static bool runImpl(CallGraphSCC &SCC, CallGraph &CG) {
SmallPtrSet<CallGraphNode *, 8> SCCNodes;
bool MadeChange = false;
// Fill SCCNodes with the elements of the SCC. Used for quickly
// looking up whether a given CallGraphNode is in this SCC.
for (CallGraphNode *I : SCC)
SCCNodes.insert(I);
// First pass, scan all of the functions in the SCC, simplifying them
// according to what we know.
for (CallGraphNode *I : SCC)
if (Function *F = I->getFunction())
MadeChange |= SimplifyFunction(F, CG);
// Next, check to see if any callees might throw or if there are any external
// functions in this SCC: if so, we cannot prune any functions in this SCC.
// Definitions that are weak and not declared non-throwing might be
// overridden at linktime with something that throws, so assume that.
// If this SCC includes the unwind instruction, we KNOW it throws, so
// obviously the SCC might throw.
//
bool SCCMightUnwind = false, SCCMightReturn = false;
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end();
(!SCCMightUnwind || !SCCMightReturn) && I != E; ++I) {
Function *F = (*I)->getFunction();
if (!F) {
SCCMightUnwind = true;
SCCMightReturn = true;
} else if (!F->hasExactDefinition()) {
SCCMightUnwind |= !F->doesNotThrow();
SCCMightReturn |= !F->doesNotReturn();
} else {
bool CheckUnwind = !SCCMightUnwind && !F->doesNotThrow();
bool CheckReturn = !SCCMightReturn && !F->doesNotReturn();
// Determine if we should scan for InlineAsm in a naked function as it
// is the only way to return without a ReturnInst. Only do this for
// no-inline functions as functions which may be inlined cannot
// meaningfully return via assembly.
bool CheckReturnViaAsm = CheckReturn &&
F->hasFnAttribute(Attribute::Naked) &&
F->hasFnAttribute(Attribute::NoInline);
if (!CheckUnwind && !CheckReturn)
continue;
for (const BasicBlock &BB : *F) {
const TerminatorInst *TI = BB.getTerminator();
if (CheckUnwind && TI->mayThrow()) {
SCCMightUnwind = true;
} else if (CheckReturn && isa<ReturnInst>(TI)) {
SCCMightReturn = true;
}
for (const Instruction &I : BB) {
if ((!CheckUnwind || SCCMightUnwind) &&
(!CheckReturnViaAsm || SCCMightReturn))
break;
// Check to see if this function performs an unwind or calls an
// unwinding function.
if (CheckUnwind && !SCCMightUnwind && I.mayThrow()) {
bool InstMightUnwind = true;
if (const auto *CI = dyn_cast<CallInst>(&I)) {
if (Function *Callee = CI->getCalledFunction()) {
CallGraphNode *CalleeNode = CG[Callee];
// If the callee is outside our current SCC then we may throw
// because it might. If it is inside, do nothing.
if (SCCNodes.count(CalleeNode) > 0)
InstMightUnwind = false;
}
}
SCCMightUnwind |= InstMightUnwind;
}
if (CheckReturnViaAsm && !SCCMightReturn)
if (auto ICS = ImmutableCallSite(&I))
if (const auto *IA = dyn_cast<InlineAsm>(ICS.getCalledValue()))
if (IA->hasSideEffects())
SCCMightReturn = true;
}
if (SCCMightUnwind && SCCMightReturn)
break;
}
}
}
// If the SCC doesn't unwind or doesn't throw, note this fact.
if (!SCCMightUnwind || !SCCMightReturn)
for (CallGraphNode *I : SCC) {
Function *F = I->getFunction();
if (!SCCMightUnwind && !F->hasFnAttribute(Attribute::NoUnwind)) {
F->addFnAttr(Attribute::NoUnwind);
MadeChange = true;
}
if (!SCCMightReturn && !F->hasFnAttribute(Attribute::NoReturn)) {
F->addFnAttr(Attribute::NoReturn);
MadeChange = true;
}
}
for (CallGraphNode *I : SCC) {
// Convert any invoke instructions to non-throwing functions in this node
// into call instructions with a branch. This makes the exception blocks
// dead.
if (Function *F = I->getFunction())
MadeChange |= SimplifyFunction(F, CG);
}
return MadeChange;
}
/// processImplicitDefs - Process IMPLICIT_DEF instructions and make sure
/// there is one implicit_def for each use. Add isUndef marker to
/// implicit_def defs and their uses.
bool ProcessImplicitDefs::runOnMachineFunction(MachineFunction &fn) {
DEBUG(dbgs() << "********** PROCESS IMPLICIT DEFS **********\n"
<< "********** Function: "
<< ((Value*)fn.getFunction())->getName() << '\n');
bool Changed = false;
TII = fn.getTarget().getInstrInfo();
TRI = fn.getTarget().getRegisterInfo();
MRI = &fn.getRegInfo();
LV = &getAnalysis<LiveVariables>();
SmallSet<unsigned, 8> ImpDefRegs;
SmallVector<MachineInstr*, 8> ImpDefMIs;
SmallVector<MachineInstr*, 4> RUses;
SmallPtrSet<MachineBasicBlock*,16> Visited;
SmallPtrSet<MachineInstr*, 8> ModInsts;
MachineBasicBlock *Entry = fn.begin();
for (df_ext_iterator<MachineBasicBlock*, SmallPtrSet<MachineBasicBlock*,16> >
DFI = df_ext_begin(Entry, Visited), E = df_ext_end(Entry, Visited);
DFI != E; ++DFI) {
MachineBasicBlock *MBB = *DFI;
for (MachineBasicBlock::iterator I = MBB->begin(), E = MBB->end();
I != E; ) {
MachineInstr *MI = &*I;
++I;
if (MI->isImplicitDef()) {
ImpDefMIs.push_back(MI);
// Is this a sub-register read-modify-write?
if (MI->getOperand(0).readsReg())
continue;
unsigned Reg = MI->getOperand(0).getReg();
ImpDefRegs.insert(Reg);
if (TargetRegisterInfo::isPhysicalRegister(Reg)) {
for (const unsigned *SS = TRI->getSubRegisters(Reg); *SS; ++SS)
ImpDefRegs.insert(*SS);
}
continue;
}
// Eliminate %reg1032:sub<def> = COPY undef.
if (MI->isCopy() && MI->getOperand(0).readsReg()) {
MachineOperand &MO = MI->getOperand(1);
if (MO.isUndef() || ImpDefRegs.count(MO.getReg())) {
if (MO.isKill()) {
LiveVariables::VarInfo& vi = LV->getVarInfo(MO.getReg());
vi.removeKill(MI);
}
unsigned Reg = MI->getOperand(0).getReg();
MI->eraseFromParent();
Changed = true;
// A REG_SEQUENCE may have been expanded into partial definitions.
// If this was the last one, mark Reg as implicitly defined.
if (TargetRegisterInfo::isVirtualRegister(Reg) && MRI->def_empty(Reg))
ImpDefRegs.insert(Reg);
continue;
}
}
bool ChangedToImpDef = false;
for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
MachineOperand& MO = MI->getOperand(i);
if (!MO.isReg() || !MO.readsReg())
continue;
unsigned Reg = MO.getReg();
if (!Reg)
continue;
if (!ImpDefRegs.count(Reg))
continue;
// Use is a copy, just turn it into an implicit_def.
if (CanTurnIntoImplicitDef(MI, Reg, i, ImpDefRegs)) {
bool isKill = MO.isKill();
MI->setDesc(TII->get(TargetOpcode::IMPLICIT_DEF));
for (int j = MI->getNumOperands() - 1, ee = 0; j > ee; --j)
MI->RemoveOperand(j);
if (isKill) {
ImpDefRegs.erase(Reg);
LiveVariables::VarInfo& vi = LV->getVarInfo(Reg);
vi.removeKill(MI);
}
ChangedToImpDef = true;
Changed = true;
break;
}
Changed = true;
MO.setIsUndef();
// This is a partial register redef of an implicit def.
// Make sure the whole register is defined by the instruction.
if (MO.isDef()) {
MI->addRegisterDefined(Reg);
continue;
}
if (MO.isKill() || MI->isRegTiedToDefOperand(i)) {
// Make sure other reads of Reg are also marked <undef>.
for (unsigned j = i+1; j != e; ++j) {
MachineOperand &MOJ = MI->getOperand(j);
if (MOJ.isReg() && MOJ.getReg() == Reg && MOJ.readsReg())
MOJ.setIsUndef();
}
ImpDefRegs.erase(Reg);
}
}
if (ChangedToImpDef) {
// Backtrack to process this new implicit_def.
--I;
} else {
for (unsigned i = 0; i != MI->getNumOperands(); ++i) {
MachineOperand& MO = MI->getOperand(i);
if (!MO.isReg() || !MO.isDef())
continue;
ImpDefRegs.erase(MO.getReg());
}
}
}
// Any outstanding liveout implicit_def's?
for (unsigned i = 0, e = ImpDefMIs.size(); i != e; ++i) {
MachineInstr *MI = ImpDefMIs[i];
unsigned Reg = MI->getOperand(0).getReg();
if (TargetRegisterInfo::isPhysicalRegister(Reg) ||
!ImpDefRegs.count(Reg)) {
// Delete all "local" implicit_def's. That include those which define
// physical registers since they cannot be liveout.
MI->eraseFromParent();
Changed = true;
continue;
}
// If there are multiple defs of the same register and at least one
// is not an implicit_def, do not insert implicit_def's before the
// uses.
bool Skip = false;
SmallVector<MachineInstr*, 4> DeadImpDefs;
for (MachineRegisterInfo::def_iterator DI = MRI->def_begin(Reg),
DE = MRI->def_end(); DI != DE; ++DI) {
MachineInstr *DeadImpDef = &*DI;
if (!DeadImpDef->isImplicitDef()) {
Skip = true;
break;
}
DeadImpDefs.push_back(DeadImpDef);
}
if (Skip)
continue;
// The only implicit_def which we want to keep are those that are live
// out of its block.
for (unsigned j = 0, ee = DeadImpDefs.size(); j != ee; ++j)
DeadImpDefs[j]->eraseFromParent();
Changed = true;
// Process each use instruction once.
for (MachineRegisterInfo::use_iterator UI = MRI->use_begin(Reg),
UE = MRI->use_end(); UI != UE; ++UI) {
if (UI.getOperand().isUndef())
continue;
MachineInstr *RMI = &*UI;
if (ModInsts.insert(RMI))
RUses.push_back(RMI);
}
for (unsigned i = 0, e = RUses.size(); i != e; ++i) {
MachineInstr *RMI = RUses[i];
// Turn a copy use into an implicit_def.
if (isUndefCopy(RMI, Reg, ImpDefRegs)) {
RMI->setDesc(TII->get(TargetOpcode::IMPLICIT_DEF));
bool isKill = false;
SmallVector<unsigned, 4> Ops;
for (unsigned j = 0, ee = RMI->getNumOperands(); j != ee; ++j) {
MachineOperand &RRMO = RMI->getOperand(j);
if (RRMO.isReg() && RRMO.getReg() == Reg) {
Ops.push_back(j);
if (RRMO.isKill())
isKill = true;
}
}
// Leave the other operands along.
for (unsigned j = 0, ee = Ops.size(); j != ee; ++j) {
unsigned OpIdx = Ops[j];
RMI->RemoveOperand(OpIdx-j);
}
// Update LiveVariables varinfo if the instruction is a kill.
if (isKill) {
LiveVariables::VarInfo& vi = LV->getVarInfo(Reg);
vi.removeKill(RMI);
}
continue;
}
// Replace Reg with a new vreg that's marked implicit.
const TargetRegisterClass* RC = MRI->getRegClass(Reg);
unsigned NewVReg = MRI->createVirtualRegister(RC);
bool isKill = true;
for (unsigned j = 0, ee = RMI->getNumOperands(); j != ee; ++j) {
MachineOperand &RRMO = RMI->getOperand(j);
if (RRMO.isReg() && RRMO.getReg() == Reg) {
RRMO.setReg(NewVReg);
RRMO.setIsUndef();
if (isKill) {
// Only the first operand of NewVReg is marked kill.
RRMO.setIsKill();
isKill = false;
}
}
}
}
RUses.clear();
ModInsts.clear();
}
ImpDefRegs.clear();
ImpDefMIs.clear();
}
return Changed;
}
/// At this point, we're committed to promoting the alloca using IDF's, and the
/// standard SSA construction algorithm. Determine which blocks need phi nodes
/// and see if we can optimize out some work by avoiding insertion of dead phi
/// nodes.
void PromoteMem2Reg::DetermineInsertionPoint(AllocaInst *AI, unsigned AllocaNum,
AllocaInfo &Info) {
// Unique the set of defining blocks for efficient lookup.
SmallPtrSet<BasicBlock *, 32> DefBlocks;
DefBlocks.insert(Info.DefiningBlocks.begin(), Info.DefiningBlocks.end());
// Determine which blocks the value is live in. These are blocks which lead
// to uses.
SmallPtrSet<BasicBlock *, 32> LiveInBlocks;
ComputeLiveInBlocks(AI, Info, DefBlocks, LiveInBlocks);
// Use a priority queue keyed on dominator tree level so that inserted nodes
// are handled from the bottom of the dominator tree upwards.
typedef std::pair<DomTreeNode *, unsigned> DomTreeNodePair;
typedef std::priority_queue<DomTreeNodePair, SmallVector<DomTreeNodePair, 32>,
less_second> IDFPriorityQueue;
IDFPriorityQueue PQ;
for (SmallPtrSet<BasicBlock *, 32>::const_iterator I = DefBlocks.begin(),
E = DefBlocks.end();
I != E; ++I) {
if (DomTreeNode *Node = DT.getNode(*I))
PQ.push(std::make_pair(Node, DomLevels[Node]));
}
SmallVector<std::pair<unsigned, BasicBlock *>, 32> DFBlocks;
SmallPtrSet<DomTreeNode *, 32> Visited;
SmallVector<DomTreeNode *, 32> Worklist;
while (!PQ.empty()) {
DomTreeNodePair RootPair = PQ.top();
PQ.pop();
DomTreeNode *Root = RootPair.first;
unsigned RootLevel = RootPair.second;
// Walk all dominator tree children of Root, inspecting their CFG edges with
// targets elsewhere on the dominator tree. Only targets whose level is at
// most Root's level are added to the iterated dominance frontier of the
// definition set.
Worklist.clear();
Worklist.push_back(Root);
while (!Worklist.empty()) {
DomTreeNode *Node = Worklist.pop_back_val();
BasicBlock *BB = Node->getBlock();
for (succ_iterator SI = succ_begin(BB), SE = succ_end(BB); SI != SE;
++SI) {
DomTreeNode *SuccNode = DT.getNode(*SI);
// Quickly skip all CFG edges that are also dominator tree edges instead
// of catching them below.
if (SuccNode->getIDom() == Node)
continue;
unsigned SuccLevel = DomLevels[SuccNode];
if (SuccLevel > RootLevel)
continue;
if (!Visited.insert(SuccNode))
continue;
BasicBlock *SuccBB = SuccNode->getBlock();
if (!LiveInBlocks.count(SuccBB))
continue;
DFBlocks.push_back(std::make_pair(BBNumbers[SuccBB], SuccBB));
if (!DefBlocks.count(SuccBB))
PQ.push(std::make_pair(SuccNode, SuccLevel));
}
for (DomTreeNode::iterator CI = Node->begin(), CE = Node->end(); CI != CE;
++CI) {
if (!Visited.count(*CI))
Worklist.push_back(*CI);
}
}
}
if (DFBlocks.size() > 1)
std::sort(DFBlocks.begin(), DFBlocks.end());
unsigned CurrentVersion = 0;
for (unsigned i = 0, e = DFBlocks.size(); i != e; ++i)
QueuePhiNode(DFBlocks[i].second, AllocaNum, CurrentVersion);
}
/// InlineCallIfPossible - If it is possible to inline the specified call site,
/// do so and update the CallGraph for this operation.
///
/// This function also does some basic book-keeping to update the IR. The
/// InlinedArrayAllocas map keeps track of any allocas that are already
/// available from other functions inlined into the caller. If we are able to
/// inline this call site we attempt to reuse already available allocas or add
/// any new allocas to the set if not possible.
static bool InlineCallIfPossible(CallSite CS, InlineFunctionInfo &IFI,
InlinedArrayAllocasTy &InlinedArrayAllocas,
int InlineHistory, bool InsertLifetime) {
Function *Callee = CS.getCalledFunction();
Function *Caller = CS.getCaller();
// Try to inline the function. Get the list of static allocas that were
// inlined.
if (!InlineFunction(CS, IFI, InsertLifetime))
return false;
// If the inlined function had a higher stack protection level than the
// calling function, then bump up the caller's stack protection level.
if (Callee->hasFnAttr(Attribute::StackProtectReq))
Caller->addFnAttr(Attribute::StackProtectReq);
else if (Callee->hasFnAttr(Attribute::StackProtect) &&
!Caller->hasFnAttr(Attribute::StackProtectReq))
Caller->addFnAttr(Attribute::StackProtect);
// Look at all of the allocas that we inlined through this call site. If we
// have already inlined other allocas through other calls into this function,
// then we know that they have disjoint lifetimes and that we can merge them.
//
// There are many heuristics possible for merging these allocas, and the
// different options have different tradeoffs. One thing that we *really*
// don't want to hurt is SRoA: once inlining happens, often allocas are no
// longer address taken and so they can be promoted.
//
// Our "solution" for that is to only merge allocas whose outermost type is an
// array type. These are usually not promoted because someone is using a
// variable index into them. These are also often the most important ones to
// merge.
//
// A better solution would be to have real memory lifetime markers in the IR
// and not have the inliner do any merging of allocas at all. This would
// allow the backend to do proper stack slot coloring of all allocas that
// *actually make it to the backend*, which is really what we want.
//
// Because we don't have this information, we do this simple and useful hack.
//
SmallPtrSet<AllocaInst*, 16> UsedAllocas;
// When processing our SCC, check to see if CS was inlined from some other
// call site. For example, if we're processing "A" in this code:
// A() { B() }
// B() { x = alloca ... C() }
// C() { y = alloca ... }
// Assume that C was not inlined into B initially, and so we're processing A
// and decide to inline B into A. Doing this makes an alloca available for
// reuse and makes a callsite (C) available for inlining. When we process
// the C call site we don't want to do any alloca merging between X and Y
// because their scopes are not disjoint. We could make this smarter by
// keeping track of the inline history for each alloca in the
// InlinedArrayAllocas but this isn't likely to be a significant win.
if (InlineHistory != -1) // Only do merging for top-level call sites in SCC.
return true;
// Loop over all the allocas we have so far and see if they can be merged with
// a previously inlined alloca. If not, remember that we had it.
for (unsigned AllocaNo = 0, e = IFI.StaticAllocas.size();
AllocaNo != e; ++AllocaNo) {
AllocaInst *AI = IFI.StaticAllocas[AllocaNo];
// Don't bother trying to merge array allocations (they will usually be
// canonicalized to be an allocation *of* an array), or allocations whose
// type is not itself an array (because we're afraid of pessimizing SRoA).
ArrayType *ATy = dyn_cast<ArrayType>(AI->getAllocatedType());
if (ATy == 0 || AI->isArrayAllocation())
continue;
// Get the list of all available allocas for this array type.
std::vector<AllocaInst*> &AllocasForType = InlinedArrayAllocas[ATy];
// Loop over the allocas in AllocasForType to see if we can reuse one. Note
// that we have to be careful not to reuse the same "available" alloca for
// multiple different allocas that we just inlined, we use the 'UsedAllocas'
// set to keep track of which "available" allocas are being used by this
// function. Also, AllocasForType can be empty of course!
bool MergedAwayAlloca = false;
for (unsigned i = 0, e = AllocasForType.size(); i != e; ++i) {
AllocaInst *AvailableAlloca = AllocasForType[i];
// The available alloca has to be in the right function, not in some other
// function in this SCC.
if (AvailableAlloca->getParent() != AI->getParent())
continue;
// If the inlined function already uses this alloca then we can't reuse
// it.
if (!UsedAllocas.insert(AvailableAlloca))
continue;
// Otherwise, we *can* reuse it, RAUW AI into AvailableAlloca and declare
// success!
DEBUG(dbgs() << " ***MERGED ALLOCA: " << *AI << "\n\t\tINTO: "
<< *AvailableAlloca << '\n');
AI->replaceAllUsesWith(AvailableAlloca);
AI->eraseFromParent();
MergedAwayAlloca = true;
++NumMergedAllocas;
IFI.StaticAllocas[AllocaNo] = 0;
break;
}
// If we already nuked the alloca, we're done with it.
if (MergedAwayAlloca)
continue;
// If we were unable to merge away the alloca either because there are no
// allocas of the right type available or because we reused them all
// already, remember that this alloca came from an inlined function and mark
// it used so we don't reuse it for other allocas from this inline
// operation.
AllocasForType.push_back(AI);
UsedAllocas.insert(AI);
}
return true;
}
bool PruneEH::runOnSCC(CallGraphSCC &SCC) {
SmallPtrSet<CallGraphNode *, 8> SCCNodes;
CallGraph &CG = getAnalysis<CallGraph>();
bool MadeChange = false;
// Fill SCCNodes with the elements of the SCC. Used for quickly
// looking up whether a given CallGraphNode is in this SCC.
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I)
SCCNodes.insert(*I);
// First pass, scan all of the functions in the SCC, simplifying them
// according to what we know.
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I)
if (Function *F = (*I)->getFunction())
MadeChange |= SimplifyFunction(F);
// Next, check to see if any callees might throw or if there are any external
// functions in this SCC: if so, we cannot prune any functions in this SCC.
// Definitions that are weak and not declared non-throwing might be
// overridden at linktime with something that throws, so assume that.
// If this SCC includes the unwind instruction, we KNOW it throws, so
// obviously the SCC might throw.
//
bool SCCMightUnwind = false, SCCMightReturn = false;
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end();
(!SCCMightUnwind || !SCCMightReturn) && I != E; ++I) {
Function *F = (*I)->getFunction();
if (F == 0) {
SCCMightUnwind = true;
SCCMightReturn = true;
} else if (F->isDeclaration() || F->mayBeOverridden()) {
SCCMightUnwind |= !F->doesNotThrow();
SCCMightReturn |= !F->doesNotReturn();
} else {
bool CheckUnwind = !SCCMightUnwind && !F->doesNotThrow();
bool CheckReturn = !SCCMightReturn && !F->doesNotReturn();
if (!CheckUnwind && !CheckReturn)
continue;
// Check to see if this function performs an unwind or calls an
// unwinding function.
for (Function::iterator BB = F->begin(), E = F->end(); BB != E; ++BB) {
if (CheckUnwind && isa<ResumeInst>(BB->getTerminator())) {
// Uses unwind / resume!
SCCMightUnwind = true;
} else if (CheckReturn && isa<ReturnInst>(BB->getTerminator())) {
SCCMightReturn = true;
}
// Invoke instructions don't allow unwinding to continue, so we are
// only interested in call instructions.
if (CheckUnwind && !SCCMightUnwind)
for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I)
if (CallInst *CI = dyn_cast<CallInst>(I)) {
if (CI->doesNotThrow()) {
// This call cannot throw.
} else if (Function *Callee = CI->getCalledFunction()) {
CallGraphNode *CalleeNode = CG[Callee];
// If the callee is outside our current SCC then we may
// throw because it might.
if (!SCCNodes.count(CalleeNode)) {
SCCMightUnwind = true;
break;
}
} else {
// Indirect call, it might throw.
SCCMightUnwind = true;
break;
}
}
if (SCCMightUnwind && SCCMightReturn) break;
}
}
}
// If the SCC doesn't unwind or doesn't throw, note this fact.
if (!SCCMightUnwind || !SCCMightReturn)
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I) {
AttrBuilder NewAttributes;
if (!SCCMightUnwind)
NewAttributes.addAttribute(Attribute::NoUnwind);
if (!SCCMightReturn)
NewAttributes.addAttribute(Attribute::NoReturn);
Function *F = (*I)->getFunction();
const AttributeSet &PAL = F->getAttributes();
const AttributeSet &NPAL =
PAL.addFnAttributes(F->getContext(),
AttributeSet::get(F->getContext(),
AttributeSet::FunctionIndex,
NewAttributes));
if (PAL != NPAL) {
MadeChange = true;
F->setAttributes(NPAL);
}
}
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I) {
// Convert any invoke instructions to non-throwing functions in this node
// into call instructions with a branch. This makes the exception blocks
// dead.
if (Function *F = (*I)->getFunction())
MadeChange |= SimplifyFunction(F);
}
return MadeChange;
}
/// OptimizeMemoryInst - Load and Store Instructions often have
/// addressing modes that can do significant amounts of computation. As such,
/// instruction selection will try to get the load or store to do as much
/// computation as possible for the program. The problem is that isel can only
/// see within a single block. As such, we sink as much legal addressing mode
/// stuff into the block as possible.
///
/// This method is used to optimize both load/store and inline asms with memory
/// operands.
bool CodeGenPrepare::OptimizeMemoryInst(Instruction *MemoryInst, Value *Addr,
Type *AccessTy) {
Value *Repl = Addr;
// 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.
Value *Consensus = 0;
unsigned NumUsesConsensus = 0;
bool IsNumUsesConsensusValid = false;
SmallVector<Instruction*, 16> AddrModeInsts;
ExtAddrMode AddrMode;
while (!worklist.empty()) {
Value *V = worklist.back();
worklist.pop_back();
// Break use-def graph loops.
if (!Visited.insert(V)) {
Consensus = 0;
break;
}
// 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,
NewAddrModeInsts, *TLI);
// This check is broken into two cases with very similar code to avoid using
// getNumUses() as much as possible. Some values have a lot of uses, so
// calling getNumUses() unconditionally caused a significant compile-time
// regression.
if (!Consensus) {
Consensus = V;
AddrMode = NewAddrMode;
AddrModeInsts = NewAddrModeInsts;
continue;
} else if (NewAddrMode == AddrMode) {
if (!IsNumUsesConsensusValid) {
NumUsesConsensus = Consensus->getNumUses();
IsNumUsesConsensusValid = true;
}
// Ensure that the obtained addressing mode is equivalent to that obtained
// for all other roots of the PHI traversal. Also, when choosing one
// such root as representative, select the one with the most uses in order
// to keep the cost modeling heuristics in AddressingModeMatcher
// applicable.
unsigned NumUses = V->getNumUses();
if (NumUses > NumUsesConsensus) {
Consensus = V;
NumUsesConsensus = NumUses;
AddrModeInsts = NewAddrModeInsts;
}
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;
for (unsigned i = 0, e = AddrModeInsts.size(); i != e; ++i) {
if (IsNonLocalValue(AddrModeInsts[i], MemoryInst->getParent())) {
AnyNonLocal = true;
break;
}
}
// If all the instructions matched are already in this BB, don't do anything.
if (!AnyNonLocal) {
DEBUG(dbgs() << "CGP: Found local addrmode: " << AddrMode << "\n");
return 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.
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
// done this for some other load/store instr in this block. If so, reuse the
// computation.
Value *&SunkAddr = SunkAddrs[Addr];
if (SunkAddr) {
DEBUG(dbgs() << "CGP: Reusing nonlocal addrmode: " << AddrMode << " for "
<< *MemoryInst);
if (SunkAddr->getType() != Addr->getType())
SunkAddr = Builder.CreateBitCast(SunkAddr, Addr->getType());
} else {
DEBUG(dbgs() << "CGP: SINKING nonlocal addrmode: " << AddrMode << " for "
<< *MemoryInst);
Type *IntPtrTy =
TLI->getTargetData()->getIntPtrType(AccessTy->getContext());
Value *Result = 0;
// Start with the base register. Do this first so that subsequent address
// matching finds it last, which will prevent it from trying to match it
// as the scaled value in case it happens to be a mul. That would be
// problematic if we've sunk a different mul for the scale, because then
// we'd end up sinking both muls.
if (AddrMode.BaseReg) {
Value *V = AddrMode.BaseReg;
if (V->getType()->isPointerTy())
V = Builder.CreatePtrToInt(V, IntPtrTy, "sunkaddr");
if (V->getType() != IntPtrTy)
V = Builder.CreateIntCast(V, IntPtrTy, /*isSigned=*/true, "sunkaddr");
Result = V;
}
// Add the scale value.
if (AddrMode.Scale) {
Value *V = AddrMode.ScaledReg;
if (V->getType() == IntPtrTy) {
// done.
} else if (V->getType()->isPointerTy()) {
V = Builder.CreatePtrToInt(V, IntPtrTy, "sunkaddr");
} else if (cast<IntegerType>(IntPtrTy)->getBitWidth() <
cast<IntegerType>(V->getType())->getBitWidth()) {
V = Builder.CreateTrunc(V, IntPtrTy, "sunkaddr");
} else {
V = Builder.CreateSExt(V, IntPtrTy, "sunkaddr");
}
if (AddrMode.Scale != 1)
V = Builder.CreateMul(V, ConstantInt::get(IntPtrTy, AddrMode.Scale),
"sunkaddr");
if (Result)
Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
// Add in the BaseGV if present.
if (AddrMode.BaseGV) {
Value *V = Builder.CreatePtrToInt(AddrMode.BaseGV, IntPtrTy, "sunkaddr");
if (Result)
Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
// Add in the Base Offset if present.
if (AddrMode.BaseOffs) {
Value *V = ConstantInt::get(IntPtrTy, AddrMode.BaseOffs);
if (Result)
Result = Builder.CreateAdd(Result, V, "sunkaddr");
else
Result = V;
}
if (Result == 0)
SunkAddr = Constant::getNullValue(Addr->getType());
else
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()) {
// 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);
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 LoopInstSimplify::runOnLoop(Loop *L, LPPassManager &LPM) {
DominatorTree *DT = getAnalysisIfAvailable<DominatorTree>();
LoopInfo *LI = &getAnalysis<LoopInfo>();
const DataLayout *TD = getAnalysisIfAvailable<DataLayout>();
const TargetLibraryInfo *TLI = &getAnalysis<TargetLibraryInfo>();
SmallVector<BasicBlock*, 8> ExitBlocks;
L->getUniqueExitBlocks(ExitBlocks);
array_pod_sort(ExitBlocks.begin(), ExitBlocks.end());
SmallPtrSet<const Instruction*, 8> S1, S2, *ToSimplify = &S1, *Next = &S2;
// The bit we are stealing from the pointer represents whether this basic
// block is the header of a subloop, in which case we only process its phis.
typedef PointerIntPair<BasicBlock*, 1> WorklistItem;
SmallVector<WorklistItem, 16> VisitStack;
SmallPtrSet<BasicBlock*, 32> Visited;
bool Changed = false;
bool LocalChanged;
do {
LocalChanged = false;
VisitStack.clear();
Visited.clear();
VisitStack.push_back(WorklistItem(L->getHeader(), false));
while (!VisitStack.empty()) {
WorklistItem Item = VisitStack.pop_back_val();
BasicBlock *BB = Item.getPointer();
bool IsSubloopHeader = Item.getInt();
// Simplify instructions in the current basic block.
for (BasicBlock::iterator BI = BB->begin(), BE = BB->end(); BI != BE;) {
Instruction *I = BI++;
// The first time through the loop ToSimplify is empty and we try to
// simplify all instructions. On later iterations ToSimplify is not
// empty and we only bother simplifying instructions that are in it.
if (!ToSimplify->empty() && !ToSimplify->count(I))
continue;
// Don't bother simplifying unused instructions.
if (!I->use_empty()) {
Value *V = SimplifyInstruction(I, TD, TLI, DT);
if (V && LI->replacementPreservesLCSSAForm(I, V)) {
// Mark all uses for resimplification next time round the loop.
for (Value::use_iterator UI = I->use_begin(), UE = I->use_end();
UI != UE; ++UI)
Next->insert(cast<Instruction>(*UI));
I->replaceAllUsesWith(V);
LocalChanged = true;
++NumSimplified;
}
}
LocalChanged |= RecursivelyDeleteTriviallyDeadInstructions(I, TLI);
if (IsSubloopHeader && !isa<PHINode>(I))
break;
}
// Add all successors to the worklist, except for loop exit blocks and the
// bodies of subloops. We visit the headers of loops so that we can process
// their phis, but we contract the rest of the subloop body and only follow
// edges leading back to the original loop.
for (succ_iterator SI = succ_begin(BB), SE = succ_end(BB); SI != SE;
++SI) {
BasicBlock *SuccBB = *SI;
if (!Visited.insert(SuccBB))
continue;
const Loop *SuccLoop = LI->getLoopFor(SuccBB);
if (SuccLoop && SuccLoop->getHeader() == SuccBB
&& L->contains(SuccLoop)) {
VisitStack.push_back(WorklistItem(SuccBB, true));
SmallVector<BasicBlock*, 8> SubLoopExitBlocks;
SuccLoop->getExitBlocks(SubLoopExitBlocks);
for (unsigned i = 0; i < SubLoopExitBlocks.size(); ++i) {
BasicBlock *ExitBB = SubLoopExitBlocks[i];
if (LI->getLoopFor(ExitBB) == L && Visited.insert(ExitBB))
VisitStack.push_back(WorklistItem(ExitBB, false));
}
continue;
}
bool IsExitBlock = std::binary_search(ExitBlocks.begin(),
ExitBlocks.end(), SuccBB);
if (IsExitBlock)
continue;
VisitStack.push_back(WorklistItem(SuccBB, false));
}
}
// Place the list of instructions to simplify on the next loop iteration
// into ToSimplify.
std::swap(ToSimplify, Next);
Next->clear();
Changed |= LocalChanged;
} while (LocalChanged);
return Changed;
}
/// PromoteValuesInLoop - Try to promote memory values to scalars by sinking
/// stores out of the loop and moving loads to before the loop. We do this by
/// looping over the stores in the loop, looking for stores to Must pointers
/// which are loop invariant. We promote these memory locations to use allocas
/// instead. These allocas can easily be raised to register values by the
/// PromoteMem2Reg functionality.
///
void LICM::PromoteValuesInLoop() {
// PromotedValues - List of values that are promoted out of the loop. Each
// value has an alloca instruction for it, and a canonical version of the
// pointer.
std::vector<std::pair<AllocaInst*, Value*> > PromotedValues;
std::map<Value*, AllocaInst*> ValueToAllocaMap; // Map of ptr to alloca
FindPromotableValuesInLoop(PromotedValues, ValueToAllocaMap);
if (ValueToAllocaMap.empty()) return; // If there are values to promote.
Changed = true;
NumPromoted += PromotedValues.size();
std::vector<Value*> PointerValueNumbers;
// Emit a copy from the value into the alloca'd value in the loop preheader
TerminatorInst *LoopPredInst = Preheader->getTerminator();
for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i) {
Value *Ptr = PromotedValues[i].second;
// If we are promoting a pointer value, update alias information for the
// inserted load.
Value *LoadValue = 0;
if (cast<PointerType>(Ptr->getType())->getElementType()->isPointerTy()) {
// Locate a load or store through the pointer, and assign the same value
// to LI as we are loading or storing. Since we know that the value is
// stored in this loop, this will always succeed.
for (Value::use_iterator UI = Ptr->use_begin(), E = Ptr->use_end();
UI != E; ++UI) {
User *U = *UI;
if (LoadInst *LI = dyn_cast<LoadInst>(U)) {
LoadValue = LI;
break;
} else if (StoreInst *SI = dyn_cast<StoreInst>(U)) {
if (SI->getOperand(1) == Ptr) {
LoadValue = SI->getOperand(0);
break;
}
}
}
assert(LoadValue && "No store through the pointer found!");
PointerValueNumbers.push_back(LoadValue); // Remember this for later.
}
// Load from the memory we are promoting.
LoadInst *LI = new LoadInst(Ptr, Ptr->getName()+".promoted", LoopPredInst);
if (LoadValue) CurAST->copyValue(LoadValue, LI);
// Store into the temporary alloca.
new StoreInst(LI, PromotedValues[i].first, LoopPredInst);
}
// Scan the basic blocks in the loop, replacing uses of our pointers with
// uses of the allocas in question.
//
for (Loop::block_iterator I = CurLoop->block_begin(),
E = CurLoop->block_end(); I != E; ++I) {
BasicBlock *BB = *I;
// Rewrite all loads and stores in the block of the pointer...
for (BasicBlock::iterator II = BB->begin(), E = BB->end(); II != E; ++II) {
if (LoadInst *L = dyn_cast<LoadInst>(II)) {
std::map<Value*, AllocaInst*>::iterator
I = ValueToAllocaMap.find(L->getOperand(0));
if (I != ValueToAllocaMap.end())
L->setOperand(0, I->second); // Rewrite load instruction...
} else if (StoreInst *S = dyn_cast<StoreInst>(II)) {
std::map<Value*, AllocaInst*>::iterator
I = ValueToAllocaMap.find(S->getOperand(1));
if (I != ValueToAllocaMap.end())
S->setOperand(1, I->second); // Rewrite store instruction...
}
}
}
// Now that the body of the loop uses the allocas instead of the original
// memory locations, insert code to copy the alloca value back into the
// original memory location on all exits from the loop. Note that we only
// want to insert one copy of the code in each exit block, though the loop may
// exit to the same block more than once.
//
SmallPtrSet<BasicBlock*, 16> ProcessedBlocks;
SmallVector<BasicBlock*, 8> ExitBlocks;
CurLoop->getExitBlocks(ExitBlocks);
for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i) {
if (!ProcessedBlocks.insert(ExitBlocks[i]))
continue;
// Copy all of the allocas into their memory locations.
BasicBlock::iterator BI = ExitBlocks[i]->getFirstNonPHI();
Instruction *InsertPos = BI;
unsigned PVN = 0;
for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i) {
// Load from the alloca.
LoadInst *LI = new LoadInst(PromotedValues[i].first, "", InsertPos);
// If this is a pointer type, update alias info appropriately.
if (LI->getType()->isPointerTy())
CurAST->copyValue(PointerValueNumbers[PVN++], LI);
// Store into the memory we promoted.
new StoreInst(LI, PromotedValues[i].second, InsertPos);
}
}
// Now that we have done the deed, use the mem2reg functionality to promote
// all of the new allocas we just created into real SSA registers.
//
std::vector<AllocaInst*> PromotedAllocas;
PromotedAllocas.reserve(PromotedValues.size());
for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i)
PromotedAllocas.push_back(PromotedValues[i].first);
PromoteMemToReg(PromotedAllocas, *DT, *DF, CurAST);
}
/// PromoteArguments - This method checks the specified function to see if there
/// are any promotable arguments and if it is safe to promote the function (for
/// example, all callers are direct). If safe to promote some arguments, it
/// calls the DoPromotion method.
///
CallGraphNode *ArgPromotion::PromoteArguments(CallGraphNode *CGN) {
Function *F = CGN->getFunction();
// Make sure that it is local to this module.
if (!F || !F->hasLocalLinkage()) return 0;
// First check: see if there are any pointer arguments! If not, quick exit.
SmallVector<std::pair<Argument*, unsigned>, 16> PointerArgs;
unsigned ArgNo = 0;
for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end();
I != E; ++I, ++ArgNo)
if (I->getType()->isPointerTy())
PointerArgs.push_back(std::pair<Argument*, unsigned>(I, ArgNo));
if (PointerArgs.empty()) return 0;
// Second check: make sure that all callers are direct callers. We can't
// transform functions that have indirect callers. Also see if the function
// is self-recursive.
bool isSelfRecursive = false;
for (Value::use_iterator UI = F->use_begin(), E = F->use_end();
UI != E; ++UI) {
CallSite CS(*UI);
// Must be a direct call.
if (CS.getInstruction() == 0 || !CS.isCallee(UI)) return 0;
if (CS.getInstruction()->getParent()->getParent() == F)
isSelfRecursive = true;
}
// Check to see which arguments are promotable. If an argument is promotable,
// add it to ArgsToPromote.
SmallPtrSet<Argument*, 8> ArgsToPromote;
SmallPtrSet<Argument*, 8> ByValArgsToTransform;
for (unsigned i = 0; i != PointerArgs.size(); ++i) {
bool isByVal = F->paramHasAttr(PointerArgs[i].second+1, Attribute::ByVal);
Argument *PtrArg = PointerArgs[i].first;
Type *AgTy = cast<PointerType>(PtrArg->getType())->getElementType();
// If this is a byval argument, and if the aggregate type is small, just
// pass the elements, which is always safe.
if (isByVal) {
if (StructType *STy = dyn_cast<StructType>(AgTy)) {
if (maxElements > 0 && STy->getNumElements() > maxElements) {
DEBUG(dbgs() << "argpromotion disable promoting argument '"
<< PtrArg->getName() << "' because it would require adding more"
<< " than " << maxElements << " arguments to the function.\n");
continue;
}
// If all the elements are single-value types, we can promote it.
bool AllSimple = true;
for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
if (!STy->getElementType(i)->isSingleValueType()) {
AllSimple = false;
break;
}
}
// Safe to transform, don't even bother trying to "promote" it.
// Passing the elements as a scalar will allow scalarrepl to hack on
// the new alloca we introduce.
if (AllSimple) {
ByValArgsToTransform.insert(PtrArg);
continue;
}
}
}
// If the argument is a recursive type and we're in a recursive
// function, we could end up infinitely peeling the function argument.
if (isSelfRecursive) {
if (StructType *STy = dyn_cast<StructType>(AgTy)) {
bool RecursiveType = false;
for (unsigned i = 0, e = STy->getNumElements(); i != e; ++i) {
if (STy->getElementType(i) == PtrArg->getType()) {
RecursiveType = true;
break;
}
}
if (RecursiveType)
continue;
}
}
// Otherwise, see if we can promote the pointer to its value.
if (isSafeToPromoteArgument(PtrArg, isByVal))
ArgsToPromote.insert(PtrArg);
}
// No promotable pointer arguments.
if (ArgsToPromote.empty() && ByValArgsToTransform.empty())
return 0;
return DoPromotion(F, ArgsToPromote, ByValArgsToTransform);
}
/// ClusterNeighboringLoads - Force nearby loads together by "gluing" them.
/// This function finds loads of the same base and different offsets. If the
/// offsets are not far apart (target specific), it add MVT::Glue inputs and
/// outputs to ensure they are scheduled together and in order. This
/// optimization may benefit some targets by improving cache locality.
void ScheduleDAGSDNodes::ClusterNeighboringLoads(SDNode *Node) {
SDNode *Chain = nullptr;
unsigned NumOps = Node->getNumOperands();
if (Node->getOperand(NumOps-1).getValueType() == MVT::Other)
Chain = Node->getOperand(NumOps-1).getNode();
if (!Chain)
return;
// Look for other loads of the same chain. Find loads that are loading from
// the same base pointer and different offsets.
SmallPtrSet<SDNode*, 16> Visited;
SmallVector<int64_t, 4> Offsets;
DenseMap<long long, SDNode*> O2SMap; // Map from offset to SDNode.
bool Cluster = false;
SDNode *Base = Node;
// This algorithm requires a reasonably low use count before finding a match
// to avoid uselessly blowing up compile time in large blocks.
unsigned UseCount = 0;
for (SDNode::use_iterator I = Chain->use_begin(), E = Chain->use_end();
I != E && UseCount < 100; ++I, ++UseCount) {
SDNode *User = *I;
if (User == Node || !Visited.insert(User))
continue;
int64_t Offset1, Offset2;
if (!TII->areLoadsFromSameBasePtr(Base, User, Offset1, Offset2) ||
Offset1 == Offset2)
// FIXME: Should be ok if they addresses are identical. But earlier
// optimizations really should have eliminated one of the loads.
continue;
if (O2SMap.insert(std::make_pair(Offset1, Base)).second)
Offsets.push_back(Offset1);
O2SMap.insert(std::make_pair(Offset2, User));
Offsets.push_back(Offset2);
if (Offset2 < Offset1)
Base = User;
Cluster = true;
// Reset UseCount to allow more matches.
UseCount = 0;
}
if (!Cluster)
return;
// Sort them in increasing order.
std::sort(Offsets.begin(), Offsets.end());
// Check if the loads are close enough.
SmallVector<SDNode*, 4> Loads;
unsigned NumLoads = 0;
int64_t BaseOff = Offsets[0];
SDNode *BaseLoad = O2SMap[BaseOff];
Loads.push_back(BaseLoad);
for (unsigned i = 1, e = Offsets.size(); i != e; ++i) {
int64_t Offset = Offsets[i];
SDNode *Load = O2SMap[Offset];
if (!TII->shouldScheduleLoadsNear(BaseLoad, Load, BaseOff, Offset,NumLoads))
break; // Stop right here. Ignore loads that are further away.
Loads.push_back(Load);
++NumLoads;
}
if (NumLoads == 0)
return;
// Cluster loads by adding MVT::Glue outputs and inputs. This also
// ensure they are scheduled in order of increasing addresses.
SDNode *Lead = Loads[0];
SDValue InGlue = SDValue(nullptr, 0);
if (AddGlue(Lead, InGlue, true, DAG))
InGlue = SDValue(Lead, Lead->getNumValues() - 1);
for (unsigned I = 1, E = Loads.size(); I != E; ++I) {
bool OutGlue = I < E - 1;
SDNode *Load = Loads[I];
// If AddGlue fails, we could leave an unsused glue value. This should not
// cause any
if (AddGlue(Load, InGlue, OutGlue, DAG)) {
if (OutGlue)
InGlue = SDValue(Load, Load->getNumValues() - 1);
++LoadsClustered;
}
else if (!OutGlue && InGlue.getNode())
RemoveUnusedGlue(InGlue.getNode(), DAG);
}
}
void ScheduleDAGSDNodes::BuildSchedUnits() {
// During scheduling, the NodeId field of SDNode is used to map SDNodes
// to their associated SUnits by holding SUnits table indices. A value
// of -1 means the SDNode does not yet have an associated SUnit.
unsigned NumNodes = 0;
for (SelectionDAG::allnodes_iterator NI = DAG->allnodes_begin(),
E = DAG->allnodes_end(); NI != E; ++NI) {
NI->setNodeId(-1);
++NumNodes;
}
// Reserve entries in the vector for each of the SUnits we are creating. This
// ensure that reallocation of the vector won't happen, so SUnit*'s won't get
// invalidated.
// FIXME: Multiply by 2 because we may clone nodes during scheduling.
// This is a temporary workaround.
SUnits.reserve(NumNodes * 2);
// Add all nodes in depth first order.
SmallVector<SDNode*, 64> Worklist;
SmallPtrSet<SDNode*, 64> Visited;
Worklist.push_back(DAG->getRoot().getNode());
Visited.insert(DAG->getRoot().getNode());
SmallVector<SUnit*, 8> CallSUnits;
while (!Worklist.empty()) {
SDNode *NI = Worklist.pop_back_val();
// Add all operands to the worklist unless they've already been added.
for (unsigned i = 0, e = NI->getNumOperands(); i != e; ++i)
if (Visited.insert(NI->getOperand(i).getNode()))
Worklist.push_back(NI->getOperand(i).getNode());
if (isPassiveNode(NI)) // Leaf node, e.g. a TargetImmediate.
continue;
// If this node has already been processed, stop now.
if (NI->getNodeId() != -1) continue;
SUnit *NodeSUnit = newSUnit(NI);
// See if anything is glued to this node, if so, add them to glued
// nodes. Nodes can have at most one glue input and one glue output. Glue
// is required to be the last operand and result of a node.
// Scan up to find glued preds.
SDNode *N = NI;
while (N->getNumOperands() &&
N->getOperand(N->getNumOperands()-1).getValueType() == MVT::Glue) {
N = N->getOperand(N->getNumOperands()-1).getNode();
assert(N->getNodeId() == -1 && "Node already inserted!");
N->setNodeId(NodeSUnit->NodeNum);
if (N->isMachineOpcode() && TII->get(N->getMachineOpcode()).isCall())
NodeSUnit->isCall = true;
}
// Scan down to find any glued succs.
N = NI;
while (N->getValueType(N->getNumValues()-1) == MVT::Glue) {
SDValue GlueVal(N, N->getNumValues()-1);
// There are either zero or one users of the Glue result.
bool HasGlueUse = false;
for (SDNode::use_iterator UI = N->use_begin(), E = N->use_end();
UI != E; ++UI)
if (GlueVal.isOperandOf(*UI)) {
HasGlueUse = true;
assert(N->getNodeId() == -1 && "Node already inserted!");
N->setNodeId(NodeSUnit->NodeNum);
N = *UI;
if (N->isMachineOpcode() && TII->get(N->getMachineOpcode()).isCall())
NodeSUnit->isCall = true;
break;
}
if (!HasGlueUse) break;
}
if (NodeSUnit->isCall)
CallSUnits.push_back(NodeSUnit);
// Schedule zero-latency TokenFactor below any nodes that may increase the
// schedule height. Otherwise, ancestors of the TokenFactor may appear to
// have false stalls.
if (NI->getOpcode() == ISD::TokenFactor)
NodeSUnit->isScheduleLow = true;
// If there are glue operands involved, N is now the bottom-most node
// of the sequence of nodes that are glued together.
// Update the SUnit.
NodeSUnit->setNode(N);
assert(N->getNodeId() == -1 && "Node already inserted!");
N->setNodeId(NodeSUnit->NodeNum);
// Compute NumRegDefsLeft. This must be done before AddSchedEdges.
InitNumRegDefsLeft(NodeSUnit);
// Assign the Latency field of NodeSUnit using target-provided information.
computeLatency(NodeSUnit);
}
// Find all call operands.
while (!CallSUnits.empty()) {
SUnit *SU = CallSUnits.pop_back_val();
for (const SDNode *SUNode = SU->getNode(); SUNode;
SUNode = SUNode->getGluedNode()) {
if (SUNode->getOpcode() != ISD::CopyToReg)
continue;
SDNode *SrcN = SUNode->getOperand(2).getNode();
if (isPassiveNode(SrcN)) continue; // Not scheduled.
SUnit *SrcSU = &SUnits[SrcN->getNodeId()];
SrcSU->isCallOp = true;
}
}
}
/// Deduce nocapture attributes for the SCC.
static bool addArgumentAttrs(const SCCNodeSet &SCCNodes) {
bool Changed = false;
ArgumentGraph AG;
AttrBuilder B;
B.addAttribute(Attribute::NoCapture);
// Check each function in turn, determining which pointer arguments are not
// captured.
for (Function *F : SCCNodes) {
// We can infer and propagate function attributes only when we know that the
// definition we'll get at link time is *exactly* the definition we see now.
// For more details, see GlobalValue::mayBeDerefined.
if (!F->hasExactDefinition())
continue;
// Functions that are readonly (or readnone) and nounwind and don't return
// a value can't capture arguments. Don't analyze them.
if (F->onlyReadsMemory() && F->doesNotThrow() &&
F->getReturnType()->isVoidTy()) {
for (Function::arg_iterator A = F->arg_begin(), E = F->arg_end(); A != E;
++A) {
if (A->getType()->isPointerTy() && !A->hasNoCaptureAttr()) {
A->addAttr(AttributeSet::get(F->getContext(), A->getArgNo() + 1, B));
++NumNoCapture;
Changed = true;
}
}
continue;
}
for (Function::arg_iterator A = F->arg_begin(), E = F->arg_end(); A != E;
++A) {
if (!A->getType()->isPointerTy())
continue;
bool HasNonLocalUses = false;
if (!A->hasNoCaptureAttr()) {
ArgumentUsesTracker Tracker(SCCNodes);
PointerMayBeCaptured(&*A, &Tracker);
if (!Tracker.Captured) {
if (Tracker.Uses.empty()) {
// If it's trivially not captured, mark it nocapture now.
A->addAttr(
AttributeSet::get(F->getContext(), A->getArgNo() + 1, B));
++NumNoCapture;
Changed = true;
} else {
// If it's not trivially captured and not trivially not captured,
// then it must be calling into another function in our SCC. Save
// its particulars for Argument-SCC analysis later.
ArgumentGraphNode *Node = AG[&*A];
for (SmallVectorImpl<Argument *>::iterator
UI = Tracker.Uses.begin(),
UE = Tracker.Uses.end();
UI != UE; ++UI) {
Node->Uses.push_back(AG[*UI]);
if (*UI != &*A)
HasNonLocalUses = true;
}
}
}
// Otherwise, it's captured. Don't bother doing SCC analysis on it.
}
if (!HasNonLocalUses && !A->onlyReadsMemory()) {
// Can we determine that it's readonly/readnone without doing an SCC?
// Note that we don't allow any calls at all here, or else our result
// will be dependent on the iteration order through the functions in the
// SCC.
SmallPtrSet<Argument *, 8> Self;
Self.insert(&*A);
Attribute::AttrKind R = determinePointerReadAttrs(&*A, Self);
if (R != Attribute::None) {
AttrBuilder B;
B.addAttribute(R);
A->addAttr(AttributeSet::get(A->getContext(), A->getArgNo() + 1, B));
Changed = true;
R == Attribute::ReadOnly ? ++NumReadOnlyArg : ++NumReadNoneArg;
}
}
}
}
// The graph we've collected is partial because we stopped scanning for
// argument uses once we solved the argument trivially. These partial nodes
// show up as ArgumentGraphNode objects with an empty Uses list, and for
// these nodes the final decision about whether they capture has already been
// made. If the definition doesn't have a 'nocapture' attribute by now, it
// captures.
for (scc_iterator<ArgumentGraph *> I = scc_begin(&AG); !I.isAtEnd(); ++I) {
const std::vector<ArgumentGraphNode *> &ArgumentSCC = *I;
if (ArgumentSCC.size() == 1) {
if (!ArgumentSCC[0]->Definition)
continue; // synthetic root node
// eg. "void f(int* x) { if (...) f(x); }"
if (ArgumentSCC[0]->Uses.size() == 1 &&
ArgumentSCC[0]->Uses[0] == ArgumentSCC[0]) {
Argument *A = ArgumentSCC[0]->Definition;
A->addAttr(AttributeSet::get(A->getContext(), A->getArgNo() + 1, B));
++NumNoCapture;
Changed = true;
}
continue;
}
bool SCCCaptured = false;
for (auto I = ArgumentSCC.begin(), E = ArgumentSCC.end();
I != E && !SCCCaptured; ++I) {
ArgumentGraphNode *Node = *I;
if (Node->Uses.empty()) {
if (!Node->Definition->hasNoCaptureAttr())
SCCCaptured = true;
}
}
if (SCCCaptured)
continue;
SmallPtrSet<Argument *, 8> ArgumentSCCNodes;
// Fill ArgumentSCCNodes with the elements of the ArgumentSCC. Used for
// quickly looking up whether a given Argument is in this ArgumentSCC.
for (auto I = ArgumentSCC.begin(), E = ArgumentSCC.end(); I != E; ++I) {
ArgumentSCCNodes.insert((*I)->Definition);
}
for (auto I = ArgumentSCC.begin(), E = ArgumentSCC.end();
I != E && !SCCCaptured; ++I) {
ArgumentGraphNode *N = *I;
for (SmallVectorImpl<ArgumentGraphNode *>::iterator UI = N->Uses.begin(),
UE = N->Uses.end();
UI != UE; ++UI) {
Argument *A = (*UI)->Definition;
if (A->hasNoCaptureAttr() || ArgumentSCCNodes.count(A))
continue;
SCCCaptured = true;
break;
}
}
if (SCCCaptured)
continue;
for (unsigned i = 0, e = ArgumentSCC.size(); i != e; ++i) {
Argument *A = ArgumentSCC[i]->Definition;
A->addAttr(AttributeSet::get(A->getContext(), A->getArgNo() + 1, B));
++NumNoCapture;
Changed = true;
}
// We also want to compute readonly/readnone. With a small number of false
// negatives, we can assume that any pointer which is captured isn't going
// to be provably readonly or readnone, since by definition we can't
// analyze all uses of a captured pointer.
//
// The false negatives happen when the pointer is captured by a function
// that promises readonly/readnone behaviour on the pointer, then the
// pointer's lifetime ends before anything that writes to arbitrary memory.
// Also, a readonly/readnone pointer may be returned, but returning a
// pointer is capturing it.
Attribute::AttrKind ReadAttr = Attribute::ReadNone;
for (unsigned i = 0, e = ArgumentSCC.size(); i != e; ++i) {
Argument *A = ArgumentSCC[i]->Definition;
Attribute::AttrKind K = determinePointerReadAttrs(A, ArgumentSCCNodes);
if (K == Attribute::ReadNone)
continue;
if (K == Attribute::ReadOnly) {
ReadAttr = Attribute::ReadOnly;
continue;
}
ReadAttr = K;
break;
}
if (ReadAttr != Attribute::None) {
AttrBuilder B, R;
B.addAttribute(ReadAttr);
R.addAttribute(Attribute::ReadOnly).addAttribute(Attribute::ReadNone);
for (unsigned i = 0, e = ArgumentSCC.size(); i != e; ++i) {
Argument *A = ArgumentSCC[i]->Definition;
// Clear out existing readonly/readnone attributes
A->removeAttr(AttributeSet::get(A->getContext(), A->getArgNo() + 1, R));
A->addAttr(AttributeSet::get(A->getContext(), A->getArgNo() + 1, B));
ReadAttr == Attribute::ReadOnly ? ++NumReadOnlyArg : ++NumReadNoneArg;
Changed = true;
}
}
}
return Changed;
}
static bool markAliveBlocks(BasicBlock *BB,
SmallPtrSet<BasicBlock*, 128> &Reachable) {
SmallVector<BasicBlock*, 128> Worklist;
Worklist.push_back(BB);
bool Changed = false;
do {
BB = Worklist.pop_back_val();
if (!Reachable.insert(BB))
continue;
// Do a quick scan of the basic block, turning any obviously unreachable
// instructions into LLVM unreachable insts. The instruction combining pass
// canonicalizes unreachable insts into stores to null or undef.
for (BasicBlock::iterator BBI = BB->begin(), E = BB->end(); BBI != E;++BBI){
if (CallInst *CI = dyn_cast<CallInst>(BBI)) {
if (CI->doesNotReturn()) {
// If we found a call to a no-return function, insert an unreachable
// instruction after it. Make sure there isn't *already* one there
// though.
++BBI;
if (!isa<UnreachableInst>(BBI)) {
// Don't insert a call to llvm.trap right before the unreachable.
changeToUnreachable(BBI, false);
Changed = true;
}
break;
}
}
// Store to undef and store to null are undefined and used to signal that
// they should be changed to unreachable by passes that can't modify the
// CFG.
if (StoreInst *SI = dyn_cast<StoreInst>(BBI)) {
// Don't touch volatile stores.
if (SI->isVolatile()) continue;
Value *Ptr = SI->getOperand(1);
if (isa<UndefValue>(Ptr) ||
(isa<ConstantPointerNull>(Ptr) &&
SI->getPointerAddressSpace() == 0)) {
changeToUnreachable(SI, true);
Changed = true;
break;
}
}
}
// Turn invokes that call 'nounwind' functions into ordinary calls.
if (InvokeInst *II = dyn_cast<InvokeInst>(BB->getTerminator())) {
Value *Callee = II->getCalledValue();
if (isa<ConstantPointerNull>(Callee) || isa<UndefValue>(Callee)) {
changeToUnreachable(II, true);
Changed = true;
} else if (II->doesNotThrow()) {
if (II->use_empty() && II->onlyReadsMemory()) {
// jump to the normal destination branch.
BranchInst::Create(II->getNormalDest(), II);
II->getUnwindDest()->removePredecessor(II->getParent());
II->eraseFromParent();
} else
changeToCall(II);
Changed = true;
}
}
Changed |= ConstantFoldTerminator(BB, true);
for (succ_iterator SI = succ_begin(BB), SE = succ_end(BB); SI != SE; ++SI)
Worklist.push_back(*SI);
} while (!Worklist.empty());
return Changed;
}
/// findValueImpl - Implementation helper for findValue.
Value *Lint::findValueImpl(Value *V, bool OffsetOk,
SmallPtrSet<Value *, 4> &Visited) const {
// Detect self-referential values.
if (!Visited.insert(V))
return UndefValue::get(V->getType());
// TODO: Look through sext or zext cast, when the result is known to
// be interpreted as signed or unsigned, respectively.
// TODO: Look through eliminable cast pairs.
// TODO: Look through calls with unique return values.
// TODO: Look through vector insert/extract/shuffle.
V = OffsetOk ? V->getUnderlyingObject() : V->stripPointerCasts();
if (LoadInst *L = dyn_cast<LoadInst>(V)) {
BasicBlock::iterator BBI = L;
BasicBlock *BB = L->getParent();
SmallPtrSet<BasicBlock *, 4> VisitedBlocks;
for (;;) {
if (!VisitedBlocks.insert(BB)) break;
if (Value *U = FindAvailableLoadedValue(L->getPointerOperand(),
BB, BBI, 6, AA))
return findValueImpl(U, OffsetOk, Visited);
if (BBI != BB->begin()) break;
BB = BB->getUniquePredecessor();
if (!BB) break;
BBI = BB->end();
}
} else if (PHINode *PN = dyn_cast<PHINode>(V)) {
if (Value *W = PN->hasConstantValue())
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
} else if (CastInst *CI = dyn_cast<CastInst>(V)) {
if (CI->isNoopCast(TD ? TD->getIntPtrType(V->getContext()) :
Type::getInt64Ty(V->getContext())))
return findValueImpl(CI->getOperand(0), OffsetOk, Visited);
} else if (ExtractValueInst *Ex = dyn_cast<ExtractValueInst>(V)) {
if (Value *W = FindInsertedValue(Ex->getAggregateOperand(),
Ex->idx_begin(),
Ex->idx_end()))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V)) {
// Same as above, but for ConstantExpr instead of Instruction.
if (Instruction::isCast(CE->getOpcode())) {
if (CastInst::isNoopCast(Instruction::CastOps(CE->getOpcode()),
CE->getOperand(0)->getType(),
CE->getType(),
TD ? TD->getIntPtrType(V->getContext()) :
Type::getInt64Ty(V->getContext())))
return findValueImpl(CE->getOperand(0), OffsetOk, Visited);
} else if (CE->getOpcode() == Instruction::ExtractValue) {
const SmallVector<unsigned, 4> &Indices = CE->getIndices();
if (Value *W = FindInsertedValue(CE->getOperand(0),
Indices.begin(),
Indices.end()))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
}
}
// As a last resort, try SimplifyInstruction or constant folding.
if (Instruction *Inst = dyn_cast<Instruction>(V)) {
if (Value *W = SimplifyInstruction(Inst, TD, DT))
return findValueImpl(W, OffsetOk, Visited);
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V)) {
if (Value *W = ConstantFoldConstantExpression(CE, TD))
if (W != V)
return findValueImpl(W, OffsetOk, Visited);
}
return V;
}
/// CanMergeBlocks - Return true if we can merge BB into DestBB if there is a
/// single uncond branch between them, and BB contains no other non-phi
/// instructions.
bool CodeGenPrepare::CanMergeBlocks(const BasicBlock *BB,
const BasicBlock *DestBB) const {
// We only want to eliminate blocks whose phi nodes are used by phi nodes in
// the successor. If there are more complex condition (e.g. preheaders),
// don't mess around with them.
BasicBlock::const_iterator BBI = BB->begin();
while (const PHINode *PN = dyn_cast<PHINode>(BBI++)) {
for (Value::const_use_iterator UI = PN->use_begin(), E = PN->use_end();
UI != E; ++UI) {
const Instruction *User = cast<Instruction>(*UI);
if (User->getParent() != DestBB || !isa<PHINode>(User))
return false;
// If User is inside DestBB block and it is a PHINode then check
// incoming value. If incoming value is not from BB then this is
// a complex condition (e.g. preheaders) we want to avoid here.
if (User->getParent() == DestBB) {
if (const PHINode *UPN = dyn_cast<PHINode>(User))
for (unsigned I = 0, E = UPN->getNumIncomingValues(); I != E; ++I) {
Instruction *Insn = dyn_cast<Instruction>(UPN->getIncomingValue(I));
if (Insn && Insn->getParent() == BB &&
Insn->getParent() != UPN->getIncomingBlock(I))
return false;
}
}
}
}
// If BB and DestBB contain any common predecessors, then the phi nodes in BB
// and DestBB may have conflicting incoming values for the block. If so, we
// can't merge the block.
const PHINode *DestBBPN = dyn_cast<PHINode>(DestBB->begin());
if (!DestBBPN) return true; // no conflict.
// Collect the preds of BB.
SmallPtrSet<const BasicBlock*, 16> BBPreds;
if (const PHINode *BBPN = dyn_cast<PHINode>(BB->begin())) {
// It is faster to get preds from a PHI than with pred_iterator.
for (unsigned i = 0, e = BBPN->getNumIncomingValues(); i != e; ++i)
BBPreds.insert(BBPN->getIncomingBlock(i));
} else {
BBPreds.insert(pred_begin(BB), pred_end(BB));
}
// Walk the preds of DestBB.
for (unsigned i = 0, e = DestBBPN->getNumIncomingValues(); i != e; ++i) {
BasicBlock *Pred = DestBBPN->getIncomingBlock(i);
if (BBPreds.count(Pred)) { // Common predecessor?
BBI = DestBB->begin();
while (const PHINode *PN = dyn_cast<PHINode>(BBI++)) {
const Value *V1 = PN->getIncomingValueForBlock(Pred);
const Value *V2 = PN->getIncomingValueForBlock(BB);
// If V2 is a phi node in BB, look up what the mapped value will be.
if (const PHINode *V2PN = dyn_cast<PHINode>(V2))
if (V2PN->getParent() == BB)
V2 = V2PN->getIncomingValueForBlock(Pred);
// If there is a conflict, bail out.
if (V1 != V2) return false;
}
}
}
return true;
}
bool ReduceCrashingBlocks::TestBlocks(std::vector<const BasicBlock*> &BBs) {
// Clone the program to try hacking it apart...
ValueToValueMapTy VMap;
Module *M = CloneModule(BD.getProgram(), VMap).release();
// Convert list to set for fast lookup...
SmallPtrSet<BasicBlock*, 8> Blocks;
for (unsigned i = 0, e = BBs.size(); i != e; ++i)
Blocks.insert(cast<BasicBlock>(VMap[BBs[i]]));
outs() << "Checking for crash with only these blocks:";
unsigned NumPrint = Blocks.size();
if (NumPrint > 10) NumPrint = 10;
for (unsigned i = 0, e = NumPrint; i != e; ++i)
outs() << " " << BBs[i]->getName();
if (NumPrint < Blocks.size())
outs() << "... <" << Blocks.size() << " total>";
outs() << ": ";
// Loop over and delete any hack up any blocks that are not listed...
for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I)
for (Function::iterator BB = I->begin(), E = I->end(); BB != E; ++BB)
if (!Blocks.count(&*BB) && BB->getTerminator()->getNumSuccessors()) {
// Loop over all of the successors of this block, deleting any PHI nodes
// that might include it.
for (succ_iterator SI = succ_begin(&*BB), E = succ_end(&*BB); SI != E;
++SI)
(*SI)->removePredecessor(&*BB);
TerminatorInst *BBTerm = BB->getTerminator();
if (BBTerm->isEHPad())
continue;
if (!BBTerm->getType()->isVoidTy() && !BBTerm->getType()->isTokenTy())
BBTerm->replaceAllUsesWith(Constant::getNullValue(BBTerm->getType()));
// Replace the old terminator instruction.
BB->getInstList().pop_back();
new UnreachableInst(BB->getContext(), &*BB);
}
// The CFG Simplifier pass may delete one of the basic blocks we are
// interested in. If it does we need to take the block out of the list. Make
// a "persistent mapping" by turning basic blocks into <function, name> pairs.
// This won't work well if blocks are unnamed, but that is just the risk we
// have to take.
std::vector<std::pair<std::string, std::string> > BlockInfo;
for (BasicBlock *BB : Blocks)
BlockInfo.emplace_back(BB->getParent()->getName(), BB->getName());
// Now run the CFG simplify pass on the function...
std::vector<std::string> Passes;
Passes.push_back("simplifycfg");
Passes.push_back("verify");
std::unique_ptr<Module> New = BD.runPassesOn(M, Passes);
delete M;
if (!New) {
errs() << "simplifycfg failed!\n";
exit(1);
}
M = New.release();
// Try running on the hacked up program...
if (TestFn(BD, M)) {
BD.setNewProgram(M); // It crashed, keep the trimmed version...
// Make sure to use basic block pointers that point into the now-current
// module, and that they don't include any deleted blocks.
BBs.clear();
const ValueSymbolTable &GST = M->getValueSymbolTable();
for (unsigned i = 0, e = BlockInfo.size(); i != e; ++i) {
Function *F = cast<Function>(GST.lookup(BlockInfo[i].first));
ValueSymbolTable &ST = F->getValueSymbolTable();
Value* V = ST.lookup(BlockInfo[i].second);
if (V && V->getType() == Type::getLabelTy(V->getContext()))
BBs.push_back(cast<BasicBlock>(V));
}
return true;
}
delete M; // It didn't crash, try something else.
return false;
}
/// DupRetToEnableTailCallOpts - Look for opportunities to duplicate return
/// instructions to the predecessor to enable tail call optimizations. The
/// case it is currently looking for is:
/// 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
///
/// =>
///
/// 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
///
bool CodeGenPrepare::DupRetToEnableTailCallOpts(ReturnInst *RI) {
if (!TLI)
return false;
Value *V = RI->getReturnValue();
PHINode *PN = V ? dyn_cast<PHINode>(V) : NULL;
if (V && !PN)
return false;
BasicBlock *BB = RI->getParent();
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();
Attributes CallerRetAttr = F->getAttributes().getRetAttributes();
if ((CallerRetAttr & Attribute::ZExt) || (CallerRetAttr & 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 != 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.
Attributes CalleeRetAttr = CS.getAttributes().getRetAttributes();
if ((CalleeRetAttr ^ CallerRetAttr) & ~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 && pred_begin(BB) == pred_end(BB))
BB->eraseFromParent();
return Changed;
}
/// ComputeLiveInBlocks - Determine which blocks the value is live in. These
/// are blocks which lead to uses. Knowing this allows us to avoid inserting
/// PHI nodes into blocks which don't lead to uses (thus, the inserted phi nodes
/// would be dead).
void PromoteMem2Reg::
ComputeLiveInBlocks(AllocaInst *AI, AllocaInfo &Info,
const SmallPtrSet<BasicBlock*, 32> &DefBlocks,
SmallPtrSet<BasicBlock*, 32> &LiveInBlocks) {
// To determine liveness, we must iterate through the predecessors of blocks
// where the def is live. Blocks are added to the worklist if we need to
// check their predecessors. Start with all the using blocks.
SmallVector<BasicBlock*, 64> LiveInBlockWorklist;
LiveInBlockWorklist.insert(LiveInBlockWorklist.end(),
Info.UsingBlocks.begin(), Info.UsingBlocks.end());
// If any of the using blocks is also a definition block, check to see if the
// definition occurs before or after the use. If it happens before the use,
// the value isn't really live-in.
for (unsigned i = 0, e = LiveInBlockWorklist.size(); i != e; ++i) {
BasicBlock *BB = LiveInBlockWorklist[i];
if (!DefBlocks.count(BB)) continue;
// Okay, this is a block that both uses and defines the value. If the first
// reference to the alloca is a def (store), then we know it isn't live-in.
for (BasicBlock::iterator I = BB->begin(); ; ++I) {
if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
if (SI->getOperand(1) != AI) continue;
// We found a store to the alloca before a load. The alloca is not
// actually live-in here.
LiveInBlockWorklist[i] = LiveInBlockWorklist.back();
LiveInBlockWorklist.pop_back();
--i, --e;
break;
}
if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
if (LI->getOperand(0) != AI) continue;
// Okay, we found a load before a store to the alloca. It is actually
// live into this block.
break;
}
}
}
// Now that we have a set of blocks where the phi is live-in, recursively add
// their predecessors until we find the full region the value is live.
while (!LiveInBlockWorklist.empty()) {
BasicBlock *BB = LiveInBlockWorklist.pop_back_val();
// The block really is live in here, insert it into the set. If already in
// the set, then it has already been processed.
if (!LiveInBlocks.insert(BB))
continue;
// Since the value is live into BB, it is either defined in a predecessor or
// live into it to. Add the preds to the worklist unless they are a
// defining block.
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
BasicBlock *P = *PI;
// The value is not live into a predecessor if it defines the value.
if (DefBlocks.count(P))
continue;
// Otherwise it is, add to the worklist.
LiveInBlockWorklist.push_back(P);
}
}
}
void TypeChecker::configureInterfaceType(AbstractFunctionDecl *func) {
Type funcTy;
Type initFuncTy;
auto *sig = func->getGenericSignature();
if (auto fn = dyn_cast<FuncDecl>(func)) {
funcTy = fn->getBodyResultTypeLoc().getType();
if (!funcTy) {
funcTy = TupleType::getEmpty(Context);
} else {
funcTy = getResultType(*this, fn, funcTy);
}
} else if (auto ctor = dyn_cast<ConstructorDecl>(func)) {
auto *dc = ctor->getDeclContext();
funcTy = dc->getSelfInterfaceType();
// Adjust result type for failability.
if (ctor->getFailability() != OTK_None)
funcTy = OptionalType::get(ctor->getFailability(), funcTy);
initFuncTy = funcTy;
} else {
assert(isa<DestructorDecl>(func));
funcTy = TupleType::getEmpty(Context);
}
auto paramLists = func->getParameterLists();
SmallVector<ParameterList*, 4> storedParamLists;
// FIXME: Destructors don't have the '()' pattern in their signature, so
// paste it here.
if (isa<DestructorDecl>(func)) {
assert(paramLists.size() == 1 && "Only the self paramlist");
storedParamLists.push_back(paramLists[0]);
storedParamLists.push_back(ParameterList::createEmpty(Context));
paramLists = storedParamLists;
}
bool hasSelf = func->getDeclContext()->isTypeContext();
for (unsigned i = 0, e = paramLists.size(); i != e; ++i) {
Type argTy;
Type initArgTy;
Type selfTy;
if (i == e-1 && hasSelf) {
selfTy = func->computeInterfaceSelfType(/*isInitializingCtor=*/false);
// Substitute in our own 'self' parameter.
argTy = selfTy;
if (initFuncTy) {
initArgTy = func->computeInterfaceSelfType(/*isInitializingCtor=*/true);
}
} else {
argTy = paramLists[e - i - 1]->getInterfaceType(func->getDeclContext());
if (initFuncTy)
initArgTy = argTy;
}
// 'throws' only applies to the innermost function.
AnyFunctionType::ExtInfo info;
if (i == 0 && func->hasThrows())
info = info.withThrows();
assert(!argTy->hasArchetype());
assert(!funcTy->hasArchetype());
if (initFuncTy)
assert(!initFuncTy->hasArchetype());
if (sig && i == e-1) {
funcTy = GenericFunctionType::get(sig, argTy, funcTy, info);
if (initFuncTy)
initFuncTy = GenericFunctionType::get(sig, initArgTy, initFuncTy, info);
} else {
funcTy = FunctionType::get(argTy, funcTy, info);
if (initFuncTy)
initFuncTy = FunctionType::get(initArgTy, initFuncTy, info);
}
}
// Record the interface type.
func->setInterfaceType(funcTy);
if (initFuncTy)
cast<ConstructorDecl>(func)->setInitializerInterfaceType(initFuncTy);
if (func->getGenericParams()) {
// Collect all generic params referenced in parameter types,
// return type or requirements.
SmallPtrSet<GenericTypeParamDecl *, 4> referencedGenericParams;
auto visitorFn = [&referencedGenericParams](Type t) {
if (auto *paramTy = t->getAs<GenericTypeParamType>())
referencedGenericParams.insert(paramTy->getDecl());
};
funcTy->castTo<AnyFunctionType>()->getInput().visit(visitorFn);
funcTy->castTo<AnyFunctionType>()->getResult().visit(visitorFn);
auto requirements = sig->getRequirements();
for (auto req : requirements) {
if (req.getKind() == RequirementKind::SameType) {
// Same type requirements may allow for generic
// inference, even if this generic parameter
// is not mentioned in the function signature.
// TODO: Make the test more precise.
auto left = req.getFirstType();
auto right = req.getSecondType();
// For now consider any references inside requirements
// as a possibility to infer the generic type.
left.visit(visitorFn);
right.visit(visitorFn);
}
}
// Find the depth of the function's own generic parameters.
unsigned fnGenericParamsDepth = func->getGenericParams()->getDepth();
// Check that every generic parameter type from the signature is
// among referencedArchetypes.
for (auto *genParam : sig->getGenericParams()) {
auto *paramDecl = genParam->getDecl();
if (paramDecl->getDepth() != fnGenericParamsDepth)
continue;
if (!referencedGenericParams.count(paramDecl)) {
// Produce an error that this generic parameter cannot be bound.
diagnose(paramDecl->getLoc(), diag::unreferenced_generic_parameter,
paramDecl->getNameStr());
func->setInvalid();
}
}
}
}
/// runOnLoop - Remove dead loops, by which we mean loops that do not impact the
/// observable behavior of the program other than finite running time. Note
/// we do ensure that this never remove a loop that might be infinite, as doing
/// so could change the halting/non-halting nature of a program.
/// NOTE: This entire process relies pretty heavily on LoopSimplify and LCSSA
/// in order to make various safety checks work.
bool LoopDeletion::runOnLoop(Loop *L, LPPassManager &LPM) {
// We can only remove the loop if there is a preheader that we can
// branch from after removing it.
BasicBlock *preheader = L->getLoopPreheader();
if (!preheader)
return false;
// If LoopSimplify form is not available, stay out of trouble.
if (!L->hasDedicatedExits())
return false;
// We can't remove loops that contain subloops. If the subloops were dead,
// they would already have been removed in earlier executions of this pass.
if (L->begin() != L->end())
return false;
SmallVector<BasicBlock*, 4> exitingBlocks;
L->getExitingBlocks(exitingBlocks);
SmallVector<BasicBlock*, 4> exitBlocks;
L->getUniqueExitBlocks(exitBlocks);
// We require that the loop only have a single exit block. Otherwise, we'd
// be in the situation of needing to be able to solve statically which exit
// block will be branched to, or trying to preserve the branching logic in
// a loop invariant manner.
if (exitBlocks.size() != 1)
return false;
// Finally, we have to check that the loop really is dead.
bool Changed = false;
if (!isLoopDead(L, exitingBlocks, exitBlocks, Changed, preheader))
return Changed;
// Don't remove loops for which we can't solve the trip count.
// They could be infinite, in which case we'd be changing program behavior.
ScalarEvolution &SE = getAnalysis<ScalarEvolution>();
const SCEV *S = SE.getMaxBackedgeTakenCount(L);
if (isa<SCEVCouldNotCompute>(S))
return Changed;
// Now that we know the removal is safe, remove the loop by changing the
// branch from the preheader to go to the single exit block.
BasicBlock *exitBlock = exitBlocks[0];
// Because we're deleting a large chunk of code at once, the sequence in which
// we remove things is very important to avoid invalidation issues. Don't
// mess with this unless you have good reason and know what you're doing.
// Tell ScalarEvolution that the loop is deleted. Do this before
// deleting the loop so that ScalarEvolution can look at the loop
// to determine what it needs to clean up.
SE.forgetLoop(L);
// Connect the preheader directly to the exit block.
TerminatorInst *TI = preheader->getTerminator();
TI->replaceUsesOfWith(L->getHeader(), exitBlock);
// Rewrite phis in the exit block to get their inputs from
// the preheader instead of the exiting block.
BasicBlock *exitingBlock = exitingBlocks[0];
BasicBlock::iterator BI = exitBlock->begin();
while (PHINode *P = dyn_cast<PHINode>(BI)) {
int j = P->getBasicBlockIndex(exitingBlock);
assert(j >= 0 && "Can't find exiting block in exit block's phi node!");
P->setIncomingBlock(j, preheader);
for (unsigned i = 1; i < exitingBlocks.size(); ++i)
P->removeIncomingValue(exitingBlocks[i]);
++BI;
}
// Update the dominator tree and remove the instructions and blocks that will
// be deleted from the reference counting scheme.
DominatorTree &DT = getAnalysis<DominatorTree>();
SmallVector<DomTreeNode*, 8> ChildNodes;
for (Loop::block_iterator LI = L->block_begin(), LE = L->block_end();
LI != LE; ++LI) {
// Move all of the block's children to be children of the preheader, which
// allows us to remove the domtree entry for the block.
ChildNodes.insert(ChildNodes.begin(), DT[*LI]->begin(), DT[*LI]->end());
for (SmallVector<DomTreeNode*, 8>::iterator DI = ChildNodes.begin(),
DE = ChildNodes.end(); DI != DE; ++DI) {
DT.changeImmediateDominator(*DI, DT[preheader]);
}
ChildNodes.clear();
DT.eraseNode(*LI);
// Remove the block from the reference counting scheme, so that we can
// delete it freely later.
(*LI)->dropAllReferences();
}
// Erase the instructions and the blocks without having to worry
// about ordering because we already dropped the references.
// NOTE: This iteration is safe because erasing the block does not remove its
// entry from the loop's block list. We do that in the next section.
for (Loop::block_iterator LI = L->block_begin(), LE = L->block_end();
LI != LE; ++LI)
(*LI)->eraseFromParent();
// Finally, the blocks from loopinfo. This has to happen late because
// otherwise our loop iterators won't work.
LoopInfo &loopInfo = getAnalysis<LoopInfo>();
SmallPtrSet<BasicBlock*, 8> blocks;
blocks.insert(L->block_begin(), L->block_end());
for (SmallPtrSet<BasicBlock*,8>::iterator I = blocks.begin(),
E = blocks.end(); I != E; ++I)
loopInfo.removeBlock(*I);
// The last step is to inform the loop pass manager that we've
// eliminated this loop.
LPM.deleteLoopFromQueue(L);
Changed = true;
++NumDeleted;
return Changed;
}
bool Inliner::runOnSCC(CallGraphSCC &SCC) {
CallGraph &CG = getAnalysis<CallGraph>();
const TargetData *TD = getAnalysisIfAvailable<TargetData>();
SmallPtrSet<Function*, 8> SCCFunctions;
DEBUG(dbgs() << "Inliner visiting SCC:");
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I) {
Function *F = (*I)->getFunction();
if (F) SCCFunctions.insert(F);
DEBUG(dbgs() << " " << (F ? F->getName() : "INDIRECTNODE"));
}
// Scan through and identify all call sites ahead of time so that we only
// inline call sites in the original functions, not call sites that result
// from inlining other functions.
SmallVector<std::pair<CallSite, int>, 16> CallSites;
// When inlining a callee produces new call sites, we want to keep track of
// the fact that they were inlined from the callee. This allows us to avoid
// infinite inlining in some obscure cases. To represent this, we use an
// index into the InlineHistory vector.
SmallVector<std::pair<Function*, int>, 8> InlineHistory;
for (CallGraphSCC::iterator I = SCC.begin(), E = SCC.end(); I != E; ++I) {
Function *F = (*I)->getFunction();
if (!F) continue;
for (Function::iterator BB = F->begin(), E = F->end(); BB != E; ++BB)
for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I) {
CallSite CS(cast<Value>(I));
// If this isn't a call, or it is a call to an intrinsic, it can
// never be inlined.
if (!CS || isa<IntrinsicInst>(I))
continue;
// If this is a direct call to an external function, we can never inline
// it. If it is an indirect call, inlining may resolve it to be a
// direct call, so we keep it.
if (CS.getCalledFunction() && CS.getCalledFunction()->isDeclaration())
continue;
CallSites.push_back(std::make_pair(CS, -1));
}
}
DEBUG(dbgs() << ": " << CallSites.size() << " call sites.\n");
// If there are no calls in this function, exit early.
if (CallSites.empty())
return false;
// Now that we have all of the call sites, move the ones to functions in the
// current SCC to the end of the list.
unsigned FirstCallInSCC = CallSites.size();
for (unsigned i = 0; i < FirstCallInSCC; ++i)
if (Function *F = CallSites[i].first.getCalledFunction())
if (SCCFunctions.count(F))
std::swap(CallSites[i--], CallSites[--FirstCallInSCC]);
InlinedArrayAllocasTy InlinedArrayAllocas;
InlineFunctionInfo InlineInfo(&CG, TD);
// Now that we have all of the call sites, loop over them and inline them if
// it looks profitable to do so.
bool Changed = false;
bool LocalChange;
do {
LocalChange = false;
// Iterate over the outer loop because inlining functions can cause indirect
// calls to become direct calls.
for (unsigned CSi = 0; CSi != CallSites.size(); ++CSi) {
CallSite CS = CallSites[CSi].first;
Function *Caller = CS.getCaller();
Function *Callee = CS.getCalledFunction();
// If this call site is dead and it is to a readonly function, we should
// just delete the call instead of trying to inline it, regardless of
// size. This happens because IPSCCP propagates the result out of the
// call and then we're left with the dead call.
if (isInstructionTriviallyDead(CS.getInstruction())) {
DEBUG(dbgs() << " -> Deleting dead call: "
<< *CS.getInstruction() << "\n");
// Update the call graph by deleting the edge from Callee to Caller.
CG[Caller]->removeCallEdgeFor(CS);
CS.getInstruction()->eraseFromParent();
++NumCallsDeleted;
} else {
// We can only inline direct calls to non-declarations.
if (Callee == 0 || Callee->isDeclaration()) continue;
// If this call site was obtained by inlining another function, verify
// that the include path for the function did not include the callee
// itself. If so, we'd be recursively inlining the same function,
// which would provide the same callsites, which would cause us to
// infinitely inline.
int InlineHistoryID = CallSites[CSi].second;
if (InlineHistoryID != -1 &&
InlineHistoryIncludes(Callee, InlineHistoryID, InlineHistory))
continue;
// If the policy determines that we should inline this function,
// try to do so.
if (!shouldInline(CS))
continue;
// Attempt to inline the function.
if (!InlineCallIfPossible(CS, InlineInfo, InlinedArrayAllocas,
InlineHistoryID, InsertLifetime))
continue;
++NumInlined;
// If inlining this function gave us any new call sites, throw them
// onto our worklist to process. They are useful inline candidates.
if (!InlineInfo.InlinedCalls.empty()) {
// Create a new inline history entry for this, so that we remember
// that these new callsites came about due to inlining Callee.
int NewHistoryID = InlineHistory.size();
InlineHistory.push_back(std::make_pair(Callee, InlineHistoryID));
for (unsigned i = 0, e = InlineInfo.InlinedCalls.size();
i != e; ++i) {
Value *Ptr = InlineInfo.InlinedCalls[i];
CallSites.push_back(std::make_pair(CallSite(Ptr), NewHistoryID));
}
}
}
// If we inlined or deleted the last possible call site to the function,
// delete the function body now.
if (Callee && Callee->use_empty() && Callee->hasLocalLinkage() &&
// TODO: Can remove if in SCC now.
!SCCFunctions.count(Callee) &&
// The function may be apparently dead, but if there are indirect
// callgraph references to the node, we cannot delete it yet, this
// could invalidate the CGSCC iterator.
CG[Callee]->getNumReferences() == 0) {
DEBUG(dbgs() << " -> Deleting dead function: "
<< Callee->getName() << "\n");
CallGraphNode *CalleeNode = CG[Callee];
// Remove any call graph edges from the callee to its callees.
CalleeNode->removeAllCalledFunctions();
// Removing the node for callee from the call graph and delete it.
delete CG.removeFunctionFromModule(CalleeNode);
++NumDeleted;
}
// Remove this call site from the list. If possible, use
// swap/pop_back for efficiency, but do not use it if doing so would
// move a call site to a function in this SCC before the
// 'FirstCallInSCC' barrier.
if (SCC.isSingular()) {
CallSites[CSi] = CallSites.back();
CallSites.pop_back();
} else {
CallSites.erase(CallSites.begin()+CSi);
}
--CSi;
Changed = true;
LocalChange = true;
}
} while (LocalChange);
return Changed;
}
// If we can determine that all possible objects pointed to by the provided
// pointer value are, not only dereferenceable, but also definitively less than
// or equal to the provided maximum size, then return true. Otherwise, return
// false (constant global values and allocas fall into this category).
//
// FIXME: This should probably live in ValueTracking (or similar).
static bool isObjectSizeLessThanOrEq(Value *V, uint64_t MaxSize,
const DataLayout &DL) {
SmallPtrSet<Value *, 4> Visited;
SmallVector<Value *, 4> Worklist(1, V);
do {
Value *P = Worklist.pop_back_val();
P = P->stripPointerCasts();
if (!Visited.insert(P).second)
continue;
if (SelectInst *SI = dyn_cast<SelectInst>(P)) {
Worklist.push_back(SI->getTrueValue());
Worklist.push_back(SI->getFalseValue());
continue;
}
if (PHINode *PN = dyn_cast<PHINode>(P)) {
for (Value *IncValue : PN->incoming_values())
Worklist.push_back(IncValue);
continue;
}
if (GlobalAlias *GA = dyn_cast<GlobalAlias>(P)) {
if (GA->mayBeOverridden())
return false;
Worklist.push_back(GA->getAliasee());
continue;
}
// If we know how big this object is, and it is less than MaxSize, continue
// searching. Otherwise, return false.
if (AllocaInst *AI = dyn_cast<AllocaInst>(P)) {
if (!AI->getAllocatedType()->isSized())
return false;
ConstantInt *CS = dyn_cast<ConstantInt>(AI->getArraySize());
if (!CS)
return false;
uint64_t TypeSize = DL.getTypeAllocSize(AI->getAllocatedType());
// Make sure that, even if the multiplication below would wrap as an
// uint64_t, we still do the right thing.
if ((CS->getValue().zextOrSelf(128)*APInt(128, TypeSize)).ugt(MaxSize))
return false;
continue;
}
if (GlobalVariable *GV = dyn_cast<GlobalVariable>(P)) {
if (!GV->hasDefinitiveInitializer() || !GV->isConstant())
return false;
uint64_t InitSize = DL.getTypeAllocSize(GV->getValueType());
if (InitSize > MaxSize)
return false;
continue;
}
return false;
} while (!Worklist.empty());
return true;
}
bool BDCE::runOnFunction(Function& F) {
if (skipOptnoneFunction(F))
return false;
AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
DenseMap<Instruction *, APInt> AliveBits;
SmallVector<Instruction*, 128> Worklist;
// The set of visited instructions (non-integer-typed only).
SmallPtrSet<Instruction*, 128> Visited;
// Collect the set of "root" instructions that are known live.
for (Instruction &I : inst_range(F)) {
if (!isAlwaysLive(&I))
continue;
DEBUG(dbgs() << "BDCE: Root: " << I << "\n");
// For integer-valued instructions, set up an initial empty set of alive
// bits and add the instruction to the work list. For other instructions
// add their operands to the work list (for integer values operands, mark
// all bits as live).
if (IntegerType *IT = dyn_cast<IntegerType>(I.getType())) {
if (!AliveBits.count(&I)) {
AliveBits[&I] = APInt(IT->getBitWidth(), 0);
Worklist.push_back(&I);
}
continue;
}
// Non-integer-typed instructions...
for (Use &OI : I.operands()) {
if (Instruction *J = dyn_cast<Instruction>(OI)) {
if (IntegerType *IT = dyn_cast<IntegerType>(J->getType()))
AliveBits[J] = APInt::getAllOnesValue(IT->getBitWidth());
Worklist.push_back(J);
}
}
// To save memory, we don't add I to the Visited set here. Instead, we
// check isAlwaysLive on every instruction when searching for dead
// instructions later (we need to check isAlwaysLive for the
// integer-typed instructions anyway).
}
// Propagate liveness backwards to operands.
while (!Worklist.empty()) {
Instruction *UserI = Worklist.pop_back_val();
DEBUG(dbgs() << "BDCE: Visiting: " << *UserI);
APInt AOut;
if (UserI->getType()->isIntegerTy()) {
AOut = AliveBits[UserI];
DEBUG(dbgs() << " Alive Out: " << AOut);
}
DEBUG(dbgs() << "\n");
if (!UserI->getType()->isIntegerTy())
Visited.insert(UserI);
APInt KnownZero, KnownOne, KnownZero2, KnownOne2;
// Compute the set of alive bits for each operand. These are anded into the
// existing set, if any, and if that changes the set of alive bits, the
// operand is added to the work-list.
for (Use &OI : UserI->operands()) {
if (Instruction *I = dyn_cast<Instruction>(OI)) {
if (IntegerType *IT = dyn_cast<IntegerType>(I->getType())) {
unsigned BitWidth = IT->getBitWidth();
APInt AB = APInt::getAllOnesValue(BitWidth);
if (UserI->getType()->isIntegerTy() && !AOut &&
!isAlwaysLive(UserI)) {
AB = APInt(BitWidth, 0);
} else {
// If all bits of the output are dead, then all bits of the input
// Bits of each operand that are used to compute alive bits of the
// output are alive, all others are dead.
determineLiveOperandBits(UserI, I, OI.getOperandNo(), AOut, AB,
KnownZero, KnownOne,
KnownZero2, KnownOne2);
}
// If we've added to the set of alive bits (or the operand has not
// been previously visited), then re-queue the operand to be visited
// again.
APInt ABPrev(BitWidth, 0);
auto ABI = AliveBits.find(I);
if (ABI != AliveBits.end())
ABPrev = ABI->second;
APInt ABNew = AB | ABPrev;
if (ABNew != ABPrev || ABI == AliveBits.end()) {
AliveBits[I] = std::move(ABNew);
Worklist.push_back(I);
}
} else if (!Visited.count(I)) {
Worklist.push_back(I);
}
}
}
}
bool Changed = false;
// The inverse of the live set is the dead set. These are those instructions
// which have no side effects and do not influence the control flow or return
// value of the function, and may therefore be deleted safely.
// NOTE: We reuse the Worklist vector here for memory efficiency.
for (Instruction &I : inst_range(F)) {
// For live instructions that have all dead bits, first make them dead by
// replacing all uses with something else. Then, if they don't need to
// remain live (because they have side effects, etc.) we can remove them.
if (I.getType()->isIntegerTy()) {
auto ABI = AliveBits.find(&I);
if (ABI != AliveBits.end()) {
if (ABI->second.getBoolValue())
continue;
DEBUG(dbgs() << "BDCE: Trivializing: " << I << " (all bits dead)\n");
// FIXME: In theory we could substitute undef here instead of zero.
// This should be reconsidered once we settle on the semantics of
// undef, poison, etc.
Value *Zero = ConstantInt::get(I.getType(), 0);
++NumSimplified;
I.replaceAllUsesWith(Zero);
Changed = true;
}
} else if (Visited.count(&I)) {
continue;
}
if (isAlwaysLive(&I))
continue;
Worklist.push_back(&I);
I.dropAllReferences();
Changed = true;
}
for (Instruction *&I : Worklist) {
++NumRemoved;
I->eraseFromParent();
}
return Changed;
}
/// HandleURoRInvokes - Handle invokes of "_Unwind_Resume_or_Rethrow" calls. The
/// "unwind" part of these invokes jump to a landing pad within the current
/// function. This is a candidate to merge the selector associated with the URoR
/// invoke with the one from the URoR's landing pad.
bool DwarfEHPrepare::HandleURoRInvokes() {
if (!EHCatchAllValue) {
EHCatchAllValue =
F->getParent()->getNamedGlobal("llvm.eh.catch.all.value");
if (!EHCatchAllValue) return false;
}
if (!SelectorIntrinsic) {
SelectorIntrinsic =
Intrinsic::getDeclaration(F->getParent(), Intrinsic::eh_selector);
if (!SelectorIntrinsic) return false;
}
SmallPtrSet<IntrinsicInst*, 32> Sels;
SmallPtrSet<IntrinsicInst*, 32> CatchAllSels;
FindAllCleanupSelectors(Sels, CatchAllSels);
if (!DT)
// We require DominatorTree information.
return CleanupSelectors(CatchAllSels);
if (!URoR) {
URoR = F->getParent()->getFunction("_Unwind_Resume_or_Rethrow");
if (!URoR) return CleanupSelectors(CatchAllSels);
}
SmallPtrSet<InvokeInst*, 32> URoRInvokes;
FindAllURoRInvokes(URoRInvokes);
SmallPtrSet<IntrinsicInst*, 32> SelsToConvert;
for (SmallPtrSet<IntrinsicInst*, 32>::iterator
SI = Sels.begin(), SE = Sels.end(); SI != SE; ++SI) {
const BasicBlock *SelBB = (*SI)->getParent();
for (SmallPtrSet<InvokeInst*, 32>::iterator
UI = URoRInvokes.begin(), UE = URoRInvokes.end(); UI != UE; ++UI) {
const BasicBlock *URoRBB = (*UI)->getParent();
if (DT->dominates(SelBB, URoRBB)) {
SelsToConvert.insert(*SI);
break;
}
}
}
bool Changed = false;
if (Sels.size() != SelsToConvert.size()) {
// If we haven't been able to convert all of the clean-up selectors, then
// loop through the slow way to see if they still need to be converted.
if (!ExceptionValueIntrinsic) {
ExceptionValueIntrinsic =
Intrinsic::getDeclaration(F->getParent(), Intrinsic::eh_exception);
if (!ExceptionValueIntrinsic)
return CleanupSelectors(CatchAllSels);
}
for (Value::use_iterator
I = ExceptionValueIntrinsic->use_begin(),
E = ExceptionValueIntrinsic->use_end(); I != E; ++I) {
IntrinsicInst *EHPtr = dyn_cast<IntrinsicInst>(*I);
if (!EHPtr || EHPtr->getParent()->getParent() != F) continue;
Changed |= PromoteEHPtrStore(EHPtr);
bool URoRInvoke = false;
SmallPtrSet<IntrinsicInst*, 8> SelCalls;
Changed |= FindSelectorAndURoR(EHPtr, URoRInvoke, SelCalls);
if (URoRInvoke) {
// This EH pointer is being used by an invoke of an URoR instruction and
// an eh.selector intrinsic call. If the eh.selector is a 'clean-up', we
// need to convert it to a 'catch-all'.
for (SmallPtrSet<IntrinsicInst*, 8>::iterator
SI = SelCalls.begin(), SE = SelCalls.end(); SI != SE; ++SI)
if (!HasCatchAllInSelector(*SI))
SelsToConvert.insert(*SI);
}
}
}
if (!SelsToConvert.empty()) {
// Convert all clean-up eh.selectors, which are associated with "invokes" of
// URoR calls, into catch-all eh.selectors.
Changed = true;
for (SmallPtrSet<IntrinsicInst*, 8>::iterator
SI = SelsToConvert.begin(), SE = SelsToConvert.end();
SI != SE; ++SI) {
IntrinsicInst *II = *SI;
// Use the exception object pointer and the personality function
// from the original selector.
CallSite CS(II);
IntrinsicInst::op_iterator I = CS.arg_begin();
IntrinsicInst::op_iterator E = CS.arg_end();
IntrinsicInst::op_iterator B = prior(E);
// Exclude last argument if it is an integer.
if (isa<ConstantInt>(B)) E = B;
// Add exception object pointer (front).
// Add personality function (next).
// Add in any filter IDs (rest).
SmallVector<Value*, 8> Args(I, E);
Args.push_back(EHCatchAllValue->getInitializer()); // Catch-all indicator.
CallInst *NewSelector =
CallInst::Create(SelectorIntrinsic, Args.begin(), Args.end(),
"eh.sel.catch.all", II);
NewSelector->setTailCall(II->isTailCall());
NewSelector->setAttributes(II->getAttributes());
NewSelector->setCallingConv(II->getCallingConv());
II->replaceAllUsesWith(NewSelector);
II->eraseFromParent();
}
}
Changed |= CleanupSelectors(CatchAllSels);
return Changed;
}
// Look up multiple symbols in the symbol table and return a set of
// Modules that define those symbols.
bool
Archive::findModulesDefiningSymbols(std::set<std::string>& symbols,
SmallVectorImpl<Module*>& result,
std::string* error) {
if (!mapfile || !base) {
if (error)
*error = "Empty archive invalid for finding modules defining symbols";
return false;
}
if (symTab.empty()) {
// We don't have a symbol table, so we must build it now but lets also
// make sure that we populate the modules table as we do this to ensure
// that we don't load them twice when findModuleDefiningSymbol is called
// below.
// Get a pointer to the first file
const char* At = base + firstFileOffset;
const char* End = mapfile->getBufferEnd();
while ( At < End) {
// Compute the offset to be put in the symbol table
unsigned offset = At - base - firstFileOffset;
// Parse the file's header
ArchiveMember* mbr = parseMemberHeader(At, End, error);
if (!mbr)
return false;
// If it contains symbols
if (mbr->isBitcode()) {
// Get the symbols
std::vector<std::string> symbols;
std::string FullMemberName = archPath.str() + "(" +
mbr->getPath().str() + ")";
Module* M =
GetBitcodeSymbols(At, mbr->getSize(), FullMemberName, Context,
symbols, error);
if (M) {
// Insert the module's symbols into the symbol table
for (std::vector<std::string>::iterator I = symbols.begin(),
E=symbols.end(); I != E; ++I ) {
symTab.insert(std::make_pair(*I, offset));
}
// Insert the Module and the ArchiveMember into the table of
// modules.
modules.insert(std::make_pair(offset, std::make_pair(M, mbr)));
} else {
if (error)
*error = "Can't parse bitcode member: " +
mbr->getPath().str() + ": " + *error;
delete mbr;
return false;
}
}
// Go to the next file location
At += mbr->getSize();
if ((intptr_t(At) & 1) == 1)
At++;
}
}
// At this point we have a valid symbol table (one way or another) so we
// just use it to quickly find the symbols requested.
SmallPtrSet<Module*, 16> Added;
for (std::set<std::string>::iterator I=symbols.begin(),
Next = I,
E=symbols.end(); I != E; I = Next) {
// Increment Next before we invalidate it.
++Next;
// See if this symbol exists
Module* m = findModuleDefiningSymbol(*I,error);
if (!m)
continue;
bool NewMember = Added.insert(m);
if (!NewMember)
continue;
// The symbol exists, insert the Module into our result.
result.push_back(m);
// Remove the symbol now that its been resolved.
symbols.erase(I);
}
return true;
}
/// HandlePHINodesInSuccessorBlocks - Handle PHI nodes in successor blocks.
/// Emit code to ensure constants are copied into registers when needed.
/// Remember the virtual registers that need to be added to the Machine PHI
/// nodes as input. We cannot just directly add them, because expansion
/// might result in multiple MBB's for one BB. As such, the start of the
/// BB might correspond to a different MBB than the end.
bool FastISel::HandlePHINodesInSuccessorBlocks(const BasicBlock *LLVMBB) {
const TerminatorInst *TI = LLVMBB->getTerminator();
SmallPtrSet<MachineBasicBlock *, 4> SuccsHandled;
unsigned OrigNumPHINodesToUpdate = FuncInfo.PHINodesToUpdate.size();
// Check successor nodes' PHI nodes that expect a constant to be available
// from this block.
for (unsigned succ = 0, e = TI->getNumSuccessors(); succ != e; ++succ) {
const BasicBlock *SuccBB = TI->getSuccessor(succ);
if (!isa<PHINode>(SuccBB->begin())) continue;
MachineBasicBlock *SuccMBB = FuncInfo.MBBMap[SuccBB];
// If this terminator has multiple identical successors (common for
// switches), only handle each succ once.
if (!SuccsHandled.insert(SuccMBB)) continue;
MachineBasicBlock::iterator MBBI = SuccMBB->begin();
// At this point we know that there is a 1-1 correspondence between LLVM PHI
// nodes and Machine PHI nodes, but the incoming operands have not been
// emitted yet.
for (BasicBlock::const_iterator I = SuccBB->begin();
const PHINode *PN = dyn_cast<PHINode>(I); ++I) {
// Ignore dead phi's.
if (PN->use_empty()) continue;
// Only handle legal types. Two interesting things to note here. First,
// by bailing out early, we may leave behind some dead instructions,
// since SelectionDAG's HandlePHINodesInSuccessorBlocks will insert its
// own moves. Second, this check is necessary because FastISel doesn't
// use CreateRegs to create registers, so it always creates
// exactly one register for each non-void instruction.
EVT VT = TLI.getValueType(PN->getType(), /*AllowUnknown=*/true);
if (VT == MVT::Other || !TLI.isTypeLegal(VT)) {
// Promote MVT::i1.
if (VT == MVT::i1)
VT = TLI.getTypeToTransformTo(LLVMBB->getContext(), VT);
else {
FuncInfo.PHINodesToUpdate.resize(OrigNumPHINodesToUpdate);
return false;
}
}
const Value *PHIOp = PN->getIncomingValueForBlock(LLVMBB);
// Set the DebugLoc for the copy. Prefer the location of the operand
// if there is one; use the location of the PHI otherwise.
DL = PN->getDebugLoc();
if (const Instruction *Inst = dyn_cast<Instruction>(PHIOp))
DL = Inst->getDebugLoc();
unsigned Reg = getRegForValue(PHIOp);
if (Reg == 0) {
FuncInfo.PHINodesToUpdate.resize(OrigNumPHINodesToUpdate);
return false;
}
FuncInfo.PHINodesToUpdate.push_back(std::make_pair(MBBI++, Reg));
DL = DebugLoc();
}
}
return true;
}
void
MachineVerifier::visitMachineBasicBlockBefore(const MachineBasicBlock *MBB) {
FirstTerminator = 0;
if (MRI->isSSA()) {
// If this block has allocatable physical registers live-in, check that
// it is an entry block or landing pad.
for (MachineBasicBlock::livein_iterator LI = MBB->livein_begin(),
LE = MBB->livein_end();
LI != LE; ++LI) {
unsigned reg = *LI;
if (isAllocatable(reg) && !MBB->isLandingPad() &&
MBB != MBB->getParent()->begin()) {
report("MBB has allocable live-in, but isn't entry or landing-pad.", MBB);
}
}
}
// Count the number of landing pad successors.
SmallPtrSet<MachineBasicBlock*, 4> LandingPadSuccs;
for (MachineBasicBlock::const_succ_iterator I = MBB->succ_begin(),
E = MBB->succ_end(); I != E; ++I) {
if ((*I)->isLandingPad())
LandingPadSuccs.insert(*I);
}
const MCAsmInfo *AsmInfo = TM->getMCAsmInfo();
const BasicBlock *BB = MBB->getBasicBlock();
if (LandingPadSuccs.size() > 1 &&
!(AsmInfo &&
AsmInfo->getExceptionHandlingType() == ExceptionHandling::SjLj &&
BB && isa<SwitchInst>(BB->getTerminator())))
report("MBB has more than one landing pad successor", MBB);
// Call AnalyzeBranch. If it succeeds, there several more conditions to check.
MachineBasicBlock *TBB = 0, *FBB = 0;
SmallVector<MachineOperand, 4> Cond;
if (!TII->AnalyzeBranch(*const_cast<MachineBasicBlock *>(MBB),
TBB, FBB, Cond)) {
// Ok, AnalyzeBranch thinks it knows what's going on with this block. Let's
// check whether its answers match up with reality.
if (!TBB && !FBB) {
// Block falls through to its successor.
MachineFunction::const_iterator MBBI = MBB;
++MBBI;
if (MBBI == MF->end()) {
// It's possible that the block legitimately ends with a noreturn
// call or an unreachable, in which case it won't actually fall
// out the bottom of the function.
} else if (MBB->succ_size() == LandingPadSuccs.size()) {
// It's possible that the block legitimately ends with a noreturn
// call or an unreachable, in which case it won't actuall fall
// out of the block.
} else if (MBB->succ_size() != 1+LandingPadSuccs.size()) {
report("MBB exits via unconditional fall-through but doesn't have "
"exactly one CFG successor!", MBB);
} else if (!MBB->isSuccessor(MBBI)) {
report("MBB exits via unconditional fall-through but its successor "
"differs from its CFG successor!", MBB);
}
if (!MBB->empty() && MBB->back().isBarrier() &&
!TII->isPredicated(&MBB->back())) {
report("MBB exits via unconditional fall-through but ends with a "
"barrier instruction!", MBB);
}
if (!Cond.empty()) {
report("MBB exits via unconditional fall-through but has a condition!",
MBB);
}
} else if (TBB && !FBB && Cond.empty()) {
// Block unconditionally branches somewhere.
if (MBB->succ_size() != 1+LandingPadSuccs.size()) {
report("MBB exits via unconditional branch but doesn't have "
"exactly one CFG successor!", MBB);
} else if (!MBB->isSuccessor(TBB)) {
report("MBB exits via unconditional branch but the CFG "
"successor doesn't match the actual successor!", MBB);
}
if (MBB->empty()) {
report("MBB exits via unconditional branch but doesn't contain "
"any instructions!", MBB);
} else if (!MBB->back().isBarrier()) {
report("MBB exits via unconditional branch but doesn't end with a "
"barrier instruction!", MBB);
} else if (!MBB->back().isTerminator()) {
report("MBB exits via unconditional branch but the branch isn't a "
"terminator instruction!", MBB);
}
} else if (TBB && !FBB && !Cond.empty()) {
// Block conditionally branches somewhere, otherwise falls through.
MachineFunction::const_iterator MBBI = MBB;
++MBBI;
if (MBBI == MF->end()) {
report("MBB conditionally falls through out of function!", MBB);
} if (MBB->succ_size() != 2) {
report("MBB exits via conditional branch/fall-through but doesn't have "
"exactly two CFG successors!", MBB);
} else if (!matchPair(MBB->succ_begin(), TBB, MBBI)) {
report("MBB exits via conditional branch/fall-through but the CFG "
"successors don't match the actual successors!", MBB);
}
if (MBB->empty()) {
report("MBB exits via conditional branch/fall-through but doesn't "
"contain any instructions!", MBB);
} else if (MBB->back().isBarrier()) {
report("MBB exits via conditional branch/fall-through but ends with a "
"barrier instruction!", MBB);
} else if (!MBB->back().isTerminator()) {
report("MBB exits via conditional branch/fall-through but the branch "
"isn't a terminator instruction!", MBB);
}
} else if (TBB && FBB) {
// Block conditionally branches somewhere, otherwise branches
// somewhere else.
if (MBB->succ_size() != 2) {
report("MBB exits via conditional branch/branch but doesn't have "
"exactly two CFG successors!", MBB);
} else if (!matchPair(MBB->succ_begin(), TBB, FBB)) {
report("MBB exits via conditional branch/branch but the CFG "
"successors don't match the actual successors!", MBB);
}
if (MBB->empty()) {
report("MBB exits via conditional branch/branch but doesn't "
"contain any instructions!", MBB);
} else if (!MBB->back().isBarrier()) {
report("MBB exits via conditional branch/branch but doesn't end with a "
"barrier instruction!", MBB);
} else if (!MBB->back().isTerminator()) {
report("MBB exits via conditional branch/branch but the branch "
"isn't a terminator instruction!", MBB);
}
if (Cond.empty()) {
report("MBB exits via conditinal branch/branch but there's no "
"condition!", MBB);
}
} else {
report("AnalyzeBranch returned invalid data!", MBB);
}
}
regsLive.clear();
for (MachineBasicBlock::livein_iterator I = MBB->livein_begin(),
E = MBB->livein_end(); I != E; ++I) {
if (!TargetRegisterInfo::isPhysicalRegister(*I)) {
report("MBB live-in list contains non-physical register", MBB);
continue;
}
regsLive.insert(*I);
for (const uint16_t *R = TRI->getSubRegisters(*I); *R; R++)
regsLive.insert(*R);
}
regsLiveInButUnused = regsLive;
const MachineFrameInfo *MFI = MF->getFrameInfo();
assert(MFI && "Function has no frame info");
BitVector PR = MFI->getPristineRegs(MBB);
for (int I = PR.find_first(); I>0; I = PR.find_next(I)) {
regsLive.insert(I);
for (const uint16_t *R = TRI->getSubRegisters(I); *R; R++)
regsLive.insert(*R);
}
regsKilled.clear();
regsDefined.clear();
if (Indexes)
lastIndex = Indexes->getMBBStartIdx(MBB);
}
/// lowerAcrossUnwindEdges - Find all variables which are alive across an unwind
/// edge and spill them.
void SjLjEHPrepare::lowerAcrossUnwindEdges(Function &F,
ArrayRef<InvokeInst *> Invokes) {
// Finally, scan the code looking for instructions with bad live ranges.
for (BasicBlock &BB : F) {
for (Instruction &Inst : BB) {
// Ignore obvious cases we don't have to handle. In particular, most
// instructions either have no uses or only have a single use inside the
// current block. Ignore them quickly.
if (Inst.use_empty())
continue;
if (Inst.hasOneUse() &&
cast<Instruction>(Inst.user_back())->getParent() == &BB &&
!isa<PHINode>(Inst.user_back()))
continue;
// If this is an alloca in the entry block, it's not a real register
// value.
if (auto *AI = dyn_cast<AllocaInst>(&Inst))
if (AI->isStaticAlloca())
continue;
// Avoid iterator invalidation by copying users to a temporary vector.
SmallVector<Instruction *, 16> Users;
for (User *U : Inst.users()) {
Instruction *UI = cast<Instruction>(U);
if (UI->getParent() != &BB || isa<PHINode>(UI))
Users.push_back(UI);
}
// Find all of the blocks that this value is live in.
SmallPtrSet<BasicBlock *, 32> LiveBBs;
LiveBBs.insert(&BB);
while (!Users.empty()) {
Instruction *U = Users.pop_back_val();
if (!isa<PHINode>(U)) {
MarkBlocksLiveIn(U->getParent(), LiveBBs);
} else {
// Uses for a PHI node occur in their predecessor block.
PHINode *PN = cast<PHINode>(U);
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
if (PN->getIncomingValue(i) == &Inst)
MarkBlocksLiveIn(PN->getIncomingBlock(i), LiveBBs);
}
}
// Now that we know all of the blocks that this thing is live in, see if
// it includes any of the unwind locations.
bool NeedsSpill = false;
for (InvokeInst *Invoke : Invokes) {
BasicBlock *UnwindBlock = Invoke->getUnwindDest();
if (UnwindBlock != &BB && LiveBBs.count(UnwindBlock)) {
LLVM_DEBUG(dbgs() << "SJLJ Spill: " << Inst << " around "
<< UnwindBlock->getName() << "\n");
NeedsSpill = true;
break;
}
}
// If we decided we need a spill, do it.
// FIXME: Spilling this way is overkill, as it forces all uses of
// the value to be reloaded from the stack slot, even those that aren't
// in the unwind blocks. We should be more selective.
if (NeedsSpill) {
DemoteRegToStack(Inst, true);
++NumSpilled;
}
}
}
// Go through the landing pads and remove any PHIs there.
for (InvokeInst *Invoke : Invokes) {
BasicBlock *UnwindBlock = Invoke->getUnwindDest();
LandingPadInst *LPI = UnwindBlock->getLandingPadInst();
// Place PHIs into a set to avoid invalidating the iterator.
SmallPtrSet<PHINode *, 8> PHIsToDemote;
for (BasicBlock::iterator PN = UnwindBlock->begin(); isa<PHINode>(PN); ++PN)
PHIsToDemote.insert(cast<PHINode>(PN));
if (PHIsToDemote.empty())
continue;
// Demote the PHIs to the stack.
for (PHINode *PN : PHIsToDemote)
DemotePHIToStack(PN);
// Move the landingpad instruction back to the top of the landing pad block.
LPI->moveBefore(&UnwindBlock->front());
}
}
/// \brief Recursively traverse the CFG of the function, renaming loads and
/// stores to the allocas which we are promoting.
///
/// IncomingVals indicates what value each Alloca contains on exit from the
/// predecessor block Pred.
void PromoteMem2Reg::RenamePass(BasicBlock *BB, BasicBlock *Pred,
RenamePassData::ValVector &IncomingVals,
std::vector<RenamePassData> &Worklist) {
NextIteration:
// If we are inserting any phi nodes into this BB, they will already be in the
// block.
if (PHINode *APN = dyn_cast<PHINode>(BB->begin())) {
// If we have PHI nodes to update, compute the number of edges from Pred to
// BB.
if (PhiToAllocaMap.count(APN)) {
// We want to be able to distinguish between PHI nodes being inserted by
// this invocation of mem2reg from those phi nodes that already existed in
// the IR before mem2reg was run. We determine that APN is being inserted
// because it is missing incoming edges. All other PHI nodes being
// inserted by this pass of mem2reg will have the same number of incoming
// operands so far. Remember this count.
unsigned NewPHINumOperands = APN->getNumOperands();
unsigned NumEdges = std::count(succ_begin(Pred), succ_end(Pred), BB);
assert(NumEdges && "Must be at least one edge from Pred to BB!");
// Add entries for all the phis.
BasicBlock::iterator PNI = BB->begin();
do {
unsigned AllocaNo = PhiToAllocaMap[APN];
// Add N incoming values to the PHI node.
for (unsigned i = 0; i != NumEdges; ++i)
APN->addIncoming(IncomingVals[AllocaNo], Pred);
// The currently active variable for this block is now the PHI.
IncomingVals[AllocaNo] = APN;
// Get the next phi node.
++PNI;
APN = dyn_cast<PHINode>(PNI);
if (!APN)
break;
// Verify that it is missing entries. If not, it is not being inserted
// by this mem2reg invocation so we want to ignore it.
} while (APN->getNumOperands() == NewPHINumOperands);
}
}
// Don't revisit blocks.
if (!Visited.insert(BB))
return;
for (BasicBlock::iterator II = BB->begin(); !isa<TerminatorInst>(II);) {
Instruction *I = II++; // get the instruction, increment iterator
if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
AllocaInst *Src = dyn_cast<AllocaInst>(LI->getPointerOperand());
if (!Src)
continue;
DenseMap<AllocaInst *, unsigned>::iterator AI = AllocaLookup.find(Src);
if (AI == AllocaLookup.end())
continue;
Value *V = IncomingVals[AI->second];
// Anything using the load now uses the current value.
LI->replaceAllUsesWith(V);
if (AST && LI->getType()->isPointerTy())
AST->deleteValue(LI);
BB->getInstList().erase(LI);
} else if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
// Delete this instruction and mark the name as the current holder of the
// value
AllocaInst *Dest = dyn_cast<AllocaInst>(SI->getPointerOperand());
if (!Dest)
continue;
DenseMap<AllocaInst *, unsigned>::iterator ai = AllocaLookup.find(Dest);
if (ai == AllocaLookup.end())
continue;
// what value were we writing?
IncomingVals[ai->second] = SI->getOperand(0);
// Record debuginfo for the store before removing it.
if (DbgDeclareInst *DDI = AllocaDbgDeclares[ai->second])
ConvertDebugDeclareToDebugValue(DDI, SI, DIB);
BB->getInstList().erase(SI);
}
}
// 'Recurse' to our successors.
succ_iterator I = succ_begin(BB), E = succ_end(BB);
if (I == E)
return;
// Keep track of the successors so we don't visit the same successor twice
SmallPtrSet<BasicBlock *, 8> VisitedSuccs;
// Handle the first successor without using the worklist.
VisitedSuccs.insert(*I);
Pred = BB;
BB = *I;
++I;
for (; I != E; ++I)
if (VisitedSuccs.insert(*I))
Worklist.push_back(RenamePassData(*I, Pred, IncomingVals));
goto NextIteration;
}
bool TypeChecker::validateGenericFuncSignature(AbstractFunctionDecl *func) {
bool invalid = false;
// Create the archetype builder.
ArchetypeBuilder builder = createArchetypeBuilder(func->getParentModule());
// Type check the function declaration, treating all generic type
// parameters as dependent, unresolved.
DependentGenericTypeResolver dependentResolver(builder);
if (checkGenericFuncSignature(*this, &builder, func, dependentResolver))
invalid = true;
// If this triggered a recursive validation, back out: we're done.
// FIXME: This is an awful hack.
if (func->hasType())
return !func->isInvalid();
// Finalize the generic requirements.
(void)builder.finalize(func->getLoc());
// The archetype builder now has all of the requirements, although there might
// still be errors that have not yet been diagnosed. Revert the generic
// function signature and type-check it again, completely.
revertGenericFuncSignature(func);
CompleteGenericTypeResolver completeResolver(*this, builder);
if (checkGenericFuncSignature(*this, nullptr, func, completeResolver))
invalid = true;
// The generic function signature is complete and well-formed. Determine
// the type of the generic function.
// Collect the complete set of generic parameter types.
SmallVector<GenericTypeParamType *, 4> allGenericParams;
collectGenericParamTypes(func->getGenericParams(),
func->getDeclContext()->getGenericSignatureOfContext(),
allGenericParams);
auto sig = builder.getGenericSignature(allGenericParams);
// Debugging of the archetype builder and generic signature generation.
if (sig && Context.LangOpts.DebugGenericSignatures) {
func->dumpRef(llvm::errs());
llvm::errs() << "\n";
builder.dump(llvm::errs());
llvm::errs() << "Generic signature: ";
sig->print(llvm::errs());
llvm::errs() << "\n";
llvm::errs() << "Canonical generic signature: ";
sig->getCanonicalSignature()->print(llvm::errs());
llvm::errs() << "\n";
llvm::errs() << "Canonical generic signature for mangling: ";
sig->getCanonicalManglingSignature(*func->getParentModule())
->print(llvm::errs());
llvm::errs() << "\n";
}
func->setGenericSignature(sig);
if (invalid) {
func->overwriteType(ErrorType::get(Context));
return true;
}
// Compute the function type.
Type funcTy;
Type initFuncTy;
if (auto fn = dyn_cast<FuncDecl>(func)) {
funcTy = fn->getBodyResultTypeLoc().getType();
if (!funcTy) {
funcTy = TupleType::getEmpty(Context);
} else {
funcTy = getResultType(*this, fn, funcTy);
}
} else if (auto ctor = dyn_cast<ConstructorDecl>(func)) {
// FIXME: shouldn't this just be
// ctor->getDeclContext()->getDeclaredInterfaceType()?
if (ctor->getDeclContext()->getAsProtocolOrProtocolExtensionContext()) {
funcTy = ctor->getDeclContext()->getProtocolSelf()->getDeclaredType();
} else {
funcTy = ctor->getExtensionType()->getAnyNominal()
->getDeclaredInterfaceType();
}
// Adjust result type for failability.
if (ctor->getFailability() != OTK_None)
funcTy = OptionalType::get(ctor->getFailability(), funcTy);
initFuncTy = funcTy;
} else {
assert(isa<DestructorDecl>(func));
funcTy = TupleType::getEmpty(Context);
}
auto paramLists = func->getParameterLists();
SmallVector<ParameterList*, 4> storedParamLists;
// FIXME: Destructors don't have the '()' pattern in their signature, so
// paste it here.
if (isa<DestructorDecl>(func)) {
assert(paramLists.size() == 1 && "Only the self paramlist");
storedParamLists.push_back(paramLists[0]);
storedParamLists.push_back(ParameterList::createEmpty(Context));
paramLists = storedParamLists;
}
bool hasSelf = func->getDeclContext()->isTypeContext();
for (unsigned i = 0, e = paramLists.size(); i != e; ++i) {
Type argTy;
Type initArgTy;
Type selfTy;
if (i == e-1 && hasSelf) {
selfTy = func->computeInterfaceSelfType(/*isInitializingCtor=*/false);
// Substitute in our own 'self' parameter.
argTy = selfTy;
if (initFuncTy) {
initArgTy = func->computeInterfaceSelfType(/*isInitializingCtor=*/true);
}
} else {
argTy = paramLists[e - i - 1]->getType(Context);
// For an implicit declaration, our argument type will be in terms of
// archetypes rather than dependent types. Replace the
// archetypes with their corresponding dependent types.
if (func->isImplicit()) {
argTy = ArchetypeBuilder::mapTypeOutOfContext(func, argTy);
}
if (initFuncTy)
initArgTy = argTy;
}
auto info = applyFunctionTypeAttributes(func, i);
// FIXME: We shouldn't even get here if the function isn't locally generic
// to begin with, but fixing that requires a lot of reengineering for local
// definitions in generic contexts.
if (sig && i == e-1) {
if (func->getGenericParams()) {
// Collect all generic params referenced in parameter types,
// return type or requirements.
SmallPtrSet<GenericTypeParamDecl *, 4> referencedGenericParams;
argTy.visit([&referencedGenericParams](Type t) {
if (isa<GenericTypeParamType>(t.getCanonicalTypeOrNull())) {
referencedGenericParams.insert(
t->castTo<GenericTypeParamType>()->getDecl());
}
});
funcTy.visit([&referencedGenericParams](Type t) {
if (isa<GenericTypeParamType>(t.getCanonicalTypeOrNull())) {
referencedGenericParams.insert(
t->castTo<GenericTypeParamType>()->getDecl());
}
});
auto requirements = sig->getRequirements();
for (auto req : requirements) {
if (req.getKind() == RequirementKind::SameType) {
// Same type requirements may allow for generic
// inference, even if this generic parameter
// is not mentioned in the function signature.
// TODO: Make the test more precise.
auto left = req.getFirstType();
auto right = req.getSecondType();
// For now consider any references inside requirements
// as a possibility to infer the generic type.
left.visit([&referencedGenericParams](Type t) {
if (isa<GenericTypeParamType>(t.getCanonicalTypeOrNull())) {
referencedGenericParams.insert(
t->castTo<GenericTypeParamType>()->getDecl());
}
});
right.visit([&referencedGenericParams](Type t) {
if (isa<GenericTypeParamType>(t.getCanonicalTypeOrNull())) {
referencedGenericParams.insert(
t->castTo<GenericTypeParamType>()->getDecl());
}
});
}
}
// Find the depth of the function's own generic parameters.
unsigned fnGenericParamsDepth = func->getGenericParams()->getDepth();
// Check that every generic parameter type from the signature is
// among referencedArchetypes.
for (auto *genParam : sig->getGenericParams()) {
auto *paramDecl = genParam->getDecl();
if (paramDecl->getDepth() != fnGenericParamsDepth)
continue;
if (!referencedGenericParams.count(paramDecl)) {
// Produce an error that this generic parameter cannot be bound.
diagnose(paramDecl->getLoc(), diag::unreferenced_generic_parameter,
paramDecl->getNameStr());
func->setInvalid();
}
}
}
funcTy = GenericFunctionType::get(sig, argTy, funcTy, info);
if (initFuncTy)
initFuncTy = GenericFunctionType::get(sig, initArgTy, initFuncTy, info);
} else {
funcTy = FunctionType::get(argTy, funcTy, info);
if (initFuncTy)
initFuncTy = FunctionType::get(initArgTy, initFuncTy, info);
}
}
// Record the interface type.
func->setInterfaceType(funcTy);
if (initFuncTy)
cast<ConstructorDecl>(func)->setInitializerInterfaceType(initFuncTy);
return false;
}