1
//===- JumpThreading.cpp - Thread control through conditional blocks ------===//
3
// The LLVM Compiler Infrastructure
5
// This file is distributed under the University of Illinois Open Source
6
// License. See LICENSE.TXT for details.
8
//===----------------------------------------------------------------------===//
10
// This file implements the Jump Threading pass.
12
//===----------------------------------------------------------------------===//
14
#define DEBUG_TYPE "jump-threading"
15
#include "llvm/Transforms/Scalar.h"
16
#include "llvm/IntrinsicInst.h"
17
#include "llvm/LLVMContext.h"
18
#include "llvm/Pass.h"
19
#include "llvm/Analysis/InstructionSimplify.h"
20
#include "llvm/Analysis/LazyValueInfo.h"
21
#include "llvm/Analysis/Loads.h"
22
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
23
#include "llvm/Transforms/Utils/Local.h"
24
#include "llvm/Transforms/Utils/SSAUpdater.h"
25
#include "llvm/Target/TargetData.h"
26
#include "llvm/ADT/DenseMap.h"
27
#include "llvm/ADT/DenseSet.h"
28
#include "llvm/ADT/Statistic.h"
29
#include "llvm/ADT/STLExtras.h"
30
#include "llvm/ADT/SmallPtrSet.h"
31
#include "llvm/ADT/SmallSet.h"
32
#include "llvm/Support/CommandLine.h"
33
#include "llvm/Support/Debug.h"
34
#include "llvm/Support/ValueHandle.h"
35
#include "llvm/Support/raw_ostream.h"
38
STATISTIC(NumThreads, "Number of jumps threaded");
39
STATISTIC(NumFolds, "Number of terminators folded");
40
STATISTIC(NumDupes, "Number of branch blocks duplicated to eliminate phi");
42
static cl::opt<unsigned>
43
Threshold("jump-threading-threshold",
44
cl::desc("Max block size to duplicate for jump threading"),
45
cl::init(6), cl::Hidden);
47
// Turn on use of LazyValueInfo.
49
EnableLVI("enable-jump-threading-lvi",
50
cl::desc("Use LVI for jump threading"),
57
/// This pass performs 'jump threading', which looks at blocks that have
58
/// multiple predecessors and multiple successors. If one or more of the
59
/// predecessors of the block can be proven to always jump to one of the
60
/// successors, we forward the edge from the predecessor to the successor by
61
/// duplicating the contents of this block.
63
/// An example of when this can occur is code like this:
70
/// In this case, the unconditional branch at the end of the first if can be
71
/// revectored to the false side of the second if.
73
class JumpThreading : public FunctionPass {
77
SmallPtrSet<BasicBlock*, 16> LoopHeaders;
79
SmallSet<AssertingVH<BasicBlock>, 16> LoopHeaders;
81
DenseSet<std::pair<Value*, BasicBlock*> > RecursionSet;
83
// RAII helper for updating the recursion stack.
84
struct RecursionSetRemover {
85
DenseSet<std::pair<Value*, BasicBlock*> > &TheSet;
86
std::pair<Value*, BasicBlock*> ThePair;
88
RecursionSetRemover(DenseSet<std::pair<Value*, BasicBlock*> > &S,
89
std::pair<Value*, BasicBlock*> P)
90
: TheSet(S), ThePair(P) { }
92
~RecursionSetRemover() {
93
TheSet.erase(ThePair);
97
static char ID; // Pass identification
98
JumpThreading() : FunctionPass(ID) {}
100
bool runOnFunction(Function &F);
102
virtual void getAnalysisUsage(AnalysisUsage &AU) const {
104
AU.addRequired<LazyValueInfo>();
105
AU.addPreserved<LazyValueInfo>();
109
void FindLoopHeaders(Function &F);
110
bool ProcessBlock(BasicBlock *BB);
111
bool ThreadEdge(BasicBlock *BB, const SmallVectorImpl<BasicBlock*> &PredBBs,
113
bool DuplicateCondBranchOnPHIIntoPred(BasicBlock *BB,
114
const SmallVectorImpl<BasicBlock *> &PredBBs);
116
typedef SmallVectorImpl<std::pair<ConstantInt*,
117
BasicBlock*> > PredValueInfo;
119
bool ComputeValueKnownInPredecessors(Value *V, BasicBlock *BB,
120
PredValueInfo &Result);
121
bool ProcessThreadableEdges(Value *Cond, BasicBlock *BB);
124
bool ProcessBranchOnDuplicateCond(BasicBlock *PredBB, BasicBlock *DestBB);
125
bool ProcessSwitchOnDuplicateCond(BasicBlock *PredBB, BasicBlock *DestBB);
127
bool ProcessBranchOnPHI(PHINode *PN);
128
bool ProcessBranchOnXOR(BinaryOperator *BO);
130
bool SimplifyPartiallyRedundantLoad(LoadInst *LI);
134
char JumpThreading::ID = 0;
135
INITIALIZE_PASS(JumpThreading, "jump-threading",
136
"Jump Threading", false, false);
138
// Public interface to the Jump Threading pass
139
FunctionPass *llvm::createJumpThreadingPass() { return new JumpThreading(); }
141
/// runOnFunction - Top level algorithm.
143
bool JumpThreading::runOnFunction(Function &F) {
144
DEBUG(dbgs() << "Jump threading on function '" << F.getName() << "'\n");
145
TD = getAnalysisIfAvailable<TargetData>();
146
LVI = EnableLVI ? &getAnalysis<LazyValueInfo>() : 0;
150
bool Changed, EverChanged = false;
153
for (Function::iterator I = F.begin(), E = F.end(); I != E;) {
155
// Thread all of the branches we can over this block.
156
while (ProcessBlock(BB))
161
// If the block is trivially dead, zap it. This eliminates the successor
162
// edges which simplifies the CFG.
163
if (pred_begin(BB) == pred_end(BB) &&
164
BB != &BB->getParent()->getEntryBlock()) {
165
DEBUG(dbgs() << " JT: Deleting dead block '" << BB->getName()
166
<< "' with terminator: " << *BB->getTerminator() << '\n');
167
LoopHeaders.erase(BB);
168
if (LVI) LVI->eraseBlock(BB);
171
} else if (BranchInst *BI = dyn_cast<BranchInst>(BB->getTerminator())) {
172
// Can't thread an unconditional jump, but if the block is "almost
173
// empty", we can replace uses of it with uses of the successor and make
175
if (BI->isUnconditional() &&
176
BB != &BB->getParent()->getEntryBlock()) {
177
BasicBlock::iterator BBI = BB->getFirstNonPHI();
178
// Ignore dbg intrinsics.
179
while (isa<DbgInfoIntrinsic>(BBI))
181
// If the terminator is the only non-phi instruction, try to nuke it.
182
if (BBI->isTerminator()) {
183
// Since TryToSimplifyUncondBranchFromEmptyBlock may delete the
184
// block, we have to make sure it isn't in the LoopHeaders set. We
185
// reinsert afterward if needed.
186
bool ErasedFromLoopHeaders = LoopHeaders.erase(BB);
187
BasicBlock *Succ = BI->getSuccessor(0);
189
// FIXME: It is always conservatively correct to drop the info
190
// for a block even if it doesn't get erased. This isn't totally
191
// awesome, but it allows us to use AssertingVH to prevent nasty
192
// dangling pointer issues within LazyValueInfo.
193
if (LVI) LVI->eraseBlock(BB);
194
if (TryToSimplifyUncondBranchFromEmptyBlock(BB)) {
196
// If we deleted BB and BB was the header of a loop, then the
197
// successor is now the header of the loop.
201
if (ErasedFromLoopHeaders)
202
LoopHeaders.insert(BB);
207
EverChanged |= Changed;
214
/// getJumpThreadDuplicationCost - Return the cost of duplicating this block to
215
/// thread across it.
216
static unsigned getJumpThreadDuplicationCost(const BasicBlock *BB) {
217
/// Ignore PHI nodes, these will be flattened when duplication happens.
218
BasicBlock::const_iterator I = BB->getFirstNonPHI();
220
// FIXME: THREADING will delete values that are just used to compute the
221
// branch, so they shouldn't count against the duplication cost.
224
// Sum up the cost of each instruction until we get to the terminator. Don't
225
// include the terminator because the copy won't include it.
227
for (; !isa<TerminatorInst>(I); ++I) {
228
// Debugger intrinsics don't incur code size.
229
if (isa<DbgInfoIntrinsic>(I)) continue;
231
// If this is a pointer->pointer bitcast, it is free.
232
if (isa<BitCastInst>(I) && I->getType()->isPointerTy())
235
// All other instructions count for at least one unit.
238
// Calls are more expensive. If they are non-intrinsic calls, we model them
239
// as having cost of 4. If they are a non-vector intrinsic, we model them
240
// as having cost of 2 total, and if they are a vector intrinsic, we model
241
// them as having cost 1.
242
if (const CallInst *CI = dyn_cast<CallInst>(I)) {
243
if (!isa<IntrinsicInst>(CI))
245
else if (!CI->getType()->isVectorTy())
250
// Threading through a switch statement is particularly profitable. If this
251
// block ends in a switch, decrease its cost to make it more likely to happen.
252
if (isa<SwitchInst>(I))
253
Size = Size > 6 ? Size-6 : 0;
258
/// FindLoopHeaders - We do not want jump threading to turn proper loop
259
/// structures into irreducible loops. Doing this breaks up the loop nesting
260
/// hierarchy and pessimizes later transformations. To prevent this from
261
/// happening, we first have to find the loop headers. Here we approximate this
262
/// by finding targets of backedges in the CFG.
264
/// Note that there definitely are cases when we want to allow threading of
265
/// edges across a loop header. For example, threading a jump from outside the
266
/// loop (the preheader) to an exit block of the loop is definitely profitable.
267
/// It is also almost always profitable to thread backedges from within the loop
268
/// to exit blocks, and is often profitable to thread backedges to other blocks
269
/// within the loop (forming a nested loop). This simple analysis is not rich
270
/// enough to track all of these properties and keep it up-to-date as the CFG
271
/// mutates, so we don't allow any of these transformations.
273
void JumpThreading::FindLoopHeaders(Function &F) {
274
SmallVector<std::pair<const BasicBlock*,const BasicBlock*>, 32> Edges;
275
FindFunctionBackedges(F, Edges);
277
for (unsigned i = 0, e = Edges.size(); i != e; ++i)
278
LoopHeaders.insert(const_cast<BasicBlock*>(Edges[i].second));
281
// Helper method for ComputeValueKnownInPredecessors. If Value is a
282
// ConstantInt, push it. If it's an undef, push 0. Otherwise, do nothing.
283
static void PushConstantIntOrUndef(SmallVectorImpl<std::pair<ConstantInt*,
284
BasicBlock*> > &Result,
285
Constant *Value, BasicBlock* BB){
286
if (ConstantInt *FoldedCInt = dyn_cast<ConstantInt>(Value))
287
Result.push_back(std::make_pair(FoldedCInt, BB));
288
else if (isa<UndefValue>(Value))
289
Result.push_back(std::make_pair((ConstantInt*)0, BB));
292
/// ComputeValueKnownInPredecessors - Given a basic block BB and a value V, see
293
/// if we can infer that the value is a known ConstantInt in any of our
294
/// predecessors. If so, return the known list of value and pred BB in the
295
/// result vector. If a value is known to be undef, it is returned as null.
297
/// This returns true if there were any known values.
300
ComputeValueKnownInPredecessors(Value *V, BasicBlock *BB,PredValueInfo &Result){
301
// This method walks up use-def chains recursively. Because of this, we could
302
// get into an infinite loop going around loops in the use-def chain. To
303
// prevent this, keep track of what (value, block) pairs we've already visited
304
// and terminate the search if we loop back to them
305
if (!RecursionSet.insert(std::make_pair(V, BB)).second)
308
// An RAII help to remove this pair from the recursion set once the recursion
309
// stack pops back out again.
310
RecursionSetRemover remover(RecursionSet, std::make_pair(V, BB));
312
// If V is a constantint, then it is known in all predecessors.
313
if (isa<ConstantInt>(V) || isa<UndefValue>(V)) {
314
ConstantInt *CI = dyn_cast<ConstantInt>(V);
316
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI)
317
Result.push_back(std::make_pair(CI, *PI));
322
// If V is a non-instruction value, or an instruction in a different block,
323
// then it can't be derived from a PHI.
324
Instruction *I = dyn_cast<Instruction>(V);
325
if (I == 0 || I->getParent() != BB) {
327
// Okay, if this is a live-in value, see if it has a known value at the end
328
// of any of our predecessors.
330
// FIXME: This should be an edge property, not a block end property.
331
/// TODO: Per PR2563, we could infer value range information about a
332
/// predecessor based on its terminator.
335
// FIXME: change this to use the more-rich 'getPredicateOnEdge' method if
336
// "I" is a non-local compare-with-a-constant instruction. This would be
337
// able to handle value inequalities better, for example if the compare is
338
// "X < 4" and "X < 3" is known true but "X < 4" itself is not available.
339
// Perhaps getConstantOnEdge should be smart enough to do this?
341
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
343
// If the value is known by LazyValueInfo to be a constant in a
344
// predecessor, use that information to try to thread this block.
345
Constant *PredCst = LVI->getConstantOnEdge(V, P, BB);
347
(!isa<ConstantInt>(PredCst) && !isa<UndefValue>(PredCst)))
350
Result.push_back(std::make_pair(dyn_cast<ConstantInt>(PredCst), P));
353
return !Result.empty();
359
/// If I is a PHI node, then we know the incoming values for any constants.
360
if (PHINode *PN = dyn_cast<PHINode>(I)) {
361
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
362
Value *InVal = PN->getIncomingValue(i);
363
if (isa<ConstantInt>(InVal) || isa<UndefValue>(InVal)) {
364
ConstantInt *CI = dyn_cast<ConstantInt>(InVal);
365
Result.push_back(std::make_pair(CI, PN->getIncomingBlock(i)));
367
Constant *CI = LVI->getConstantOnEdge(InVal,
368
PN->getIncomingBlock(i), BB);
369
// LVI returns null is no value could be determined.
371
PushConstantIntOrUndef(Result, CI, PN->getIncomingBlock(i));
375
return !Result.empty();
378
SmallVector<std::pair<ConstantInt*, BasicBlock*>, 8> LHSVals, RHSVals;
380
// Handle some boolean conditions.
381
if (I->getType()->getPrimitiveSizeInBits() == 1) {
383
// X & false -> false
384
if (I->getOpcode() == Instruction::Or ||
385
I->getOpcode() == Instruction::And) {
386
ComputeValueKnownInPredecessors(I->getOperand(0), BB, LHSVals);
387
ComputeValueKnownInPredecessors(I->getOperand(1), BB, RHSVals);
389
if (LHSVals.empty() && RHSVals.empty())
392
ConstantInt *InterestingVal;
393
if (I->getOpcode() == Instruction::Or)
394
InterestingVal = ConstantInt::getTrue(I->getContext());
396
InterestingVal = ConstantInt::getFalse(I->getContext());
398
SmallPtrSet<BasicBlock*, 4> LHSKnownBBs;
400
// Scan for the sentinel. If we find an undef, force it to the
401
// interesting value: x|undef -> true and x&undef -> false.
402
for (unsigned i = 0, e = LHSVals.size(); i != e; ++i)
403
if (LHSVals[i].first == InterestingVal || LHSVals[i].first == 0) {
404
Result.push_back(LHSVals[i]);
405
Result.back().first = InterestingVal;
406
LHSKnownBBs.insert(LHSVals[i].second);
408
for (unsigned i = 0, e = RHSVals.size(); i != e; ++i)
409
if (RHSVals[i].first == InterestingVal || RHSVals[i].first == 0) {
410
// If we already inferred a value for this block on the LHS, don't
412
if (!LHSKnownBBs.count(RHSVals[i].second)) {
413
Result.push_back(RHSVals[i]);
414
Result.back().first = InterestingVal;
418
return !Result.empty();
421
// Handle the NOT form of XOR.
422
if (I->getOpcode() == Instruction::Xor &&
423
isa<ConstantInt>(I->getOperand(1)) &&
424
cast<ConstantInt>(I->getOperand(1))->isOne()) {
425
ComputeValueKnownInPredecessors(I->getOperand(0), BB, Result);
429
// Invert the known values.
430
for (unsigned i = 0, e = Result.size(); i != e; ++i)
433
cast<ConstantInt>(ConstantExpr::getNot(Result[i].first));
438
// Try to simplify some other binary operator values.
439
} else if (BinaryOperator *BO = dyn_cast<BinaryOperator>(I)) {
440
if (ConstantInt *CI = dyn_cast<ConstantInt>(BO->getOperand(1))) {
441
SmallVector<std::pair<ConstantInt*, BasicBlock*>, 8> LHSVals;
442
ComputeValueKnownInPredecessors(BO->getOperand(0), BB, LHSVals);
444
// Try to use constant folding to simplify the binary operator.
445
for (unsigned i = 0, e = LHSVals.size(); i != e; ++i) {
446
Constant *V = LHSVals[i].first ? LHSVals[i].first :
447
cast<Constant>(UndefValue::get(BO->getType()));
448
Constant *Folded = ConstantExpr::get(BO->getOpcode(), V, CI);
450
PushConstantIntOrUndef(Result, Folded, LHSVals[i].second);
454
return !Result.empty();
457
// Handle compare with phi operand, where the PHI is defined in this block.
458
if (CmpInst *Cmp = dyn_cast<CmpInst>(I)) {
459
PHINode *PN = dyn_cast<PHINode>(Cmp->getOperand(0));
460
if (PN && PN->getParent() == BB) {
461
// We can do this simplification if any comparisons fold to true or false.
463
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
464
BasicBlock *PredBB = PN->getIncomingBlock(i);
465
Value *LHS = PN->getIncomingValue(i);
466
Value *RHS = Cmp->getOperand(1)->DoPHITranslation(BB, PredBB);
468
Value *Res = SimplifyCmpInst(Cmp->getPredicate(), LHS, RHS, TD);
470
if (!LVI || !isa<Constant>(RHS))
473
LazyValueInfo::Tristate
474
ResT = LVI->getPredicateOnEdge(Cmp->getPredicate(), LHS,
475
cast<Constant>(RHS), PredBB, BB);
476
if (ResT == LazyValueInfo::Unknown)
478
Res = ConstantInt::get(Type::getInt1Ty(LHS->getContext()), ResT);
481
if (Constant *ConstRes = dyn_cast<Constant>(Res))
482
PushConstantIntOrUndef(Result, ConstRes, PredBB);
485
return !Result.empty();
489
// If comparing a live-in value against a constant, see if we know the
490
// live-in value on any predecessors.
491
if (LVI && isa<Constant>(Cmp->getOperand(1)) &&
492
Cmp->getType()->isIntegerTy()) {
493
if (!isa<Instruction>(Cmp->getOperand(0)) ||
494
cast<Instruction>(Cmp->getOperand(0))->getParent() != BB) {
495
Constant *RHSCst = cast<Constant>(Cmp->getOperand(1));
497
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB);PI != E; ++PI){
499
// If the value is known by LazyValueInfo to be a constant in a
500
// predecessor, use that information to try to thread this block.
501
LazyValueInfo::Tristate Res =
502
LVI->getPredicateOnEdge(Cmp->getPredicate(), Cmp->getOperand(0),
504
if (Res == LazyValueInfo::Unknown)
507
Constant *ResC = ConstantInt::get(Cmp->getType(), Res);
508
Result.push_back(std::make_pair(cast<ConstantInt>(ResC), P));
511
return !Result.empty();
514
// Try to find a constant value for the LHS of a comparison,
515
// and evaluate it statically if we can.
516
if (Constant *CmpConst = dyn_cast<Constant>(Cmp->getOperand(1))) {
517
SmallVector<std::pair<ConstantInt*, BasicBlock*>, 8> LHSVals;
518
ComputeValueKnownInPredecessors(I->getOperand(0), BB, LHSVals);
520
for (unsigned i = 0, e = LHSVals.size(); i != e; ++i) {
521
Constant *V = LHSVals[i].first ? LHSVals[i].first :
522
cast<Constant>(UndefValue::get(CmpConst->getType()));
523
Constant *Folded = ConstantExpr::getCompare(Cmp->getPredicate(),
525
PushConstantIntOrUndef(Result, Folded, LHSVals[i].second);
528
return !Result.empty();
534
// If all else fails, see if LVI can figure out a constant value for us.
535
Constant *CI = LVI->getConstant(V, BB);
536
ConstantInt *CInt = dyn_cast_or_null<ConstantInt>(CI);
538
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI)
539
Result.push_back(std::make_pair(CInt, *PI));
542
return !Result.empty();
550
/// GetBestDestForBranchOnUndef - If we determine that the specified block ends
551
/// in an undefined jump, decide which block is best to revector to.
553
/// Since we can pick an arbitrary destination, we pick the successor with the
554
/// fewest predecessors. This should reduce the in-degree of the others.
556
static unsigned GetBestDestForJumpOnUndef(BasicBlock *BB) {
557
TerminatorInst *BBTerm = BB->getTerminator();
558
unsigned MinSucc = 0;
559
BasicBlock *TestBB = BBTerm->getSuccessor(MinSucc);
560
// Compute the successor with the minimum number of predecessors.
561
unsigned MinNumPreds = std::distance(pred_begin(TestBB), pred_end(TestBB));
562
for (unsigned i = 1, e = BBTerm->getNumSuccessors(); i != e; ++i) {
563
TestBB = BBTerm->getSuccessor(i);
564
unsigned NumPreds = std::distance(pred_begin(TestBB), pred_end(TestBB));
565
if (NumPreds < MinNumPreds)
572
/// ProcessBlock - If there are any predecessors whose control can be threaded
573
/// through to a successor, transform them now.
574
bool JumpThreading::ProcessBlock(BasicBlock *BB) {
575
// If the block is trivially dead, just return and let the caller nuke it.
576
// This simplifies other transformations.
577
if (pred_begin(BB) == pred_end(BB) &&
578
BB != &BB->getParent()->getEntryBlock())
581
// If this block has a single predecessor, and if that pred has a single
582
// successor, merge the blocks. This encourages recursive jump threading
583
// because now the condition in this block can be threaded through
584
// predecessors of our predecessor block.
585
if (BasicBlock *SinglePred = BB->getSinglePredecessor()) {
586
if (SinglePred->getTerminator()->getNumSuccessors() == 1 &&
588
// If SinglePred was a loop header, BB becomes one.
589
if (LoopHeaders.erase(SinglePred))
590
LoopHeaders.insert(BB);
592
// Remember if SinglePred was the entry block of the function. If so, we
593
// will need to move BB back to the entry position.
594
bool isEntry = SinglePred == &SinglePred->getParent()->getEntryBlock();
595
if (LVI) LVI->eraseBlock(SinglePred);
596
MergeBasicBlockIntoOnlyPred(BB);
598
if (isEntry && BB != &BB->getParent()->getEntryBlock())
599
BB->moveBefore(&BB->getParent()->getEntryBlock());
604
// Look to see if the terminator is a branch of switch, if not we can't thread
607
if (BranchInst *BI = dyn_cast<BranchInst>(BB->getTerminator())) {
608
// Can't thread an unconditional jump.
609
if (BI->isUnconditional()) return false;
610
Condition = BI->getCondition();
611
} else if (SwitchInst *SI = dyn_cast<SwitchInst>(BB->getTerminator()))
612
Condition = SI->getCondition();
614
return false; // Must be an invoke.
616
// If the terminator of this block is branching on a constant, simplify the
617
// terminator to an unconditional branch. This can occur due to threading in
619
if (isa<ConstantInt>(Condition)) {
620
DEBUG(dbgs() << " In block '" << BB->getName()
621
<< "' folding terminator: " << *BB->getTerminator() << '\n');
623
ConstantFoldTerminator(BB);
627
// If the terminator is branching on an undef, we can pick any of the
628
// successors to branch to. Let GetBestDestForJumpOnUndef decide.
629
if (isa<UndefValue>(Condition)) {
630
unsigned BestSucc = GetBestDestForJumpOnUndef(BB);
632
// Fold the branch/switch.
633
TerminatorInst *BBTerm = BB->getTerminator();
634
for (unsigned i = 0, e = BBTerm->getNumSuccessors(); i != e; ++i) {
635
if (i == BestSucc) continue;
636
RemovePredecessorAndSimplify(BBTerm->getSuccessor(i), BB, TD);
639
DEBUG(dbgs() << " In block '" << BB->getName()
640
<< "' folding undef terminator: " << *BBTerm << '\n');
641
BranchInst::Create(BBTerm->getSuccessor(BestSucc), BBTerm);
642
BBTerm->eraseFromParent();
646
Instruction *CondInst = dyn_cast<Instruction>(Condition);
648
// If the condition is an instruction defined in another block, see if a
649
// predecessor has the same condition:
654
!Condition->hasOneUse() && // Multiple uses.
655
(CondInst == 0 || CondInst->getParent() != BB)) { // Non-local definition.
656
pred_iterator PI = pred_begin(BB), E = pred_end(BB);
657
if (isa<BranchInst>(BB->getTerminator())) {
658
for (; PI != E; ++PI) {
660
if (BranchInst *PBI = dyn_cast<BranchInst>(P->getTerminator()))
661
if (PBI->isConditional() && PBI->getCondition() == Condition &&
662
ProcessBranchOnDuplicateCond(P, BB))
666
assert(isa<SwitchInst>(BB->getTerminator()) && "Unknown jump terminator");
667
for (; PI != E; ++PI) {
669
if (SwitchInst *PSI = dyn_cast<SwitchInst>(P->getTerminator()))
670
if (PSI->getCondition() == Condition &&
671
ProcessSwitchOnDuplicateCond(P, BB))
677
// All the rest of our checks depend on the condition being an instruction.
679
// FIXME: Unify this with code below.
680
if (LVI && ProcessThreadableEdges(Condition, BB))
686
if (CmpInst *CondCmp = dyn_cast<CmpInst>(CondInst)) {
688
(!isa<PHINode>(CondCmp->getOperand(0)) ||
689
cast<PHINode>(CondCmp->getOperand(0))->getParent() != BB)) {
690
// If we have a comparison, loop over the predecessors to see if there is
691
// a condition with a lexically identical value.
692
pred_iterator PI = pred_begin(BB), E = pred_end(BB);
693
for (; PI != E; ++PI) {
695
if (BranchInst *PBI = dyn_cast<BranchInst>(P->getTerminator()))
696
if (PBI->isConditional() && P != BB) {
697
if (CmpInst *CI = dyn_cast<CmpInst>(PBI->getCondition())) {
698
if (CI->getOperand(0) == CondCmp->getOperand(0) &&
699
CI->getOperand(1) == CondCmp->getOperand(1) &&
700
CI->getPredicate() == CondCmp->getPredicate()) {
701
// TODO: Could handle things like (x != 4) --> (x == 17)
702
if (ProcessBranchOnDuplicateCond(P, BB))
710
// For a comparison where the LHS is outside this block, it's possible
711
// that we've branched on it before. Used LVI to see if we can simplify
712
// the branch based on that.
713
BranchInst *CondBr = dyn_cast<BranchInst>(BB->getTerminator());
714
Constant *CondConst = dyn_cast<Constant>(CondCmp->getOperand(1));
715
pred_iterator PI = pred_begin(BB), PE = pred_end(BB);
716
if (LVI && CondBr && CondConst && CondBr->isConditional() && PI != PE &&
717
(!isa<Instruction>(CondCmp->getOperand(0)) ||
718
cast<Instruction>(CondCmp->getOperand(0))->getParent() != BB)) {
719
// For predecessor edge, determine if the comparison is true or false
720
// on that edge. If they're all true or all false, we can simplify the
722
// FIXME: We could handle mixed true/false by duplicating code.
723
LazyValueInfo::Tristate Baseline =
724
LVI->getPredicateOnEdge(CondCmp->getPredicate(), CondCmp->getOperand(0),
726
if (Baseline != LazyValueInfo::Unknown) {
727
// Check that all remaining incoming values match the first one.
729
LazyValueInfo::Tristate Ret = LVI->getPredicateOnEdge(
730
CondCmp->getPredicate(),
731
CondCmp->getOperand(0),
733
if (Ret != Baseline) break;
736
// If we terminated early, then one of the values didn't match.
738
unsigned ToRemove = Baseline == LazyValueInfo::True ? 1 : 0;
739
unsigned ToKeep = Baseline == LazyValueInfo::True ? 0 : 1;
740
RemovePredecessorAndSimplify(CondBr->getSuccessor(ToRemove), BB, TD);
741
BranchInst::Create(CondBr->getSuccessor(ToKeep), CondBr);
742
CondBr->eraseFromParent();
749
// Check for some cases that are worth simplifying. Right now we want to look
750
// for loads that are used by a switch or by the condition for the branch. If
751
// we see one, check to see if it's partially redundant. If so, insert a PHI
752
// which can then be used to thread the values.
754
Value *SimplifyValue = CondInst;
755
if (CmpInst *CondCmp = dyn_cast<CmpInst>(SimplifyValue))
756
if (isa<Constant>(CondCmp->getOperand(1)))
757
SimplifyValue = CondCmp->getOperand(0);
759
// TODO: There are other places where load PRE would be profitable, such as
760
// more complex comparisons.
761
if (LoadInst *LI = dyn_cast<LoadInst>(SimplifyValue))
762
if (SimplifyPartiallyRedundantLoad(LI))
766
// Handle a variety of cases where we are branching on something derived from
767
// a PHI node in the current block. If we can prove that any predecessors
768
// compute a predictable value based on a PHI node, thread those predecessors.
770
if (ProcessThreadableEdges(CondInst, BB))
773
// If this is an otherwise-unfoldable branch on a phi node in the current
774
// block, see if we can simplify.
775
if (PHINode *PN = dyn_cast<PHINode>(CondInst))
776
if (PN->getParent() == BB && isa<BranchInst>(BB->getTerminator()))
777
return ProcessBranchOnPHI(PN);
780
// If this is an otherwise-unfoldable branch on a XOR, see if we can simplify.
781
if (CondInst->getOpcode() == Instruction::Xor &&
782
CondInst->getParent() == BB && isa<BranchInst>(BB->getTerminator()))
783
return ProcessBranchOnXOR(cast<BinaryOperator>(CondInst));
786
// TODO: If we have: "br (X > 0)" and we have a predecessor where we know
787
// "(X == 4)", thread through this block.
792
/// ProcessBranchOnDuplicateCond - We found a block and a predecessor of that
793
/// block that jump on exactly the same condition. This means that we almost
794
/// always know the direction of the edge in the DESTBB:
796
/// br COND, DESTBB, BBY
798
/// br COND, BBZ, BBW
800
/// If DESTBB has multiple predecessors, we can't just constant fold the branch
801
/// in DESTBB, we have to thread over it.
802
bool JumpThreading::ProcessBranchOnDuplicateCond(BasicBlock *PredBB,
804
BranchInst *PredBI = cast<BranchInst>(PredBB->getTerminator());
806
// If both successors of PredBB go to DESTBB, we don't know anything. We can
807
// fold the branch to an unconditional one, which allows other recursive
810
if (PredBI->getSuccessor(1) != BB)
812
else if (PredBI->getSuccessor(0) != BB)
815
DEBUG(dbgs() << " In block '" << PredBB->getName()
816
<< "' folding terminator: " << *PredBB->getTerminator() << '\n');
818
ConstantFoldTerminator(PredBB);
822
BranchInst *DestBI = cast<BranchInst>(BB->getTerminator());
824
// If the dest block has one predecessor, just fix the branch condition to a
825
// constant and fold it.
826
if (BB->getSinglePredecessor()) {
827
DEBUG(dbgs() << " In block '" << BB->getName()
828
<< "' folding condition to '" << BranchDir << "': "
829
<< *BB->getTerminator() << '\n');
831
Value *OldCond = DestBI->getCondition();
832
DestBI->setCondition(ConstantInt::get(Type::getInt1Ty(BB->getContext()),
834
// Delete dead instructions before we fold the branch. Folding the branch
835
// can eliminate edges from the CFG which can end up deleting OldCond.
836
RecursivelyDeleteTriviallyDeadInstructions(OldCond);
837
ConstantFoldTerminator(BB);
842
// Next, figure out which successor we are threading to.
843
BasicBlock *SuccBB = DestBI->getSuccessor(!BranchDir);
845
SmallVector<BasicBlock*, 2> Preds;
846
Preds.push_back(PredBB);
848
// Ok, try to thread it!
849
return ThreadEdge(BB, Preds, SuccBB);
852
/// ProcessSwitchOnDuplicateCond - We found a block and a predecessor of that
853
/// block that switch on exactly the same condition. This means that we almost
854
/// always know the direction of the edge in the DESTBB:
856
/// switch COND [... DESTBB, BBY ... ]
858
/// switch COND [... BBZ, BBW ]
860
/// Optimizing switches like this is very important, because simplifycfg builds
861
/// switches out of repeated 'if' conditions.
862
bool JumpThreading::ProcessSwitchOnDuplicateCond(BasicBlock *PredBB,
863
BasicBlock *DestBB) {
864
// Can't thread edge to self.
865
if (PredBB == DestBB)
868
SwitchInst *PredSI = cast<SwitchInst>(PredBB->getTerminator());
869
SwitchInst *DestSI = cast<SwitchInst>(DestBB->getTerminator());
871
// There are a variety of optimizations that we can potentially do on these
872
// blocks: we order them from most to least preferable.
874
// If DESTBB *just* contains the switch, then we can forward edges from PREDBB
875
// directly to their destination. This does not introduce *any* code size
876
// growth. Skip debug info first.
877
BasicBlock::iterator BBI = DestBB->begin();
878
while (isa<DbgInfoIntrinsic>(BBI))
881
// FIXME: Thread if it just contains a PHI.
882
if (isa<SwitchInst>(BBI)) {
883
bool MadeChange = false;
884
// Ignore the default edge for now.
885
for (unsigned i = 1, e = DestSI->getNumSuccessors(); i != e; ++i) {
886
ConstantInt *DestVal = DestSI->getCaseValue(i);
887
BasicBlock *DestSucc = DestSI->getSuccessor(i);
889
// Okay, DestSI has a case for 'DestVal' that goes to 'DestSucc'. See if
890
// PredSI has an explicit case for it. If so, forward. If it is covered
891
// by the default case, we can't update PredSI.
892
unsigned PredCase = PredSI->findCaseValue(DestVal);
893
if (PredCase == 0) continue;
895
// If PredSI doesn't go to DestBB on this value, then it won't reach the
896
// case on this condition.
897
if (PredSI->getSuccessor(PredCase) != DestBB &&
898
DestSI->getSuccessor(i) != DestBB)
901
// Do not forward this if it already goes to this destination, this would
902
// be an infinite loop.
903
if (PredSI->getSuccessor(PredCase) == DestSucc)
906
// Otherwise, we're safe to make the change. Make sure that the edge from
907
// DestSI to DestSucc is not critical and has no PHI nodes.
908
DEBUG(dbgs() << "FORWARDING EDGE " << *DestVal << " FROM: " << *PredSI);
909
DEBUG(dbgs() << "THROUGH: " << *DestSI);
911
// If the destination has PHI nodes, just split the edge for updating
913
if (isa<PHINode>(DestSucc->begin()) && !DestSucc->getSinglePredecessor()){
914
SplitCriticalEdge(DestSI, i, this);
915
DestSucc = DestSI->getSuccessor(i);
917
FoldSingleEntryPHINodes(DestSucc);
918
PredSI->setSuccessor(PredCase, DestSucc);
930
/// SimplifyPartiallyRedundantLoad - If LI is an obviously partially redundant
931
/// load instruction, eliminate it by replacing it with a PHI node. This is an
932
/// important optimization that encourages jump threading, and needs to be run
933
/// interlaced with other jump threading tasks.
934
bool JumpThreading::SimplifyPartiallyRedundantLoad(LoadInst *LI) {
935
// Don't hack volatile loads.
936
if (LI->isVolatile()) return false;
938
// If the load is defined in a block with exactly one predecessor, it can't be
939
// partially redundant.
940
BasicBlock *LoadBB = LI->getParent();
941
if (LoadBB->getSinglePredecessor())
944
Value *LoadedPtr = LI->getOperand(0);
946
// If the loaded operand is defined in the LoadBB, it can't be available.
947
// TODO: Could do simple PHI translation, that would be fun :)
948
if (Instruction *PtrOp = dyn_cast<Instruction>(LoadedPtr))
949
if (PtrOp->getParent() == LoadBB)
952
// Scan a few instructions up from the load, to see if it is obviously live at
953
// the entry to its block.
954
BasicBlock::iterator BBIt = LI;
956
if (Value *AvailableVal =
957
FindAvailableLoadedValue(LoadedPtr, LoadBB, BBIt, 6)) {
958
// If the value if the load is locally available within the block, just use
959
// it. This frequently occurs for reg2mem'd allocas.
960
//cerr << "LOAD ELIMINATED:\n" << *BBIt << *LI << "\n";
962
// If the returned value is the load itself, replace with an undef. This can
963
// only happen in dead loops.
964
if (AvailableVal == LI) AvailableVal = UndefValue::get(LI->getType());
965
LI->replaceAllUsesWith(AvailableVal);
966
LI->eraseFromParent();
970
// Otherwise, if we scanned the whole block and got to the top of the block,
971
// we know the block is locally transparent to the load. If not, something
972
// might clobber its value.
973
if (BBIt != LoadBB->begin())
977
SmallPtrSet<BasicBlock*, 8> PredsScanned;
978
typedef SmallVector<std::pair<BasicBlock*, Value*>, 8> AvailablePredsTy;
979
AvailablePredsTy AvailablePreds;
980
BasicBlock *OneUnavailablePred = 0;
982
// If we got here, the loaded value is transparent through to the start of the
983
// block. Check to see if it is available in any of the predecessor blocks.
984
for (pred_iterator PI = pred_begin(LoadBB), PE = pred_end(LoadBB);
986
BasicBlock *PredBB = *PI;
988
// If we already scanned this predecessor, skip it.
989
if (!PredsScanned.insert(PredBB))
992
// Scan the predecessor to see if the value is available in the pred.
993
BBIt = PredBB->end();
994
Value *PredAvailable = FindAvailableLoadedValue(LoadedPtr, PredBB, BBIt, 6);
995
if (!PredAvailable) {
996
OneUnavailablePred = PredBB;
1000
// If so, this load is partially redundant. Remember this info so that we
1001
// can create a PHI node.
1002
AvailablePreds.push_back(std::make_pair(PredBB, PredAvailable));
1005
// If the loaded value isn't available in any predecessor, it isn't partially
1007
if (AvailablePreds.empty()) return false;
1009
// Okay, the loaded value is available in at least one (and maybe all!)
1010
// predecessors. If the value is unavailable in more than one unique
1011
// predecessor, we want to insert a merge block for those common predecessors.
1012
// This ensures that we only have to insert one reload, thus not increasing
1014
BasicBlock *UnavailablePred = 0;
1016
// If there is exactly one predecessor where the value is unavailable, the
1017
// already computed 'OneUnavailablePred' block is it. If it ends in an
1018
// unconditional branch, we know that it isn't a critical edge.
1019
if (PredsScanned.size() == AvailablePreds.size()+1 &&
1020
OneUnavailablePred->getTerminator()->getNumSuccessors() == 1) {
1021
UnavailablePred = OneUnavailablePred;
1022
} else if (PredsScanned.size() != AvailablePreds.size()) {
1023
// Otherwise, we had multiple unavailable predecessors or we had a critical
1024
// edge from the one.
1025
SmallVector<BasicBlock*, 8> PredsToSplit;
1026
SmallPtrSet<BasicBlock*, 8> AvailablePredSet;
1028
for (unsigned i = 0, e = AvailablePreds.size(); i != e; ++i)
1029
AvailablePredSet.insert(AvailablePreds[i].first);
1031
// Add all the unavailable predecessors to the PredsToSplit list.
1032
for (pred_iterator PI = pred_begin(LoadBB), PE = pred_end(LoadBB);
1034
BasicBlock *P = *PI;
1035
// If the predecessor is an indirect goto, we can't split the edge.
1036
if (isa<IndirectBrInst>(P->getTerminator()))
1039
if (!AvailablePredSet.count(P))
1040
PredsToSplit.push_back(P);
1043
// Split them out to their own block.
1045
SplitBlockPredecessors(LoadBB, &PredsToSplit[0], PredsToSplit.size(),
1046
"thread-pre-split", this);
1049
// If the value isn't available in all predecessors, then there will be
1050
// exactly one where it isn't available. Insert a load on that edge and add
1051
// it to the AvailablePreds list.
1052
if (UnavailablePred) {
1053
assert(UnavailablePred->getTerminator()->getNumSuccessors() == 1 &&
1054
"Can't handle critical edge here!");
1055
Value *NewVal = new LoadInst(LoadedPtr, LI->getName()+".pr", false,
1057
UnavailablePred->getTerminator());
1058
AvailablePreds.push_back(std::make_pair(UnavailablePred, NewVal));
1061
// Now we know that each predecessor of this block has a value in
1062
// AvailablePreds, sort them for efficient access as we're walking the preds.
1063
array_pod_sort(AvailablePreds.begin(), AvailablePreds.end());
1065
// Create a PHI node at the start of the block for the PRE'd load value.
1066
PHINode *PN = PHINode::Create(LI->getType(), "", LoadBB->begin());
1069
// Insert new entries into the PHI for each predecessor. A single block may
1070
// have multiple entries here.
1071
for (pred_iterator PI = pred_begin(LoadBB), E = pred_end(LoadBB); PI != E;
1073
BasicBlock *P = *PI;
1074
AvailablePredsTy::iterator I =
1075
std::lower_bound(AvailablePreds.begin(), AvailablePreds.end(),
1076
std::make_pair(P, (Value*)0));
1078
assert(I != AvailablePreds.end() && I->first == P &&
1079
"Didn't find entry for predecessor!");
1081
PN->addIncoming(I->second, I->first);
1084
//cerr << "PRE: " << *LI << *PN << "\n";
1086
LI->replaceAllUsesWith(PN);
1087
LI->eraseFromParent();
1092
/// FindMostPopularDest - The specified list contains multiple possible
1093
/// threadable destinations. Pick the one that occurs the most frequently in
1096
FindMostPopularDest(BasicBlock *BB,
1097
const SmallVectorImpl<std::pair<BasicBlock*,
1098
BasicBlock*> > &PredToDestList) {
1099
assert(!PredToDestList.empty());
1101
// Determine popularity. If there are multiple possible destinations, we
1102
// explicitly choose to ignore 'undef' destinations. We prefer to thread
1103
// blocks with known and real destinations to threading undef. We'll handle
1104
// them later if interesting.
1105
DenseMap<BasicBlock*, unsigned> DestPopularity;
1106
for (unsigned i = 0, e = PredToDestList.size(); i != e; ++i)
1107
if (PredToDestList[i].second)
1108
DestPopularity[PredToDestList[i].second]++;
1110
// Find the most popular dest.
1111
DenseMap<BasicBlock*, unsigned>::iterator DPI = DestPopularity.begin();
1112
BasicBlock *MostPopularDest = DPI->first;
1113
unsigned Popularity = DPI->second;
1114
SmallVector<BasicBlock*, 4> SamePopularity;
1116
for (++DPI; DPI != DestPopularity.end(); ++DPI) {
1117
// If the popularity of this entry isn't higher than the popularity we've
1118
// seen so far, ignore it.
1119
if (DPI->second < Popularity)
1121
else if (DPI->second == Popularity) {
1122
// If it is the same as what we've seen so far, keep track of it.
1123
SamePopularity.push_back(DPI->first);
1125
// If it is more popular, remember it.
1126
SamePopularity.clear();
1127
MostPopularDest = DPI->first;
1128
Popularity = DPI->second;
1132
// Okay, now we know the most popular destination. If there is more than
1133
// destination, we need to determine one. This is arbitrary, but we need
1134
// to make a deterministic decision. Pick the first one that appears in the
1136
if (!SamePopularity.empty()) {
1137
SamePopularity.push_back(MostPopularDest);
1138
TerminatorInst *TI = BB->getTerminator();
1139
for (unsigned i = 0; ; ++i) {
1140
assert(i != TI->getNumSuccessors() && "Didn't find any successor!");
1142
if (std::find(SamePopularity.begin(), SamePopularity.end(),
1143
TI->getSuccessor(i)) == SamePopularity.end())
1146
MostPopularDest = TI->getSuccessor(i);
1151
// Okay, we have finally picked the most popular destination.
1152
return MostPopularDest;
1155
bool JumpThreading::ProcessThreadableEdges(Value *Cond, BasicBlock *BB) {
1156
// If threading this would thread across a loop header, don't even try to
1158
if (LoopHeaders.count(BB))
1161
SmallVector<std::pair<ConstantInt*, BasicBlock*>, 8> PredValues;
1162
if (!ComputeValueKnownInPredecessors(Cond, BB, PredValues))
1165
assert(!PredValues.empty() &&
1166
"ComputeValueKnownInPredecessors returned true with no values");
1168
DEBUG(dbgs() << "IN BB: " << *BB;
1169
for (unsigned i = 0, e = PredValues.size(); i != e; ++i) {
1170
dbgs() << " BB '" << BB->getName() << "': FOUND condition = ";
1171
if (PredValues[i].first)
1172
dbgs() << *PredValues[i].first;
1175
dbgs() << " for pred '" << PredValues[i].second->getName()
1179
// Decide what we want to thread through. Convert our list of known values to
1180
// a list of known destinations for each pred. This also discards duplicate
1181
// predecessors and keeps track of the undefined inputs (which are represented
1182
// as a null dest in the PredToDestList).
1183
SmallPtrSet<BasicBlock*, 16> SeenPreds;
1184
SmallVector<std::pair<BasicBlock*, BasicBlock*>, 16> PredToDestList;
1186
BasicBlock *OnlyDest = 0;
1187
BasicBlock *MultipleDestSentinel = (BasicBlock*)(intptr_t)~0ULL;
1189
for (unsigned i = 0, e = PredValues.size(); i != e; ++i) {
1190
BasicBlock *Pred = PredValues[i].second;
1191
if (!SeenPreds.insert(Pred))
1192
continue; // Duplicate predecessor entry.
1194
// If the predecessor ends with an indirect goto, we can't change its
1196
if (isa<IndirectBrInst>(Pred->getTerminator()))
1199
ConstantInt *Val = PredValues[i].first;
1202
if (Val == 0) // Undef.
1204
else if (BranchInst *BI = dyn_cast<BranchInst>(BB->getTerminator()))
1205
DestBB = BI->getSuccessor(Val->isZero());
1207
SwitchInst *SI = cast<SwitchInst>(BB->getTerminator());
1208
DestBB = SI->getSuccessor(SI->findCaseValue(Val));
1211
// If we have exactly one destination, remember it for efficiency below.
1214
else if (OnlyDest != DestBB)
1215
OnlyDest = MultipleDestSentinel;
1217
PredToDestList.push_back(std::make_pair(Pred, DestBB));
1220
// If all edges were unthreadable, we fail.
1221
if (PredToDestList.empty())
1224
// Determine which is the most common successor. If we have many inputs and
1225
// this block is a switch, we want to start by threading the batch that goes
1226
// to the most popular destination first. If we only know about one
1227
// threadable destination (the common case) we can avoid this.
1228
BasicBlock *MostPopularDest = OnlyDest;
1230
if (MostPopularDest == MultipleDestSentinel)
1231
MostPopularDest = FindMostPopularDest(BB, PredToDestList);
1233
// Now that we know what the most popular destination is, factor all
1234
// predecessors that will jump to it into a single predecessor.
1235
SmallVector<BasicBlock*, 16> PredsToFactor;
1236
for (unsigned i = 0, e = PredToDestList.size(); i != e; ++i)
1237
if (PredToDestList[i].second == MostPopularDest) {
1238
BasicBlock *Pred = PredToDestList[i].first;
1240
// This predecessor may be a switch or something else that has multiple
1241
// edges to the block. Factor each of these edges by listing them
1242
// according to # occurrences in PredsToFactor.
1243
TerminatorInst *PredTI = Pred->getTerminator();
1244
for (unsigned i = 0, e = PredTI->getNumSuccessors(); i != e; ++i)
1245
if (PredTI->getSuccessor(i) == BB)
1246
PredsToFactor.push_back(Pred);
1249
// If the threadable edges are branching on an undefined value, we get to pick
1250
// the destination that these predecessors should get to.
1251
if (MostPopularDest == 0)
1252
MostPopularDest = BB->getTerminator()->
1253
getSuccessor(GetBestDestForJumpOnUndef(BB));
1255
// Ok, try to thread it!
1256
return ThreadEdge(BB, PredsToFactor, MostPopularDest);
1259
/// ProcessBranchOnPHI - We have an otherwise unthreadable conditional branch on
1260
/// a PHI node in the current block. See if there are any simplifications we
1261
/// can do based on inputs to the phi node.
1263
bool JumpThreading::ProcessBranchOnPHI(PHINode *PN) {
1264
BasicBlock *BB = PN->getParent();
1266
// TODO: We could make use of this to do it once for blocks with common PHI
1268
SmallVector<BasicBlock*, 1> PredBBs;
1271
// If any of the predecessor blocks end in an unconditional branch, we can
1272
// *duplicate* the conditional branch into that block in order to further
1273
// encourage jump threading and to eliminate cases where we have branch on a
1274
// phi of an icmp (branch on icmp is much better).
1275
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
1276
BasicBlock *PredBB = PN->getIncomingBlock(i);
1277
if (BranchInst *PredBr = dyn_cast<BranchInst>(PredBB->getTerminator()))
1278
if (PredBr->isUnconditional()) {
1279
PredBBs[0] = PredBB;
1280
// Try to duplicate BB into PredBB.
1281
if (DuplicateCondBranchOnPHIIntoPred(BB, PredBBs))
1289
/// ProcessBranchOnXOR - We have an otherwise unthreadable conditional branch on
1290
/// a xor instruction in the current block. See if there are any
1291
/// simplifications we can do based on inputs to the xor.
1293
bool JumpThreading::ProcessBranchOnXOR(BinaryOperator *BO) {
1294
BasicBlock *BB = BO->getParent();
1296
// If either the LHS or RHS of the xor is a constant, don't do this
1298
if (isa<ConstantInt>(BO->getOperand(0)) ||
1299
isa<ConstantInt>(BO->getOperand(1)))
1302
// If the first instruction in BB isn't a phi, we won't be able to infer
1303
// anything special about any particular predecessor.
1304
if (!isa<PHINode>(BB->front()))
1307
// If we have a xor as the branch input to this block, and we know that the
1308
// LHS or RHS of the xor in any predecessor is true/false, then we can clone
1309
// the condition into the predecessor and fix that value to true, saving some
1310
// logical ops on that path and encouraging other paths to simplify.
1312
// This copies something like this:
1315
// %X = phi i1 [1], [%X']
1316
// %Y = icmp eq i32 %A, %B
1317
// %Z = xor i1 %X, %Y
1322
// %Y = icmp ne i32 %A, %B
1325
SmallVector<std::pair<ConstantInt*, BasicBlock*>, 8> XorOpValues;
1327
if (!ComputeValueKnownInPredecessors(BO->getOperand(0), BB, XorOpValues)) {
1328
assert(XorOpValues.empty());
1329
if (!ComputeValueKnownInPredecessors(BO->getOperand(1), BB, XorOpValues))
1334
assert(!XorOpValues.empty() &&
1335
"ComputeValueKnownInPredecessors returned true with no values");
1337
// Scan the information to see which is most popular: true or false. The
1338
// predecessors can be of the set true, false, or undef.
1339
unsigned NumTrue = 0, NumFalse = 0;
1340
for (unsigned i = 0, e = XorOpValues.size(); i != e; ++i) {
1341
if (!XorOpValues[i].first) continue; // Ignore undefs for the count.
1342
if (XorOpValues[i].first->isZero())
1348
// Determine which value to split on, true, false, or undef if neither.
1349
ConstantInt *SplitVal = 0;
1350
if (NumTrue > NumFalse)
1351
SplitVal = ConstantInt::getTrue(BB->getContext());
1352
else if (NumTrue != 0 || NumFalse != 0)
1353
SplitVal = ConstantInt::getFalse(BB->getContext());
1355
// Collect all of the blocks that this can be folded into so that we can
1356
// factor this once and clone it once.
1357
SmallVector<BasicBlock*, 8> BlocksToFoldInto;
1358
for (unsigned i = 0, e = XorOpValues.size(); i != e; ++i) {
1359
if (XorOpValues[i].first != SplitVal && XorOpValues[i].first != 0) continue;
1361
BlocksToFoldInto.push_back(XorOpValues[i].second);
1364
// If we inferred a value for all of the predecessors, then duplication won't
1365
// help us. However, we can just replace the LHS or RHS with the constant.
1366
if (BlocksToFoldInto.size() ==
1367
cast<PHINode>(BB->front()).getNumIncomingValues()) {
1368
if (SplitVal == 0) {
1369
// If all preds provide undef, just nuke the xor, because it is undef too.
1370
BO->replaceAllUsesWith(UndefValue::get(BO->getType()));
1371
BO->eraseFromParent();
1372
} else if (SplitVal->isZero()) {
1373
// If all preds provide 0, replace the xor with the other input.
1374
BO->replaceAllUsesWith(BO->getOperand(isLHS));
1375
BO->eraseFromParent();
1377
// If all preds provide 1, set the computed value to 1.
1378
BO->setOperand(!isLHS, SplitVal);
1384
// Try to duplicate BB into PredBB.
1385
return DuplicateCondBranchOnPHIIntoPred(BB, BlocksToFoldInto);
1389
/// AddPHINodeEntriesForMappedBlock - We're adding 'NewPred' as a new
1390
/// predecessor to the PHIBB block. If it has PHI nodes, add entries for
1391
/// NewPred using the entries from OldPred (suitably mapped).
1392
static void AddPHINodeEntriesForMappedBlock(BasicBlock *PHIBB,
1393
BasicBlock *OldPred,
1394
BasicBlock *NewPred,
1395
DenseMap<Instruction*, Value*> &ValueMap) {
1396
for (BasicBlock::iterator PNI = PHIBB->begin();
1397
PHINode *PN = dyn_cast<PHINode>(PNI); ++PNI) {
1398
// Ok, we have a PHI node. Figure out what the incoming value was for the
1400
Value *IV = PN->getIncomingValueForBlock(OldPred);
1402
// Remap the value if necessary.
1403
if (Instruction *Inst = dyn_cast<Instruction>(IV)) {
1404
DenseMap<Instruction*, Value*>::iterator I = ValueMap.find(Inst);
1405
if (I != ValueMap.end())
1409
PN->addIncoming(IV, NewPred);
1413
/// ThreadEdge - We have decided that it is safe and profitable to factor the
1414
/// blocks in PredBBs to one predecessor, then thread an edge from it to SuccBB
1415
/// across BB. Transform the IR to reflect this change.
1416
bool JumpThreading::ThreadEdge(BasicBlock *BB,
1417
const SmallVectorImpl<BasicBlock*> &PredBBs,
1418
BasicBlock *SuccBB) {
1419
// If threading to the same block as we come from, we would infinite loop.
1421
DEBUG(dbgs() << " Not threading across BB '" << BB->getName()
1422
<< "' - would thread to self!\n");
1426
// If threading this would thread across a loop header, don't thread the edge.
1427
// See the comments above FindLoopHeaders for justifications and caveats.
1428
if (LoopHeaders.count(BB)) {
1429
DEBUG(dbgs() << " Not threading across loop header BB '" << BB->getName()
1430
<< "' to dest BB '" << SuccBB->getName()
1431
<< "' - it might create an irreducible loop!\n");
1435
unsigned JumpThreadCost = getJumpThreadDuplicationCost(BB);
1436
if (JumpThreadCost > Threshold) {
1437
DEBUG(dbgs() << " Not threading BB '" << BB->getName()
1438
<< "' - Cost is too high: " << JumpThreadCost << "\n");
1442
// And finally, do it! Start by factoring the predecessors is needed.
1444
if (PredBBs.size() == 1)
1445
PredBB = PredBBs[0];
1447
DEBUG(dbgs() << " Factoring out " << PredBBs.size()
1448
<< " common predecessors.\n");
1449
PredBB = SplitBlockPredecessors(BB, &PredBBs[0], PredBBs.size(),
1453
// And finally, do it!
1454
DEBUG(dbgs() << " Threading edge from '" << PredBB->getName() << "' to '"
1455
<< SuccBB->getName() << "' with cost: " << JumpThreadCost
1456
<< ", across block:\n "
1460
LVI->threadEdge(PredBB, BB, SuccBB);
1462
// We are going to have to map operands from the original BB block to the new
1463
// copy of the block 'NewBB'. If there are PHI nodes in BB, evaluate them to
1464
// account for entry from PredBB.
1465
DenseMap<Instruction*, Value*> ValueMapping;
1467
BasicBlock *NewBB = BasicBlock::Create(BB->getContext(),
1468
BB->getName()+".thread",
1469
BB->getParent(), BB);
1470
NewBB->moveAfter(PredBB);
1472
BasicBlock::iterator BI = BB->begin();
1473
for (; PHINode *PN = dyn_cast<PHINode>(BI); ++BI)
1474
ValueMapping[PN] = PN->getIncomingValueForBlock(PredBB);
1476
// Clone the non-phi instructions of BB into NewBB, keeping track of the
1477
// mapping and using it to remap operands in the cloned instructions.
1478
for (; !isa<TerminatorInst>(BI); ++BI) {
1479
Instruction *New = BI->clone();
1480
New->setName(BI->getName());
1481
NewBB->getInstList().push_back(New);
1482
ValueMapping[BI] = New;
1484
// Remap operands to patch up intra-block references.
1485
for (unsigned i = 0, e = New->getNumOperands(); i != e; ++i)
1486
if (Instruction *Inst = dyn_cast<Instruction>(New->getOperand(i))) {
1487
DenseMap<Instruction*, Value*>::iterator I = ValueMapping.find(Inst);
1488
if (I != ValueMapping.end())
1489
New->setOperand(i, I->second);
1493
// We didn't copy the terminator from BB over to NewBB, because there is now
1494
// an unconditional jump to SuccBB. Insert the unconditional jump.
1495
BranchInst::Create(SuccBB, NewBB);
1497
// Check to see if SuccBB has PHI nodes. If so, we need to add entries to the
1498
// PHI nodes for NewBB now.
1499
AddPHINodeEntriesForMappedBlock(SuccBB, BB, NewBB, ValueMapping);
1501
// If there were values defined in BB that are used outside the block, then we
1502
// now have to update all uses of the value to use either the original value,
1503
// the cloned value, or some PHI derived value. This can require arbitrary
1504
// PHI insertion, of which we are prepared to do, clean these up now.
1505
SSAUpdater SSAUpdate;
1506
SmallVector<Use*, 16> UsesToRename;
1507
for (BasicBlock::iterator I = BB->begin(); I != BB->end(); ++I) {
1508
// Scan all uses of this instruction to see if it is used outside of its
1509
// block, and if so, record them in UsesToRename.
1510
for (Value::use_iterator UI = I->use_begin(), E = I->use_end(); UI != E;
1512
Instruction *User = cast<Instruction>(*UI);
1513
if (PHINode *UserPN = dyn_cast<PHINode>(User)) {
1514
if (UserPN->getIncomingBlock(UI) == BB)
1516
} else if (User->getParent() == BB)
1519
UsesToRename.push_back(&UI.getUse());
1522
// If there are no uses outside the block, we're done with this instruction.
1523
if (UsesToRename.empty())
1526
DEBUG(dbgs() << "JT: Renaming non-local uses of: " << *I << "\n");
1528
// We found a use of I outside of BB. Rename all uses of I that are outside
1529
// its block to be uses of the appropriate PHI node etc. See ValuesInBlocks
1530
// with the two values we know.
1531
SSAUpdate.Initialize(I->getType(), I->getName());
1532
SSAUpdate.AddAvailableValue(BB, I);
1533
SSAUpdate.AddAvailableValue(NewBB, ValueMapping[I]);
1535
while (!UsesToRename.empty())
1536
SSAUpdate.RewriteUse(*UsesToRename.pop_back_val());
1537
DEBUG(dbgs() << "\n");
1541
// Ok, NewBB is good to go. Update the terminator of PredBB to jump to
1542
// NewBB instead of BB. This eliminates predecessors from BB, which requires
1543
// us to simplify any PHI nodes in BB.
1544
TerminatorInst *PredTerm = PredBB->getTerminator();
1545
for (unsigned i = 0, e = PredTerm->getNumSuccessors(); i != e; ++i)
1546
if (PredTerm->getSuccessor(i) == BB) {
1547
RemovePredecessorAndSimplify(BB, PredBB, TD);
1548
PredTerm->setSuccessor(i, NewBB);
1551
// At this point, the IR is fully up to date and consistent. Do a quick scan
1552
// over the new instructions and zap any that are constants or dead. This
1553
// frequently happens because of phi translation.
1554
SimplifyInstructionsInBlock(NewBB, TD);
1556
// Threaded an edge!
1561
/// DuplicateCondBranchOnPHIIntoPred - PredBB contains an unconditional branch
1562
/// to BB which contains an i1 PHI node and a conditional branch on that PHI.
1563
/// If we can duplicate the contents of BB up into PredBB do so now, this
1564
/// improves the odds that the branch will be on an analyzable instruction like
1566
bool JumpThreading::DuplicateCondBranchOnPHIIntoPred(BasicBlock *BB,
1567
const SmallVectorImpl<BasicBlock *> &PredBBs) {
1568
assert(!PredBBs.empty() && "Can't handle an empty set");
1570
// If BB is a loop header, then duplicating this block outside the loop would
1571
// cause us to transform this into an irreducible loop, don't do this.
1572
// See the comments above FindLoopHeaders for justifications and caveats.
1573
if (LoopHeaders.count(BB)) {
1574
DEBUG(dbgs() << " Not duplicating loop header '" << BB->getName()
1575
<< "' into predecessor block '" << PredBBs[0]->getName()
1576
<< "' - it might create an irreducible loop!\n");
1580
unsigned DuplicationCost = getJumpThreadDuplicationCost(BB);
1581
if (DuplicationCost > Threshold) {
1582
DEBUG(dbgs() << " Not duplicating BB '" << BB->getName()
1583
<< "' - Cost is too high: " << DuplicationCost << "\n");
1587
// And finally, do it! Start by factoring the predecessors is needed.
1589
if (PredBBs.size() == 1)
1590
PredBB = PredBBs[0];
1592
DEBUG(dbgs() << " Factoring out " << PredBBs.size()
1593
<< " common predecessors.\n");
1594
PredBB = SplitBlockPredecessors(BB, &PredBBs[0], PredBBs.size(),
1598
// Okay, we decided to do this! Clone all the instructions in BB onto the end
1600
DEBUG(dbgs() << " Duplicating block '" << BB->getName() << "' into end of '"
1601
<< PredBB->getName() << "' to eliminate branch on phi. Cost: "
1602
<< DuplicationCost << " block is:" << *BB << "\n");
1604
// Unless PredBB ends with an unconditional branch, split the edge so that we
1605
// can just clone the bits from BB into the end of the new PredBB.
1606
BranchInst *OldPredBranch = dyn_cast<BranchInst>(PredBB->getTerminator());
1608
if (OldPredBranch == 0 || !OldPredBranch->isUnconditional()) {
1609
PredBB = SplitEdge(PredBB, BB, this);
1610
OldPredBranch = cast<BranchInst>(PredBB->getTerminator());
1613
// We are going to have to map operands from the original BB block into the
1614
// PredBB block. Evaluate PHI nodes in BB.
1615
DenseMap<Instruction*, Value*> ValueMapping;
1617
BasicBlock::iterator BI = BB->begin();
1618
for (; PHINode *PN = dyn_cast<PHINode>(BI); ++BI)
1619
ValueMapping[PN] = PN->getIncomingValueForBlock(PredBB);
1621
// Clone the non-phi instructions of BB into PredBB, keeping track of the
1622
// mapping and using it to remap operands in the cloned instructions.
1623
for (; BI != BB->end(); ++BI) {
1624
Instruction *New = BI->clone();
1626
// Remap operands to patch up intra-block references.
1627
for (unsigned i = 0, e = New->getNumOperands(); i != e; ++i)
1628
if (Instruction *Inst = dyn_cast<Instruction>(New->getOperand(i))) {
1629
DenseMap<Instruction*, Value*>::iterator I = ValueMapping.find(Inst);
1630
if (I != ValueMapping.end())
1631
New->setOperand(i, I->second);
1634
// If this instruction can be simplified after the operands are updated,
1635
// just use the simplified value instead. This frequently happens due to
1637
if (Value *IV = SimplifyInstruction(New, TD)) {
1639
ValueMapping[BI] = IV;
1641
// Otherwise, insert the new instruction into the block.
1642
New->setName(BI->getName());
1643
PredBB->getInstList().insert(OldPredBranch, New);
1644
ValueMapping[BI] = New;
1648
// Check to see if the targets of the branch had PHI nodes. If so, we need to
1649
// add entries to the PHI nodes for branch from PredBB now.
1650
BranchInst *BBBranch = cast<BranchInst>(BB->getTerminator());
1651
AddPHINodeEntriesForMappedBlock(BBBranch->getSuccessor(0), BB, PredBB,
1653
AddPHINodeEntriesForMappedBlock(BBBranch->getSuccessor(1), BB, PredBB,
1656
// If there were values defined in BB that are used outside the block, then we
1657
// now have to update all uses of the value to use either the original value,
1658
// the cloned value, or some PHI derived value. This can require arbitrary
1659
// PHI insertion, of which we are prepared to do, clean these up now.
1660
SSAUpdater SSAUpdate;
1661
SmallVector<Use*, 16> UsesToRename;
1662
for (BasicBlock::iterator I = BB->begin(); I != BB->end(); ++I) {
1663
// Scan all uses of this instruction to see if it is used outside of its
1664
// block, and if so, record them in UsesToRename.
1665
for (Value::use_iterator UI = I->use_begin(), E = I->use_end(); UI != E;
1667
Instruction *User = cast<Instruction>(*UI);
1668
if (PHINode *UserPN = dyn_cast<PHINode>(User)) {
1669
if (UserPN->getIncomingBlock(UI) == BB)
1671
} else if (User->getParent() == BB)
1674
UsesToRename.push_back(&UI.getUse());
1677
// If there are no uses outside the block, we're done with this instruction.
1678
if (UsesToRename.empty())
1681
DEBUG(dbgs() << "JT: Renaming non-local uses of: " << *I << "\n");
1683
// We found a use of I outside of BB. Rename all uses of I that are outside
1684
// its block to be uses of the appropriate PHI node etc. See ValuesInBlocks
1685
// with the two values we know.
1686
SSAUpdate.Initialize(I->getType(), I->getName());
1687
SSAUpdate.AddAvailableValue(BB, I);
1688
SSAUpdate.AddAvailableValue(PredBB, ValueMapping[I]);
1690
while (!UsesToRename.empty())
1691
SSAUpdate.RewriteUse(*UsesToRename.pop_back_val());
1692
DEBUG(dbgs() << "\n");
1695
// PredBB no longer jumps to BB, remove entries in the PHI node for the edge
1697
RemovePredecessorAndSimplify(BB, PredBB, TD);
1699
// Remove the unconditional branch at the end of the PredBB block.
1700
OldPredBranch->eraseFromParent();
added
removed
libclamav/c++/llvm/lib/Transforms/Scalar/JumpThreading.cpp
