeaderVal];
- // If there are no uses of the value (e.g. because it returns void), there
- // is nothing to rewrite.
- if (OrigHeaderVal->use_empty() && OrigPreHeaderVal->use_empty())
- continue;
-
- // The value now exits in two versions: the initial value in the preheader
- // and the loop "next" value in the original header.
- SSA.Initialize(OrigHeaderVal->getType(), OrigHeaderVal->getName());
- SSA.AddAvailableValue(OrigHeader, OrigHeaderVal);
- SSA.AddAvailableValue(OrigPreHeader, OrigPreHeaderVal);
-
- // Visit each use of the OrigHeader instruction.
- 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();
+ // If there were any uses of instructions in the duplicated block outside the
+ // loop, update them, inserting PHI nodes as required
+ RewriteUsesOfClonedInstructions(OrigHeader, OrigPreheader, ValueMap);
- // Increment the iterator before removing the use from the list.
- ++UI;
-
- // SSAUpdater can't handle a non-PHI use in the same block as an
- // earlier def. We can easily handle those cases manually.
- Instruction *UserInst = cast<Instruction>(U.getUser());
- if (!isa<PHINode>(UserInst)) {
- BasicBlock *UserBB = UserInst->getParent();
+ // NewHeader is now the header of the loop.
+ L->moveToHeader(NewHeader);
+ assert(L->getHeader() == NewHeader && "Latch block is our new header");
- // The original users in the OrigHeader are already using the
- // original definitions.
- if (UserBB == OrigHeader)
- continue;
- // Users in the OrigPreHeader need to use the value to which the
- // original definitions are mapped.
- if (UserBB == OrigPreHeader) {
- U = OrigPreHeaderVal;
- continue;
- }
- }
+ // At this point, we've finished our major CFG changes. As part of cloning
+ // the loop into the preheader we've simplified instructions and the
+ // duplicated conditional branch may now be branching on a constant. If it is
+ // branching on a constant and if that constant means that we enter the loop,
+ // then we fold away the cond branch to an uncond branch. This simplifies the
+ // loop in cases important for nested loops, and it also means we don't have
+ // to split as many edges.
+ BranchInst *PHBI = cast<BranchInst>(OrigPreheader->getTerminator());
+ assert(PHBI->isConditional() && "Should be clone of BI condbr!");
+ if (!isa<ConstantInt>(PHBI->getCondition()) ||
+ PHBI->getSuccessor(cast<ConstantInt>(PHBI->getCondition())->isZero())
+ != NewHeader) {
+ // 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>()) {
+ // Since OrigPreheader now has the conditional branch to Exit block, it is
+ // the dominator of Exit.
+ DT->changeImmediateDominator(Exit, OrigPreheader);
+ DT->changeImmediateDominator(NewHeader, OrigPreheader);
+
+ // Update OrigHeader to be dominated by the new header block.
+ DT->changeImmediateDominator(OrigHeader, OrigLatch);
+ }
- // Anything else can be handled by SSAUpdater.
- SSA.RewriteUse(U);
+ // Right now OrigPreHeader has two successors, NewHeader and ExitBlock, and
+ // thus is not a preheader anymore.
+ // Split the edge to form a real preheader.
+ BasicBlock *NewPH = SplitCriticalEdge(OrigPreheader, NewHeader, this);
+ 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);
+ } 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.
+ Exit->removePredecessor(OrigPreheader, true /*preserve LCSSA*/);
+ BranchInst *NewBI = BranchInst::Create(NewHeader, PHBI);
+ NewBI->setDebugLoc(PHBI->getDebugLoc());
+ PHBI->eraseFromParent();
+
+ // With our CFG finalized, update DomTree if it is available.
+ if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>()) {
+ // Update OrigHeader to be dominated by the new header block.
+ DT->changeImmediateDominator(NewHeader, OrigPreheader);
+ DT->changeImmediateDominator(OrigHeader, OrigLatch);
}
}
- // NewHeader is now the header of the loop.
- L->moveToHeader(NewHeader);
+ assert(L->getLoopPreheader() && "Invalid loop preheader after loop rotation");
+ assert(L->getLoopLatch() && "Invalid loop latch after loop rotation");
- // Move the original header to the bottom of the loop, where it now more
- // naturally belongs. This isn't necessary for correctness, and CodeGen can
- // usually reorder blocks on its own to fix things like this up, but it's
- // still nice to keep the IR readable.
- //
- // The original header should have only one predecessor at this point, since
- // we checked that the loop had a proper preheader and unique backedge before
- // we started.
- assert(OrigHeader->getSinglePredecessor() &&
- "Original loop header has too many predecessors after loop rotation!");
- OrigHeader->moveAfter(OrigHeader->getSinglePredecessor());
-
- // Also, since this original header only has one predecessor, zap its
- // PHI nodes, which are now trivial.
- FoldSingleEntryPHINodes(OrigHeader);
-
- // TODO: We could just go ahead and merge OrigHeader into its predecessor
- // at this point, if we don't mind updating dominator info.
-
- // Establish a new preheader, update dominators, etc.
- preserveCanonicalLoopForm(L, OrigHeader, OrigPreHeader, OrigLatch,
- NewHeader, Exit);
+ // Now that the CFG and DomTree are in a consistent state again, try to merge
+ // the OrigHeader block into OrigLatch. This will succeed if they are
+ // connected by an unconditional branch. This is just a cleanup so the
+ // emitted code isn't too gross in this common case.
+ MergeBlockIntoPredecessor(OrigHeader, this);
++NumRotated;
return true;
}
-
-/// Update LoopInfo, DominatorTree, and DomFrontiers to reflect the CFG change
-/// we just made. Then split edges as necessary to preserve LoopSimplify form.
-void LoopRotate::preserveCanonicalLoopForm(Loop *L, BasicBlock *OrigHeader,
- BasicBlock *OrigPreHeader,
- BasicBlock *OrigLatch,
- BasicBlock *NewHeader,
- BasicBlock *Exit) {
- assert(L->getHeader() == NewHeader && "Latch block is our new header");
-
- if (DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>()) {
- // Since OrigPreheader now has the conditional branch to Exit block, it is
- // the dominator of Exit.
- DT->changeImmediateDominator(Exit, OrigPreHeader);
- DT->changeImmediateDominator(NewHeader, OrigPreHeader);
-
- // Update OrigHeader to be dominated by the new header block.
- DT->changeImmediateDominator(OrigHeader, OrigLatch);
- }
-
- // Right now OrigPreHeader has two successors, NewHeader and ExitBlock, and
- // thus is not a preheader anymore. Split the edge to form a real preheader.
- BasicBlock *NewPH = SplitCriticalEdge(OrigPreHeader, NewHeader, this);
- NewPH->setName(NewHeader->getName() + ".lr.ph");
-
- // Preserve canonical loop form, which means Exit block should have only one
- // predecessor.
- SplitCriticalEdge(L->getLoopLatch(), Exit, this);
-
- assert(NewHeader && L->getHeader() == NewHeader &&
- "Invalid loop header after loop rotation");
- assert(L->getLoopPreheader() == NewPH &&
- "Invalid loop preheader after loop rotation");
- assert(L->getLoopLatch() && "Invalid loop latch after loop rotation");
-}
projects / oota-llvm.git / blobdiff