#define DEBUG_TYPE "codegenprepare"
#include "llvm/Transforms/Scalar.h"
+#include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/SmallSet.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/Analysis/DominatorInternals.h"
+#include "llvm/Analysis/Dominators.h"
+#include "llvm/Analysis/InstructionSimplify.h"
+#include "llvm/Analysis/ProfileInfo.h"
+#include "llvm/Assembly/Writer.h"
#include "llvm/Constants.h"
+#include "llvm/DataLayout.h"
#include "llvm/DerivedTypes.h"
#include "llvm/Function.h"
+#include "llvm/IRBuilder.h"
#include "llvm/InlineAsm.h"
#include "llvm/Instructions.h"
#include "llvm/IntrinsicInst.h"
#include "llvm/Pass.h"
-#include "llvm/Analysis/Dominators.h"
-#include "llvm/Analysis/InstructionSimplify.h"
-#include "llvm/Analysis/ProfileInfo.h"
-#include "llvm/Target/TargetData.h"
-#include "llvm/Target/TargetLowering.h"
-#include "llvm/Transforms/Utils/AddrModeMatcher.h"
-#include "llvm/Transforms/Utils/BasicBlockUtils.h"
-#include "llvm/Transforms/Utils/Local.h"
-#include "llvm/Transforms/Utils/BuildLibCalls.h"
-#include "llvm/ADT/DenseMap.h"
-#include "llvm/ADT/SmallSet.h"
-#include "llvm/ADT/Statistic.h"
-#include "llvm/Assembly/Writer.h"
#include "llvm/Support/CallSite.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/GetElementPtrTypeIterator.h"
#include "llvm/Support/PatternMatch.h"
-#include "llvm/Support/raw_ostream.h"
-#include "llvm/Support/IRBuilder.h"
#include "llvm/Support/ValueHandle.h"
+#include "llvm/Support/raw_ostream.h"
+#include "llvm/Target/TargetLibraryInfo.h"
+#include "llvm/Target/TargetLowering.h"
+#include "llvm/Transforms/Utils/AddrModeMatcher.h"
+#include "llvm/Transforms/Utils/BasicBlockUtils.h"
+#include "llvm/Transforms/Utils/BuildLibCalls.h"
+#include "llvm/Transforms/Utils/BypassSlowDivision.h"
+#include "llvm/Transforms/Utils/Local.h"
using namespace llvm;
using namespace llvm::PatternMatch;
STATISTIC(NumBlocksElim, "Number of blocks eliminated");
-STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
-STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
+STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
+STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
STATISTIC(NumCmpUses, "Number of uses of Cmp expressions replaced with uses of "
"sunken Cmps");
STATISTIC(NumCastUses, "Number of uses of Cast expressions replaced with uses "
"of sunken Casts");
STATISTIC(NumMemoryInsts, "Number of memory instructions whose address "
"computations were sunk");
-STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
-STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
+STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
+STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
+STATISTIC(NumRetsDup, "Number of return instructions duplicated");
+STATISTIC(NumDbgValueMoved, "Number of debug value instructions moved");
+STATISTIC(NumSelectsExpanded, "Number of selects turned into branches");
+
+static cl::opt<bool> DisableBranchOpts(
+ "disable-cgp-branch-opts", cl::Hidden, cl::init(false),
+ cl::desc("Disable branch optimizations in CodeGenPrepare"));
+
+static cl::opt<bool> DisableSelectToBranch(
+ "disable-cgp-select2branch", cl::Hidden, cl::init(false),
+ cl::desc("Disable select to branch conversion."));
namespace {
class CodeGenPrepare : public FunctionPass {
/// TLI - Keep a pointer of a TargetLowering to consult for determining
/// transformation profitability.
const TargetLowering *TLI;
+ const TargetLibraryInfo *TLInfo;
DominatorTree *DT;
ProfileInfo *PFI;
-
+
/// CurInstIterator - As we scan instructions optimizing them, this is the
/// next instruction to optimize. Xforms that can invalidate this should
/// update it.
BasicBlock::iterator CurInstIterator;
- // Keeps track of non-local addresses that have been sunk into a block. This
- // allows us to avoid inserting duplicate code for blocks with multiple
- // load/stores of the same address.
+ /// Keeps track of non-local addresses that have been sunk into a block.
+ /// This allows us to avoid inserting duplicate code for blocks with
+ /// multiple load/stores of the same address.
DenseMap<Value*, Value*> SunkAddrs;
+ /// ModifiedDT - If CFG is modified in anyway, dominator tree may need to
+ /// be updated.
+ bool ModifiedDT;
+
+ /// OptSize - True if optimizing for size.
+ bool OptSize;
+
public:
static char ID; // Pass identification, replacement for typeid
explicit CodeGenPrepare(const TargetLowering *tli = 0)
}
bool runOnFunction(Function &F);
+ const char *getPassName() const { return "CodeGen Prepare"; }
+
virtual void getAnalysisUsage(AnalysisUsage &AU) const {
AU.addPreserved<DominatorTree>();
AU.addPreserved<ProfileInfo>();
+ AU.addRequired<TargetLibraryInfo>();
}
private:
+ bool EliminateFallThrough(Function &F);
bool EliminateMostlyEmptyBlocks(Function &F);
bool CanMergeBlocks(const BasicBlock *BB, const BasicBlock *DestBB) const;
void EliminateMostlyEmptyBlock(BasicBlock *BB);
bool OptimizeBlock(BasicBlock &BB);
bool OptimizeInst(Instruction *I);
- bool OptimizeMemoryInst(Instruction *I, Value *Addr, const Type *AccessTy);
+ bool OptimizeMemoryInst(Instruction *I, Value *Addr, Type *AccessTy);
bool OptimizeInlineAsmInst(CallInst *CS);
bool OptimizeCallInst(CallInst *CI);
bool MoveExtToFormExtLoad(Instruction *I);
bool OptimizeExtUses(Instruction *I);
+ bool OptimizeSelectInst(SelectInst *SI);
+ bool DupRetToEnableTailCallOpts(BasicBlock *BB);
+ bool PlaceDbgValues(Function &F);
};
}
char CodeGenPrepare::ID = 0;
-INITIALIZE_PASS(CodeGenPrepare, "codegenprepare",
+INITIALIZE_PASS_BEGIN(CodeGenPrepare, "codegenprepare",
+ "Optimize for code generation", false, false)
+INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfo)
+INITIALIZE_PASS_END(CodeGenPrepare, "codegenprepare",
"Optimize for code generation", false, false)
FunctionPass *llvm::createCodeGenPreparePass(const TargetLowering *TLI) {
bool CodeGenPrepare::runOnFunction(Function &F) {
bool EverMadeChange = false;
+ ModifiedDT = false;
+ TLInfo = &getAnalysis<TargetLibraryInfo>();
DT = getAnalysisIfAvailable<DominatorTree>();
PFI = getAnalysisIfAvailable<ProfileInfo>();
- // First pass, eliminate blocks that contain only PHI nodes and an
+ OptSize = F.getAttributes().hasAttribute(AttributeSet::FunctionIndex,
+ Attribute::OptimizeForSize);
+
+ /// This optimization identifies DIV instructions that can be
+ /// profitably bypassed and carried out with a shorter, faster divide.
+ if (TLI && TLI->isSlowDivBypassed()) {
+ const DenseMap<unsigned int, unsigned int> &BypassWidths =
+ TLI->getBypassSlowDivWidths();
+ for (Function::iterator I = F.begin(); I != F.end(); I++)
+ EverMadeChange |= bypassSlowDivision(F, I, BypassWidths);
+ }
+
+ // Eliminate blocks that contain only PHI nodes and an
// unconditional branch.
EverMadeChange |= EliminateMostlyEmptyBlocks(F);
+ // llvm.dbg.value is far away from the value then iSel may not be able
+ // handle it properly. iSel will drop llvm.dbg.value if it can not
+ // find a node corresponding to the value.
+ EverMadeChange |= PlaceDbgValues(F);
+
bool MadeChange = true;
while (MadeChange) {
MadeChange = false;
- for (Function::iterator BB = F.begin(), E = F.end(); BB != E; ++BB)
+ for (Function::iterator I = F.begin(); I != F.end(); ) {
+ BasicBlock *BB = I++;
MadeChange |= OptimizeBlock(*BB);
+ }
EverMadeChange |= MadeChange;
}
SunkAddrs.clear();
+ if (!DisableBranchOpts) {
+ MadeChange = false;
+ SmallPtrSet<BasicBlock*, 8> WorkList;
+ for (Function::iterator BB = F.begin(), E = F.end(); BB != E; ++BB) {
+ SmallVector<BasicBlock*, 2> Successors(succ_begin(BB), succ_end(BB));
+ MadeChange |= ConstantFoldTerminator(BB, true);
+ if (!MadeChange) continue;
+
+ for (SmallVectorImpl<BasicBlock*>::iterator
+ II = Successors.begin(), IE = Successors.end(); II != IE; ++II)
+ if (pred_begin(*II) == pred_end(*II))
+ WorkList.insert(*II);
+ }
+
+ // Delete the dead blocks and any of their dead successors.
+ MadeChange |= !WorkList.empty();
+ while (!WorkList.empty()) {
+ BasicBlock *BB = *WorkList.begin();
+ WorkList.erase(BB);
+ SmallVector<BasicBlock*, 2> Successors(succ_begin(BB), succ_end(BB));
+
+ DeleteDeadBlock(BB);
+
+ for (SmallVectorImpl<BasicBlock*>::iterator
+ II = Successors.begin(), IE = Successors.end(); II != IE; ++II)
+ if (pred_begin(*II) == pred_end(*II))
+ WorkList.insert(*II);
+ }
+
+ // Merge pairs of basic blocks with unconditional branches, connected by
+ // a single edge.
+ if (EverMadeChange || MadeChange)
+ MadeChange |= EliminateFallThrough(F);
+
+ if (MadeChange)
+ ModifiedDT = true;
+ EverMadeChange |= MadeChange;
+ }
+
+ if (ModifiedDT && DT)
+ DT->DT->recalculate(F);
+
return EverMadeChange;
}
+/// EliminateFallThrough - Merge basic blocks which are connected
+/// by a single edge, where one of the basic blocks has a single successor
+/// pointing to the other basic block, which has a single predecessor.
+bool CodeGenPrepare::EliminateFallThrough(Function &F) {
+ bool Changed = false;
+ // Scan all of the blocks in the function, except for the entry block.
+ for (Function::iterator I = ++F.begin(), E = F.end(); I != E; ) {
+ BasicBlock *BB = I++;
+ // If the destination block has a single pred, then this is a trivial
+ // edge, just collapse it.
+ BasicBlock *SinglePred = BB->getSinglePredecessor();
+
+ // Don't merge if BB's address is taken.
+ if (!SinglePred || SinglePred == BB || BB->hasAddressTaken()) continue;
+
+ BranchInst *Term = dyn_cast<BranchInst>(SinglePred->getTerminator());
+ if (Term && !Term->isConditional()) {
+ Changed = true;
+ DEBUG(dbgs() << "To merge:\n"<< *SinglePred << "\n\n\n");
+ // Remember if SinglePred was the entry block of the function.
+ // If so, we will need to move BB back to the entry position.
+ bool isEntry = SinglePred == &SinglePred->getParent()->getEntryBlock();
+ MergeBasicBlockIntoOnlyPred(BB, this);
+
+ if (isEntry && BB != &BB->getParent()->getEntryBlock())
+ BB->moveBefore(&BB->getParent()->getEntryBlock());
+
+ // We have erased a block. Update the iterator.
+ I = BB;
+ }
+ }
+ return Changed;
+}
+
/// EliminateMostlyEmptyBlocks - eliminate blocks that contain only PHI nodes,
/// debug info directives, and an unconditional branch. Passes before isel
/// (e.g. LSR/loopsimplify) often split edges in ways that are non-optimal for
if (isEntry && BB != &BB->getParent()->getEntryBlock())
BB->moveBefore(&BB->getParent()->getEntryBlock());
-
+
DEBUG(dbgs() << "AFTER:\n" << *DestBB << "\n\n\n");
return;
}
// The PHIs are now updated, change everything that refers to BB to use
// DestBB and remove BB.
BB->replaceAllUsesWith(DestBB);
- if (DT) {
+ if (DT && !ModifiedDT) {
BasicBlock *BBIDom = DT->getNode(BB)->getIDom()->getBlock();
BasicBlock *DestBBIDom = DT->getNode(DestBB)->getIDom()->getBlock();
BasicBlock *NewIDom = DT->findNearestCommonDominator(BBIDom, DestBBIDom);
// If these values will be promoted, find out what they will be promoted
// to. This helps us consider truncates on PPC as noop copies when they
// are.
- if (TLI.getTypeAction(SrcVT) == TargetLowering::Promote)
+ if (TLI.getTypeAction(CI->getContext(), SrcVT) ==
+ TargetLowering::TypePromoteInteger)
SrcVT = TLI.getTypeToTransformTo(CI->getContext(), SrcVT);
- if (TLI.getTypeAction(DstVT) == TargetLowering::Promote)
+ if (TLI.getTypeAction(CI->getContext(), DstVT) ==
+ TargetLowering::TypePromoteInteger)
DstVT = TLI.getTypeToTransformTo(CI->getContext(), DstVT);
// If, after promotion, these are the same types, this is a noop copy.
CastInst *&InsertedCast = InsertedCasts[UserBB];
if (!InsertedCast) {
- BasicBlock::iterator InsertPt = UserBB->getFirstNonPHI();
-
+ BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
InsertedCast =
CastInst::Create(CI->getOpcode(), CI->getOperand(0), CI->getType(), "",
InsertPt);
CmpInst *&InsertedCmp = InsertedCmps[UserBB];
if (!InsertedCmp) {
- BasicBlock::iterator InsertPt = UserBB->getFirstNonPHI();
-
+ BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
InsertedCmp =
CmpInst::Create(CI->getOpcode(),
CI->getPredicate(), CI->getOperand(0),
bool CodeGenPrepare::OptimizeCallInst(CallInst *CI) {
BasicBlock *BB = CI->getParent();
-
+
// Lower inline assembly if we can.
// If we found an inline asm expession, and if the target knows how to
// lower it to normal LLVM code, do so now.
if (OptimizeInlineAsmInst(CI))
return true;
}
-
+
// Lower all uses of llvm.objectsize.*
IntrinsicInst *II = dyn_cast<IntrinsicInst>(CI);
if (II && II->getIntrinsicID() == Intrinsic::objectsize) {
bool Min = (cast<ConstantInt>(II->getArgOperand(1))->getZExtValue() == 1);
- const Type *ReturnTy = CI->getType();
- Constant *RetVal = ConstantInt::get(ReturnTy, Min ? 0 : -1ULL);
-
+ Type *ReturnTy = CI->getType();
+ Constant *RetVal = ConstantInt::get(ReturnTy, Min ? 0 : -1ULL);
+
// Substituting this can cause recursive simplifications, which can
// invalidate our iterator. Use a WeakVH to hold onto it in case this
// happens.
WeakVH IterHandle(CurInstIterator);
-
- ReplaceAndSimplifyAllUses(CI, RetVal, TLI ? TLI->getTargetData() : 0, DT);
+
+ replaceAndRecursivelySimplify(CI, RetVal, TLI ? TLI->getDataLayout() : 0,
+ TLInfo, ModifiedDT ? 0 : DT);
// If the iterator instruction was recursively deleted, start over at the
// start of the block.
return true;
}
+ if (II && TLI) {
+ SmallVector<Value*, 2> PtrOps;
+ Type *AccessTy;
+ if (TLI->GetAddrModeArguments(II, PtrOps, AccessTy))
+ while (!PtrOps.empty())
+ if (OptimizeMemoryInst(II, PtrOps.pop_back_val(), AccessTy))
+ return true;
+ }
+
// From here on out we're working with named functions.
if (CI->getCalledFunction() == 0) return false;
-
- // We'll need TargetData from here on out.
- const TargetData *TD = TLI ? TLI->getTargetData() : 0;
+
+ // We'll need DataLayout from here on out.
+ const DataLayout *TD = TLI ? TLI->getDataLayout() : 0;
if (!TD) return false;
-
+
// Lower all default uses of _chk calls. This is very similar
// to what InstCombineCalls does, but here we are only lowering calls
// that have the default "don't know" as the objectsize. Anything else
// should be left alone.
CodeGenPrepareFortifiedLibCalls Simplifier;
- return Simplifier.fold(CI, TD);
+ return Simplifier.fold(CI, TD, TLInfo);
+}
+
+/// DupRetToEnableTailCallOpts - Look for opportunities to duplicate return
+/// instructions to the predecessor to enable tail call optimizations. The
+/// case it is currently looking for is:
+/// @code
+/// bb0:
+/// %tmp0 = tail call i32 @f0()
+/// br label %return
+/// bb1:
+/// %tmp1 = tail call i32 @f1()
+/// br label %return
+/// bb2:
+/// %tmp2 = tail call i32 @f2()
+/// br label %return
+/// return:
+/// %retval = phi i32 [ %tmp0, %bb0 ], [ %tmp1, %bb1 ], [ %tmp2, %bb2 ]
+/// ret i32 %retval
+/// @endcode
+///
+/// =>
+///
+/// @code
+/// bb0:
+/// %tmp0 = tail call i32 @f0()
+/// ret i32 %tmp0
+/// bb1:
+/// %tmp1 = tail call i32 @f1()
+/// ret i32 %tmp1
+/// bb2:
+/// %tmp2 = tail call i32 @f2()
+/// ret i32 %tmp2
+/// @endcode
+bool CodeGenPrepare::DupRetToEnableTailCallOpts(BasicBlock *BB) {
+ if (!TLI)
+ return false;
+
+ ReturnInst *RI = dyn_cast<ReturnInst>(BB->getTerminator());
+ if (!RI)
+ return false;
+
+ PHINode *PN = 0;
+ BitCastInst *BCI = 0;
+ Value *V = RI->getReturnValue();
+ if (V) {
+ BCI = dyn_cast<BitCastInst>(V);
+ if (BCI)
+ V = BCI->getOperand(0);
+
+ PN = dyn_cast<PHINode>(V);
+ if (!PN)
+ return false;
+ }
+
+ if (PN && PN->getParent() != BB)
+ return false;
+
+ // It's not safe to eliminate the sign / zero extension of the return value.
+ // See llvm::isInTailCallPosition().
+ const Function *F = BB->getParent();
+ Attribute CallerRetAttr = F->getAttributes().getRetAttributes();
+ if (CallerRetAttr.hasAttribute(Attribute::ZExt) ||
+ CallerRetAttr.hasAttribute(Attribute::SExt))
+ return false;
+
+ // Make sure there are no instructions between the PHI and return, or that the
+ // return is the first instruction in the block.
+ if (PN) {
+ BasicBlock::iterator BI = BB->begin();
+ do { ++BI; } while (isa<DbgInfoIntrinsic>(BI));
+ if (&*BI == BCI)
+ // Also skip over the bitcast.
+ ++BI;
+ if (&*BI != RI)
+ return false;
+ } else {
+ BasicBlock::iterator BI = BB->begin();
+ while (isa<DbgInfoIntrinsic>(BI)) ++BI;
+ if (&*BI != RI)
+ return false;
+ }
+
+ /// Only dup the ReturnInst if the CallInst is likely to be emitted as a tail
+ /// call.
+ SmallVector<CallInst*, 4> TailCalls;
+ if (PN) {
+ for (unsigned I = 0, E = PN->getNumIncomingValues(); I != E; ++I) {
+ CallInst *CI = dyn_cast<CallInst>(PN->getIncomingValue(I));
+ // Make sure the phi value is indeed produced by the tail call.
+ if (CI && CI->hasOneUse() && CI->getParent() == PN->getIncomingBlock(I) &&
+ TLI->mayBeEmittedAsTailCall(CI))
+ TailCalls.push_back(CI);
+ }
+ } else {
+ SmallPtrSet<BasicBlock*, 4> VisitedBBs;
+ for (pred_iterator PI = pred_begin(BB), PE = pred_end(BB); PI != PE; ++PI) {
+ if (!VisitedBBs.insert(*PI))
+ continue;
+
+ BasicBlock::InstListType &InstList = (*PI)->getInstList();
+ BasicBlock::InstListType::reverse_iterator RI = InstList.rbegin();
+ BasicBlock::InstListType::reverse_iterator RE = InstList.rend();
+ do { ++RI; } while (RI != RE && isa<DbgInfoIntrinsic>(&*RI));
+ if (RI == RE)
+ continue;
+
+ CallInst *CI = dyn_cast<CallInst>(&*RI);
+ if (CI && CI->use_empty() && TLI->mayBeEmittedAsTailCall(CI))
+ TailCalls.push_back(CI);
+ }
+ }
+
+ bool Changed = false;
+ for (unsigned i = 0, e = TailCalls.size(); i != e; ++i) {
+ CallInst *CI = TailCalls[i];
+ CallSite CS(CI);
+
+ // Conservatively require the attributes of the call to match those of the
+ // return. Ignore noalias because it doesn't affect the call sequence.
+ Attribute CalleeRetAttr = CS.getAttributes().getRetAttributes();
+ if (AttrBuilder(CalleeRetAttr).
+ removeAttribute(Attribute::NoAlias) !=
+ AttrBuilder(CallerRetAttr).
+ removeAttribute(Attribute::NoAlias))
+ continue;
+
+ // Make sure the call instruction is followed by an unconditional branch to
+ // the return block.
+ BasicBlock *CallBB = CI->getParent();
+ BranchInst *BI = dyn_cast<BranchInst>(CallBB->getTerminator());
+ if (!BI || !BI->isUnconditional() || BI->getSuccessor(0) != BB)
+ continue;
+
+ // Duplicate the return into CallBB.
+ (void)FoldReturnIntoUncondBranch(RI, BB, CallBB);
+ ModifiedDT = Changed = true;
+ ++NumRetsDup;
+ }
+
+ // If we eliminated all predecessors of the block, delete the block now.
+ if (Changed && !BB->hasAddressTaken() && pred_begin(BB) == pred_end(BB))
+ BB->eraseFromParent();
+
+ return Changed;
}
//===----------------------------------------------------------------------===//
/// This method is used to optimize both load/store and inline asms with memory
/// operands.
bool CodeGenPrepare::OptimizeMemoryInst(Instruction *MemoryInst, Value *Addr,
- const Type *AccessTy) {
+ Type *AccessTy) {
Value *Repl = Addr;
-
- // Try to collapse single-value PHI nodes. This is necessary to undo
+
+ // Try to collapse single-value PHI nodes. This is necessary to undo
// unprofitable PRE transformations.
SmallVector<Value*, 8> worklist;
SmallPtrSet<Value*, 16> Visited;
worklist.push_back(Addr);
-
+
// Use a worklist to iteratively look through PHI nodes, and ensure that
// the addressing mode obtained from the non-PHI roots of the graph
// are equivalent.
while (!worklist.empty()) {
Value *V = worklist.back();
worklist.pop_back();
-
+
// Break use-def graph loops.
- if (Visited.count(V)) {
+ if (!Visited.insert(V)) {
Consensus = 0;
break;
}
-
- Visited.insert(V);
-
+
// For a PHI node, push all of its incoming values.
if (PHINode *P = dyn_cast<PHINode>(V)) {
for (unsigned i = 0, e = P->getNumIncomingValues(); i != e; ++i)
worklist.push_back(P->getIncomingValue(i));
continue;
}
-
+
// For non-PHIs, determine the addressing mode being computed.
SmallVector<Instruction*, 16> NewAddrModeInsts;
ExtAddrMode NewAddrMode =
- AddressingModeMatcher::Match(V, AccessTy,MemoryInst,
+ AddressingModeMatcher::Match(V, AccessTy, MemoryInst,
NewAddrModeInsts, *TLI);
// This check is broken into two cases with very similar code to avoid using
}
continue;
}
-
+
Consensus = 0;
break;
}
-
+
// If the addressing mode couldn't be determined, or if multiple different
// ones were determined, bail out now.
if (!Consensus) return false;
-
+
// Check to see if any of the instructions supersumed by this addr mode are
// non-local to I's BB.
bool AnyNonLocal = false;
// Insert this computation right after this user. Since our caller is
// scanning from the top of the BB to the bottom, reuse of the expr are
// guaranteed to happen later.
- BasicBlock::iterator InsertPt = MemoryInst;
+ IRBuilder<> Builder(MemoryInst);
// Now that we determined the addressing expression we want to use and know
// that we have to sink it into this block. Check to see if we have already
DEBUG(dbgs() << "CGP: Reusing nonlocal addrmode: " << AddrMode << " for "
<< *MemoryInst);
if (SunkAddr->getType() != Addr->getType())
- SunkAddr = new BitCastInst(SunkAddr, Addr->getType(), "tmp", InsertPt);
+ SunkAddr = Builder.CreateBitCast(SunkAddr, Addr->getType());
} else {
DEBUG(dbgs() << "CGP: SINKING nonlocal addrmode: " << AddrMode << " for "
<< *MemoryInst);
- const Type *IntPtrTy =
- TLI->getTargetData()->getIntPtrType(AccessTy->getContext());
+ Type *IntPtrTy =
+ TLI->getDataLayout()->getIntPtrType(AccessTy->getContext());
Value *Result = 0;
if (AddrMode.BaseReg) {
Value *V = AddrMode.BaseReg;
if (V->getType()->isPointerTy())
- V = new PtrToIntInst(V, IntPtrTy, "sunkaddr", InsertPt);
+ V = Builder.CreatePtrToInt(V, IntPtrTy, "sunkaddr");
if (V->getType() != IntPtrTy)
- V = CastInst::CreateIntegerCast(V, IntPtrTy, /*isSigned=*/true,
- "sunkaddr", InsertPt);
+ V = Builder.CreateIntCast(V, IntPtrTy, /*isSigned=*/true, "sunkaddr");
Result = V;
}
if (V->getType() == IntPtrTy) {
// done.
} else if (V->getType()->isPointerTy()) {
- V = new PtrToIntInst(V, IntPtrTy, "sunkaddr", InsertPt);
+ V = Builder.CreatePtrToInt(V, IntPtrTy, "sunkaddr");
} else if (cast<IntegerType>(IntPtrTy)->getBitWidth() <
cast<IntegerType>(V->getType())->getBitWidth()) {
- V = new TruncInst(V, IntPtrTy, "sunkaddr", InsertPt);
+ V = Builder.CreateTrunc(V, IntPtrTy, "sunkaddr");
} else {
- V = new SExtInst(V, IntPtrTy, "sunkaddr", InsertPt);
+ V = Builder.CreateSExt(V, IntPtrTy, "sunkaddr");
}
if (AddrMode.Scale != 1)
- V = BinaryOperator::CreateMul(V, ConstantInt::get(IntPtrTy,
- AddrMode.Scale),
- "sunkaddr", InsertPt);
+ V = Builder.CreateMul(V, ConstantInt::get(IntPtrTy, AddrMode.Scale),
+ "sunkaddr");
if (Result)
- Result = BinaryOperator::CreateAdd(Result, V, "sunkaddr", InsertPt);
+ Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
// Add in the BaseGV if present.
if (AddrMode.BaseGV) {
- Value *V = new PtrToIntInst(AddrMode.BaseGV, IntPtrTy, "sunkaddr",
- InsertPt);
+ Value *V = Builder.CreatePtrToInt(AddrMode.BaseGV, IntPtrTy, "sunkaddr");
if (Result)
- Result = BinaryOperator::CreateAdd(Result, V, "sunkaddr", InsertPt);
+ Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
if (AddrMode.BaseOffs) {
Value *V = ConstantInt::get(IntPtrTy, AddrMode.BaseOffs);
if (Result)
- Result = BinaryOperator::CreateAdd(Result, V, "sunkaddr", InsertPt);
+ Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
if (Result == 0)
SunkAddr = Constant::getNullValue(Addr->getType());
else
- SunkAddr = new IntToPtrInst(Result, Addr->getType(), "sunkaddr",InsertPt);
+ SunkAddr = Builder.CreateIntToPtr(Result, Addr->getType(), "sunkaddr");
}
MemoryInst->replaceUsesOfWith(Repl, SunkAddr);
+ // If we have no uses, recursively delete the value and all dead instructions
+ // using it.
if (Repl->use_empty()) {
- RecursivelyDeleteTriviallyDeadInstructions(Repl);
- // This address is now available for reassignment, so erase the table entry;
- // we don't want to match some completely different instruction.
- SunkAddrs[Addr] = 0;
+ // This can cause recursive deletion, which can invalidate our iterator.
+ // Use a WeakVH to hold onto it in case this happens.
+ WeakVH IterHandle(CurInstIterator);
+ BasicBlock *BB = CurInstIterator->getParent();
+
+ RecursivelyDeleteTriviallyDeadInstructions(Repl, TLInfo);
+
+ if (IterHandle != CurInstIterator) {
+ // If the iterator instruction was recursively deleted, start over at the
+ // start of the block.
+ CurInstIterator = BB->begin();
+ SunkAddrs.clear();
+ } else {
+ // This address is now available for reassignment, so erase the table
+ // entry; we don't want to match some completely different instruction.
+ SunkAddrs[Addr] = 0;
+ }
}
++NumMemoryInsts;
return true;
bool CodeGenPrepare::OptimizeInlineAsmInst(CallInst *CS) {
bool MadeChange = false;
- TargetLowering::AsmOperandInfoVector
+ TargetLowering::AsmOperandInfoVector
TargetConstraints = TLI->ParseConstraints(CS);
unsigned ArgNo = 0;
for (unsigned i = 0, e = TargetConstraints.size(); i != e; ++i) {
TargetLowering::AsmOperandInfo &OpInfo = TargetConstraints[i];
-
+
// Compute the constraint code and ConstraintType to use.
TLI->ComputeConstraintToUse(OpInfo, SDValue());
Instruction *&InsertedTrunc = InsertedTruncs[UserBB];
if (!InsertedTrunc) {
- BasicBlock::iterator InsertPt = UserBB->getFirstNonPHI();
-
+ BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
InsertedTrunc = new TruncInst(I, Src->getType(), "", InsertPt);
}
return MadeChange;
}
+/// isFormingBranchFromSelectProfitable - Returns true if a SelectInst should be
+/// turned into an explicit branch.
+static bool isFormingBranchFromSelectProfitable(SelectInst *SI) {
+ // FIXME: This should use the same heuristics as IfConversion to determine
+ // whether a select is better represented as a branch. This requires that
+ // branch probability metadata is preserved for the select, which is not the
+ // case currently.
+
+ CmpInst *Cmp = dyn_cast<CmpInst>(SI->getCondition());
+
+ // If the branch is predicted right, an out of order CPU can avoid blocking on
+ // the compare. Emit cmovs on compares with a memory operand as branches to
+ // avoid stalls on the load from memory. If the compare has more than one use
+ // there's probably another cmov or setcc around so it's not worth emitting a
+ // branch.
+ if (!Cmp)
+ return false;
+
+ Value *CmpOp0 = Cmp->getOperand(0);
+ Value *CmpOp1 = Cmp->getOperand(1);
+
+ // We check that the memory operand has one use to avoid uses of the loaded
+ // value directly after the compare, making branches unprofitable.
+ return Cmp->hasOneUse() &&
+ ((isa<LoadInst>(CmpOp0) && CmpOp0->hasOneUse()) ||
+ (isa<LoadInst>(CmpOp1) && CmpOp1->hasOneUse()));
+}
+
+
+/// If we have a SelectInst that will likely profit from branch prediction,
+/// turn it into a branch.
+bool CodeGenPrepare::OptimizeSelectInst(SelectInst *SI) {
+ bool VectorCond = !SI->getCondition()->getType()->isIntegerTy(1);
+
+ // Can we convert the 'select' to CF ?
+ if (DisableSelectToBranch || OptSize || !TLI || VectorCond)
+ return false;
+
+ TargetLowering::SelectSupportKind SelectKind;
+ if (VectorCond)
+ SelectKind = TargetLowering::VectorMaskSelect;
+ else if (SI->getType()->isVectorTy())
+ SelectKind = TargetLowering::ScalarCondVectorVal;
+ else
+ SelectKind = TargetLowering::ScalarValSelect;
+
+ // Do we have efficient codegen support for this kind of 'selects' ?
+ if (TLI->isSelectSupported(SelectKind)) {
+ // We have efficient codegen support for the select instruction.
+ // Check if it is profitable to keep this 'select'.
+ if (!TLI->isPredictableSelectExpensive() ||
+ !isFormingBranchFromSelectProfitable(SI))
+ return false;
+ }
+
+ ModifiedDT = true;
+
+ // First, we split the block containing the select into 2 blocks.
+ BasicBlock *StartBlock = SI->getParent();
+ BasicBlock::iterator SplitPt = ++(BasicBlock::iterator(SI));
+ BasicBlock *NextBlock = StartBlock->splitBasicBlock(SplitPt, "select.end");
+
+ // Create a new block serving as the landing pad for the branch.
+ BasicBlock *SmallBlock = BasicBlock::Create(SI->getContext(), "select.mid",
+ NextBlock->getParent(), NextBlock);
+
+ // Move the unconditional branch from the block with the select in it into our
+ // landing pad block.
+ StartBlock->getTerminator()->eraseFromParent();
+ BranchInst::Create(NextBlock, SmallBlock);
+
+ // Insert the real conditional branch based on the original condition.
+ BranchInst::Create(NextBlock, SmallBlock, SI->getCondition(), SI);
+
+ // The select itself is replaced with a PHI Node.
+ PHINode *PN = PHINode::Create(SI->getType(), 2, "", NextBlock->begin());
+ PN->takeName(SI);
+ PN->addIncoming(SI->getTrueValue(), StartBlock);
+ PN->addIncoming(SI->getFalseValue(), SmallBlock);
+ SI->replaceAllUsesWith(PN);
+ SI->eraseFromParent();
+
+ // Instruct OptimizeBlock to skip to the next block.
+ CurInstIterator = StartBlock->end();
+ ++NumSelectsExpanded;
+ return true;
+}
+
bool CodeGenPrepare::OptimizeInst(Instruction *I) {
if (PHINode *P = dyn_cast<PHINode>(I)) {
// It is possible for very late stage optimizations (such as SimplifyCFG)
}
return false;
}
-
+
if (CastInst *CI = dyn_cast<CastInst>(I)) {
// If the source of the cast is a constant, then this should have
// already been constant folded. The only reason NOT to constant fold
}
return false;
}
-
+
if (CmpInst *CI = dyn_cast<CmpInst>(I))
return OptimizeCmpExpression(CI);
-
+
if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
if (TLI)
return OptimizeMemoryInst(I, I->getOperand(0), LI->getType());
return false;
}
-
+
if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
if (TLI)
return OptimizeMemoryInst(I, SI->getOperand(1),
SI->getOperand(0)->getType());
return false;
}
-
+
if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(I)) {
if (GEPI->hasAllZeroIndices()) {
/// The GEP operand must be a pointer, so must its result -> BitCast
}
return false;
}
-
+
if (CallInst *CI = dyn_cast<CallInst>(I))
return OptimizeCallInst(CI);
+ if (SelectInst *SI = dyn_cast<SelectInst>(I))
+ return OptimizeSelectInst(SI);
+
return false;
}
bool MadeChange = false;
CurInstIterator = BB.begin();
- for (BasicBlock::iterator E = BB.end(); CurInstIterator != E; )
+ while (CurInstIterator != BB.end())
MadeChange |= OptimizeInst(CurInstIterator++);
+ MadeChange |= DupRetToEnableTailCallOpts(&BB);
+
+ return MadeChange;
+}
+
+// llvm.dbg.value is far away from the value then iSel may not be able
+// handle it properly. iSel will drop llvm.dbg.value if it can not
+// find a node corresponding to the value.
+bool CodeGenPrepare::PlaceDbgValues(Function &F) {
+ bool MadeChange = false;
+ for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) {
+ Instruction *PrevNonDbgInst = NULL;
+ for (BasicBlock::iterator BI = I->begin(), BE = I->end(); BI != BE;) {
+ Instruction *Insn = BI; ++BI;
+ DbgValueInst *DVI = dyn_cast<DbgValueInst>(Insn);
+ if (!DVI) {
+ PrevNonDbgInst = Insn;
+ continue;
+ }
+
+ Instruction *VI = dyn_cast_or_null<Instruction>(DVI->getValue());
+ if (VI && VI != PrevNonDbgInst && !VI->isTerminator()) {
+ DEBUG(dbgs() << "Moving Debug Value before :\n" << *DVI << ' ' << *VI);
+ DVI->removeFromParent();
+ if (isa<PHINode>(VI))
+ DVI->insertBefore(VI->getParent()->getFirstInsertionPt());
+ else
+ DVI->insertAfter(VI);
+ MadeChange = true;
+ ++NumDbgValueMoved;
+ }
+ }
+ }
return MadeChange;
}
projects / oota-llvm.git / blobdiff