// recurseBasicBlock() - This calculates the ProfileInfo estimation for a
// single block and then recurses into the successors.
// The algorithm preserves the flow condition, meaning that the sum of the
// weight of the incoming edges must be equal the block weight which must in
// turn be equal to the sume of the weights of the outgoing edges.
// Since the flow of an block is deterimined from the current state of the
// flow, once an edge has a flow assigned this flow is never changed again,
// otherwise it would be possible to violate the flow condition in another
// block.
void ProfileEstimatorPass::recurseBasicBlock(BasicBlock *BB) {
// Break the recursion if this BasicBlock was already visited.
if (BBToVisit.find(BB) == BBToVisit.end()) return;
// Read the LoopInfo for this block.
bool BBisHeader = LI->isLoopHeader(BB);
Loop* BBLoop = LI->getLoopFor(BB);
// To get the block weight, read all incoming edges.
double BBWeight = 0;
std::set<BasicBlock*> ProcessedPreds;
for ( pred_iterator bbi = pred_begin(BB), bbe = pred_end(BB);
bbi != bbe; ++bbi ) {
// If this block was not considered already, add weight.
Edge edge = getEdge(*bbi,BB);
double w = getEdgeWeight(edge);
if (ProcessedPreds.insert(*bbi).second) {
BBWeight += ignoreMissing(w);
}
// If this block is a loop header and the predecessor is contained in this
// loop, thus the edge is a backedge, continue and do not check if the
// value is valid.
if (BBisHeader && BBLoop->contains(*bbi)) {
printEdgeError(edge, "but is backedge, continueing");
continue;
}
// If the edges value is missing (and this is no loop header, and this is
// no backedge) return, this block is currently non estimatable.
if (w == MissingValue) {
printEdgeError(edge, "returning");
return;
}
}
if (getExecutionCount(BB) != MissingValue) {
BBWeight = getExecutionCount(BB);
}
// Fetch all necessary information for current block.
SmallVector<Edge, 8> ExitEdges;
SmallVector<Edge, 8> Edges;
if (BBLoop) {
BBLoop->getExitEdges(ExitEdges);
}
// If this is a loop header, consider the following:
// Exactly the flow that is entering this block, must exit this block too. So
// do the following:
// *) get all the exit edges, read the flow that is already leaving this
// loop, remember the edges that do not have any flow on them right now.
// (The edges that have already flow on them are most likely exiting edges of
// other loops, do not touch those flows because the previously caclulated
// loopheaders would not be exact anymore.)
// *) In case there is not a single exiting edge left, create one at the loop
// latch to prevent the flow from building up in the loop.
// *) Take the flow that is not leaving the loop already and distribute it on
// the remaining exiting edges.
// (This ensures that all flow that enters the loop also leaves it.)
// *) Increase the flow into the loop by increasing the weight of this block.
// There is at least one incoming backedge that will bring us this flow later
// on. (So that the flow condition in this node is valid again.)
if (BBisHeader) {
double incoming = BBWeight;
// Subtract the flow leaving the loop.
std::set<Edge> ProcessedExits;
for (SmallVector<Edge, 8>::iterator ei = ExitEdges.begin(),
ee = ExitEdges.end(); ei != ee; ++ei) {
if (ProcessedExits.insert(*ei).second) {
double w = getEdgeWeight(*ei);
if (w == MissingValue) {
Edges.push_back(*ei);
// Check if there is a necessary minimal weight, if yes, subtract it
// from weight.
if (MinimalWeight.find(*ei) != MinimalWeight.end()) {
incoming -= MinimalWeight[*ei];
DEBUG(dbgs() << "Reserving " << format("%.20g",MinimalWeight[*ei]) << " at " << (*ei) << "\n");
}
} else {
incoming -= w;
}
}
}
// If no exit edges, create one:
if (Edges.size() == 0) {
BasicBlock *Latch = BBLoop->getLoopLatch();
if (Latch) {
Edge edge = getEdge(Latch,0);
EdgeInformation[BB->getParent()][edge] = BBWeight;
printEdgeWeight(edge);
edge = getEdge(Latch, BB);
EdgeInformation[BB->getParent()][edge] = BBWeight * ExecCount;
printEdgeWeight(edge);
}
}
// Distribute remaining weight to the exting edges. To prevent fractions
// from building up and provoking precision problems the weight which is to
// be distributed is split and the rounded, the last edge gets a somewhat
// bigger value, but we are close enough for an estimation.
double fraction = floor(incoming/Edges.size());
for (SmallVector<Edge, 8>::iterator ei = Edges.begin(), ee = Edges.end();
ei != ee; ++ei) {
double w = 0;
if (ei != (ee-1)) {
w = fraction;
incoming -= fraction;
} else {
w = incoming;
}
EdgeInformation[BB->getParent()][*ei] += w;
// Read necessary minimal weight.
if (MinimalWeight.find(*ei) != MinimalWeight.end()) {
EdgeInformation[BB->getParent()][*ei] += MinimalWeight[*ei];
DEBUG(dbgs() << "Additionally " << format("%.20g",MinimalWeight[*ei]) << " at " << (*ei) << "\n");
}
printEdgeWeight(*ei);
// Add minimal weight to paths to all exit edges, this is used to ensure
// that enough flow is reaching this edges.
Path p;
const BasicBlock *Dest = GetPath(BB, (*ei).first, p, GetPathToDest);
while (Dest != BB) {
const BasicBlock *Parent = p.find(Dest)->second;
Edge e = getEdge(Parent, Dest);
if (MinimalWeight.find(e) == MinimalWeight.end()) {
MinimalWeight[e] = 0;
}
MinimalWeight[e] += w;
DEBUG(dbgs() << "Minimal Weight for " << e << ": " << format("%.20g",MinimalWeight[e]) << "\n");
Dest = Parent;
}
}
// Increase flow into the loop.
BBWeight *= (ExecCount+1);
}
BlockInformation[BB->getParent()][BB] = BBWeight;
// Up until now we considered only the loop exiting edges, now we have a
// definite block weight and must distribute this onto the outgoing edges.
// Since there may be already flow attached to some of the edges, read this
// flow first and remember the edges that have still now flow attached.
Edges.clear();
std::set<BasicBlock*> ProcessedSuccs;
succ_iterator bbi = succ_begin(BB), bbe = succ_end(BB);
// Also check for (BB,0) edges that may already contain some flow. (But only
// in case there are no successors.)
if (bbi == bbe) {
Edge edge = getEdge(BB,0);
EdgeInformation[BB->getParent()][edge] = BBWeight;
printEdgeWeight(edge);
}
for ( ; bbi != bbe; ++bbi ) {
if (ProcessedSuccs.insert(*bbi).second) {
Edge edge = getEdge(BB,*bbi);
double w = getEdgeWeight(edge);
if (w != MissingValue) {
BBWeight -= getEdgeWeight(edge);
} else {
Edges.push_back(edge);
// If minimal weight is necessary, reserve weight by subtracting weight
// from block weight, this is readded later on.
if (MinimalWeight.find(edge) != MinimalWeight.end()) {
BBWeight -= MinimalWeight[edge];
DEBUG(dbgs() << "Reserving " << format("%.20g",MinimalWeight[edge]) << " at " << edge << "\n");
}
}
}
}
double fraction = floor(BBWeight/Edges.size());
// Finally we know what flow is still not leaving the block, distribute this
// flow onto the empty edges.
for (SmallVector<Edge, 8>::iterator ei = Edges.begin(), ee = Edges.end();
ei != ee; ++ei) {
if (ei != (ee-1)) {
EdgeInformation[BB->getParent()][*ei] += fraction;
BBWeight -= fraction;
} else {
EdgeInformation[BB->getParent()][*ei] += BBWeight;
}
// Readd minial necessary weight.
if (MinimalWeight.find(*ei) != MinimalWeight.end()) {
EdgeInformation[BB->getParent()][*ei] += MinimalWeight[*ei];
DEBUG(dbgs() << "Additionally " << format("%.20g",MinimalWeight[*ei]) << " at " << (*ei) << "\n");
}
printEdgeWeight(*ei);
}
// This block is visited, mark this before the recursion.
BBToVisit.erase(BB);
// Recurse into successors.
for (succ_iterator bbi = succ_begin(BB), bbe = succ_end(BB);
bbi != bbe; ++bbi) {
recurseBasicBlock(*bbi);
}
}
/// Parse a function definition signature.
/// func-signature:
/// func-arguments func-throws? func-signature-result?
/// func-signature-result:
/// '->' type
///
/// Note that this leaves retType as null if unspecified.
ParserStatus
Parser::parseFunctionSignature(Identifier SimpleName,
DeclName &FullName,
SmallVectorImpl<ParameterList*> &bodyParams,
DefaultArgumentInfo &defaultArgs,
SourceLoc &throwsLoc,
bool &rethrows,
TypeRepr *&retType) {
SmallVector<Identifier, 4> NamePieces;
NamePieces.push_back(SimpleName);
FullName = SimpleName;
ParserStatus Status;
// We force first type of a func declaration to be a tuple for consistency.
if (Tok.is(tok::l_paren)) {
ParameterContextKind paramContext;
if (SimpleName.isOperator())
paramContext = ParameterContextKind::Operator;
else
paramContext = ParameterContextKind::Function;
Status = parseFunctionArguments(NamePieces, bodyParams, paramContext,
defaultArgs);
FullName = DeclName(Context, SimpleName,
llvm::makeArrayRef(NamePieces.begin() + 1,
NamePieces.end()));
if (bodyParams.empty()) {
// If we didn't get anything, add a () pattern to avoid breaking
// invariants.
assert(Status.hasCodeCompletion() || Status.isError());
bodyParams.push_back(ParameterList::createEmpty(Context));
}
} else {
diagnose(Tok, diag::func_decl_without_paren);
Status = makeParserError();
// Recover by creating a '() -> ?' signature.
bodyParams.push_back(ParameterList::createEmpty(Context, PreviousLoc,
PreviousLoc));
FullName = DeclName(Context, SimpleName, bodyParams.back());
}
// Check for the 'throws' keyword.
rethrows = false;
if (Tok.is(tok::kw_throws)) {
throwsLoc = consumeToken();
} else if (Tok.is(tok::kw_rethrows)) {
throwsLoc = consumeToken();
rethrows = true;
} else if (Tok.is(tok::kw_throw)) {
throwsLoc = consumeToken();
diagnose(throwsLoc, diag::throw_in_function_type)
.fixItReplace(throwsLoc, "throws");
}
SourceLoc arrowLoc;
// If there's a trailing arrow, parse the rest as the result type.
if (Tok.isAny(tok::arrow, tok::colon)) {
if (!consumeIf(tok::arrow, arrowLoc)) {
// FixIt ':' to '->'.
diagnose(Tok, diag::func_decl_expected_arrow)
.fixItReplace(SourceRange(Tok.getLoc()), "->");
arrowLoc = consumeToken(tok::colon);
}
ParserResult<TypeRepr> ResultType =
parseType(diag::expected_type_function_result);
if (ResultType.hasCodeCompletion())
return ResultType;
retType = ResultType.getPtrOrNull();
if (!retType) {
Status.setIsParseError();
return Status;
}
} else {
// Otherwise, we leave retType null.
retType = nullptr;
}
// Check for 'throws' and 'rethrows' after the type and correct it.
if (!throwsLoc.isValid()) {
if (Tok.is(tok::kw_throws)) {
throwsLoc = consumeToken();
} else if (Tok.is(tok::kw_rethrows)) {
throwsLoc = consumeToken();
rethrows = true;
}
if (throwsLoc.isValid()) {
assert(arrowLoc.isValid());
assert(retType);
auto diag = rethrows ? diag::rethrows_after_function_result
: diag::throws_after_function_result;
SourceLoc typeEndLoc = Lexer::getLocForEndOfToken(SourceMgr,
retType->getEndLoc());
SourceLoc throwsEndLoc = Lexer::getLocForEndOfToken(SourceMgr, throwsLoc);
diagnose(Tok, diag)
.fixItInsert(arrowLoc, rethrows ? "rethrows " : "throws ")
.fixItRemoveChars(typeEndLoc, throwsEndLoc);
}
}
return Status;
}
bool LoopInterchangeTransform::adjustLoopBranches() {
DEBUG(dbgs() << "adjustLoopBranches called\n");
// Adjust the loop preheader
BasicBlock *InnerLoopHeader = InnerLoop->getHeader();
BasicBlock *OuterLoopHeader = OuterLoop->getHeader();
BasicBlock *InnerLoopLatch = InnerLoop->getLoopLatch();
BasicBlock *OuterLoopLatch = OuterLoop->getLoopLatch();
BasicBlock *OuterLoopPreHeader = OuterLoop->getLoopPreheader();
BasicBlock *InnerLoopPreHeader = InnerLoop->getLoopPreheader();
BasicBlock *OuterLoopPredecessor = OuterLoopPreHeader->getUniquePredecessor();
BasicBlock *InnerLoopLatchPredecessor =
InnerLoopLatch->getUniquePredecessor();
BasicBlock *InnerLoopLatchSuccessor;
BasicBlock *OuterLoopLatchSuccessor;
BranchInst *OuterLoopLatchBI =
dyn_cast<BranchInst>(OuterLoopLatch->getTerminator());
BranchInst *InnerLoopLatchBI =
dyn_cast<BranchInst>(InnerLoopLatch->getTerminator());
BranchInst *OuterLoopHeaderBI =
dyn_cast<BranchInst>(OuterLoopHeader->getTerminator());
BranchInst *InnerLoopHeaderBI =
dyn_cast<BranchInst>(InnerLoopHeader->getTerminator());
if (!OuterLoopPredecessor || !InnerLoopLatchPredecessor ||
!OuterLoopLatchBI || !InnerLoopLatchBI || !OuterLoopHeaderBI ||
!InnerLoopHeaderBI)
return false;
BranchInst *InnerLoopLatchPredecessorBI =
dyn_cast<BranchInst>(InnerLoopLatchPredecessor->getTerminator());
BranchInst *OuterLoopPredecessorBI =
dyn_cast<BranchInst>(OuterLoopPredecessor->getTerminator());
if (!OuterLoopPredecessorBI || !InnerLoopLatchPredecessorBI)
return false;
BasicBlock *InnerLoopHeaderSuccessor = InnerLoopHeader->getUniqueSuccessor();
if (!InnerLoopHeaderSuccessor)
return false;
// Adjust Loop Preheader and headers
unsigned NumSucc = OuterLoopPredecessorBI->getNumSuccessors();
for (unsigned i = 0; i < NumSucc; ++i) {
if (OuterLoopPredecessorBI->getSuccessor(i) == OuterLoopPreHeader)
OuterLoopPredecessorBI->setSuccessor(i, InnerLoopPreHeader);
}
NumSucc = OuterLoopHeaderBI->getNumSuccessors();
for (unsigned i = 0; i < NumSucc; ++i) {
if (OuterLoopHeaderBI->getSuccessor(i) == OuterLoopLatch)
OuterLoopHeaderBI->setSuccessor(i, LoopExit);
else if (OuterLoopHeaderBI->getSuccessor(i) == InnerLoopPreHeader)
OuterLoopHeaderBI->setSuccessor(i, InnerLoopHeaderSuccessor);
}
// Adjust reduction PHI's now that the incoming block has changed.
updateIncomingBlock(InnerLoopHeaderSuccessor, InnerLoopHeader,
OuterLoopHeader);
BranchInst::Create(OuterLoopPreHeader, InnerLoopHeaderBI);
InnerLoopHeaderBI->eraseFromParent();
// -------------Adjust loop latches-----------
if (InnerLoopLatchBI->getSuccessor(0) == InnerLoopHeader)
InnerLoopLatchSuccessor = InnerLoopLatchBI->getSuccessor(1);
else
InnerLoopLatchSuccessor = InnerLoopLatchBI->getSuccessor(0);
NumSucc = InnerLoopLatchPredecessorBI->getNumSuccessors();
for (unsigned i = 0; i < NumSucc; ++i) {
if (InnerLoopLatchPredecessorBI->getSuccessor(i) == InnerLoopLatch)
InnerLoopLatchPredecessorBI->setSuccessor(i, InnerLoopLatchSuccessor);
}
// Adjust PHI nodes in InnerLoopLatchSuccessor. Update all uses of PHI with
// the value and remove this PHI node from inner loop.
SmallVector<PHINode *, 8> LcssaVec;
for (auto I = InnerLoopLatchSuccessor->begin(); isa<PHINode>(I); ++I) {
PHINode *LcssaPhi = cast<PHINode>(I);
LcssaVec.push_back(LcssaPhi);
}
for (auto I = LcssaVec.begin(), E = LcssaVec.end(); I != E; ++I) {
PHINode *P = *I;
Value *Incoming = P->getIncomingValueForBlock(InnerLoopLatch);
P->replaceAllUsesWith(Incoming);
P->eraseFromParent();
}
if (OuterLoopLatchBI->getSuccessor(0) == OuterLoopHeader)
OuterLoopLatchSuccessor = OuterLoopLatchBI->getSuccessor(1);
else
OuterLoopLatchSuccessor = OuterLoopLatchBI->getSuccessor(0);
if (InnerLoopLatchBI->getSuccessor(1) == InnerLoopLatchSuccessor)
InnerLoopLatchBI->setSuccessor(1, OuterLoopLatchSuccessor);
else
InnerLoopLatchBI->setSuccessor(0, OuterLoopLatchSuccessor);
updateIncomingBlock(OuterLoopLatchSuccessor, OuterLoopLatch, InnerLoopLatch);
if (OuterLoopLatchBI->getSuccessor(0) == OuterLoopLatchSuccessor) {
OuterLoopLatchBI->setSuccessor(0, InnerLoopLatch);
} else {
OuterLoopLatchBI->setSuccessor(1, InnerLoopLatch);
}
return true;
}
/// InsertUnwindResumeCalls - Convert the ResumeInsts that are still present
/// into calls to the appropriate _Unwind_Resume function.
bool DwarfEHPrepare::InsertUnwindResumeCalls(Function &Fn) {
bool UsesNewEH = false;
SmallVector<ResumeInst*, 16> Resumes;
for (Function::iterator I = Fn.begin(), E = Fn.end(); I != E; ++I) {
TerminatorInst *TI = I->getTerminator();
if (ResumeInst *RI = dyn_cast<ResumeInst>(TI))
Resumes.push_back(RI);
else if (InvokeInst *II = dyn_cast<InvokeInst>(TI))
UsesNewEH = II->getUnwindDest()->isLandingPad();
}
if (Resumes.empty())
return UsesNewEH;
// Find the rewind function if we didn't already.
if (!RewindFunction) {
LLVMContext &Ctx = Resumes[0]->getContext();
FunctionType *FTy = FunctionType::get(Type::getVoidTy(Ctx),
Type::getInt8PtrTy(Ctx), false);
const char *RewindName = TLI->getLibcallName(RTLIB::UNWIND_RESUME);
RewindFunction = Fn.getParent()->getOrInsertFunction(RewindName, FTy);
}
// Create the basic block where the _Unwind_Resume call will live.
LLVMContext &Ctx = Fn.getContext();
unsigned ResumesSize = Resumes.size();
if (ResumesSize == 1) {
// Instead of creating a new BB and PHI node, just append the call to
// _Unwind_Resume to the end of the single resume block.
ResumeInst *RI = Resumes.front();
BasicBlock *UnwindBB = RI->getParent();
Value *ExnObj = GetExceptionObject(RI);
// Call the _Unwind_Resume function.
CallInst *CI = CallInst::Create(RewindFunction, ExnObj, "", UnwindBB);
CI->setCallingConv(TLI->getLibcallCallingConv(RTLIB::UNWIND_RESUME));
// We never expect _Unwind_Resume to return.
new UnreachableInst(Ctx, UnwindBB);
return true;
}
BasicBlock *UnwindBB = BasicBlock::Create(Ctx, "unwind_resume", &Fn);
PHINode *PN = PHINode::Create(Type::getInt8PtrTy(Ctx), ResumesSize,
"exn.obj", UnwindBB);
// Extract the exception object from the ResumeInst and add it to the PHI node
// that feeds the _Unwind_Resume call.
for (SmallVectorImpl<ResumeInst*>::iterator
I = Resumes.begin(), E = Resumes.end(); I != E; ++I) {
ResumeInst *RI = *I;
BasicBlock *Parent = RI->getParent();
BranchInst::Create(UnwindBB, Parent);
Value *ExnObj = GetExceptionObject(RI);
PN->addIncoming(ExnObj, Parent);
++NumResumesLowered;
}
// Call the function.
CallInst *CI = CallInst::Create(RewindFunction, PN, "", UnwindBB);
CI->setCallingConv(TLI->getLibcallCallingConv(RTLIB::UNWIND_RESUME));
// We never expect _Unwind_Resume to return.
new UnreachableInst(Ctx, UnwindBB);
return true;
}
unsigned char* JITDwarfEmitter::EmitExceptionTable(MachineFunction* MF,
unsigned char* StartFunction,
unsigned char* EndFunction) const {
assert(MMI && "MachineModuleInfo not registered!");
// Map all labels and get rid of any dead landing pads.
MMI->TidyLandingPads(JCE->getLabelLocations());
const std::vector<const GlobalVariable *> &TypeInfos = MMI->getTypeInfos();
const std::vector<unsigned> &FilterIds = MMI->getFilterIds();
const std::vector<LandingPadInfo> &PadInfos = MMI->getLandingPads();
if (PadInfos.empty()) return 0;
// Sort the landing pads in order of their type ids. This is used to fold
// duplicate actions.
SmallVector<const LandingPadInfo *, 64> LandingPads;
LandingPads.reserve(PadInfos.size());
for (unsigned i = 0, N = PadInfos.size(); i != N; ++i)
LandingPads.push_back(&PadInfos[i]);
std::sort(LandingPads.begin(), LandingPads.end(), PadLT);
// Negative type ids index into FilterIds, positive type ids index into
// TypeInfos. The value written for a positive type id is just the type
// id itself. For a negative type id, however, the value written is the
// (negative) byte offset of the corresponding FilterIds entry. The byte
// offset is usually equal to the type id, because the FilterIds entries
// are written using a variable width encoding which outputs one byte per
// entry as long as the value written is not too large, but can differ.
// This kind of complication does not occur for positive type ids because
// type infos are output using a fixed width encoding.
// FilterOffsets[i] holds the byte offset corresponding to FilterIds[i].
SmallVector<int, 16> FilterOffsets;
FilterOffsets.reserve(FilterIds.size());
int Offset = -1;
for(std::vector<unsigned>::const_iterator I = FilterIds.begin(),
E = FilterIds.end(); I != E; ++I) {
FilterOffsets.push_back(Offset);
Offset -= MCAsmInfo::getULEB128Size(*I);
}
// Compute the actions table and gather the first action index for each
// landing pad site.
SmallVector<ActionEntry, 32> Actions;
SmallVector<unsigned, 64> FirstActions;
FirstActions.reserve(LandingPads.size());
int FirstAction = 0;
unsigned SizeActions = 0;
for (unsigned i = 0, N = LandingPads.size(); i != N; ++i) {
const LandingPadInfo *LP = LandingPads[i];
const std::vector<int> &TypeIds = LP->TypeIds;
const unsigned NumShared = i ? SharedTypeIds(LP, LandingPads[i-1]) : 0;
unsigned SizeSiteActions = 0;
if (NumShared < TypeIds.size()) {
unsigned SizeAction = 0;
ActionEntry *PrevAction = 0;
if (NumShared) {
const unsigned SizePrevIds = LandingPads[i-1]->TypeIds.size();
assert(Actions.size());
PrevAction = &Actions.back();
SizeAction = MCAsmInfo::getSLEB128Size(PrevAction->NextAction) +
MCAsmInfo::getSLEB128Size(PrevAction->ValueForTypeID);
for (unsigned j = NumShared; j != SizePrevIds; ++j) {
SizeAction -= MCAsmInfo::getSLEB128Size(PrevAction->ValueForTypeID);
SizeAction += -PrevAction->NextAction;
PrevAction = PrevAction->Previous;
}
}
// Compute the actions.
for (unsigned I = NumShared, M = TypeIds.size(); I != M; ++I) {
int TypeID = TypeIds[I];
assert(-1-TypeID < (int)FilterOffsets.size() && "Unknown filter id!");
int ValueForTypeID = TypeID < 0 ? FilterOffsets[-1 - TypeID] : TypeID;
unsigned SizeTypeID = MCAsmInfo::getSLEB128Size(ValueForTypeID);
int NextAction = SizeAction ? -(SizeAction + SizeTypeID) : 0;
SizeAction = SizeTypeID + MCAsmInfo::getSLEB128Size(NextAction);
SizeSiteActions += SizeAction;
ActionEntry Action = {ValueForTypeID, NextAction, PrevAction};
Actions.push_back(Action);
PrevAction = &Actions.back();
}
// Record the first action of the landing pad site.
FirstAction = SizeActions + SizeSiteActions - SizeAction + 1;
} // else identical - re-use previous FirstAction
FirstActions.push_back(FirstAction);
// Compute this sites contribution to size.
SizeActions += SizeSiteActions;
}
// Compute the call-site table. Entries must be ordered by address.
SmallVector<CallSiteEntry, 64> CallSites;
RangeMapType PadMap;
for (unsigned i = 0, N = LandingPads.size(); i != N; ++i) {
const LandingPadInfo *LandingPad = LandingPads[i];
for (unsigned j=0, E = LandingPad->BeginLabels.size(); j != E; ++j) {
MCSymbol *BeginLabel = LandingPad->BeginLabels[j];
assert(!PadMap.count(BeginLabel) && "Duplicate landing pad labels!");
PadRange P = { i, j };
PadMap[BeginLabel] = P;
}
}
bool MayThrow = false;
MCSymbol *LastLabel = 0;
for (MachineFunction::const_iterator I = MF->begin(), E = MF->end();
I != E; ++I) {
for (MachineBasicBlock::const_iterator MI = I->begin(), E = I->end();
MI != E; ++MI) {
if (!MI->isLabel()) {
MayThrow |= MI->isCall();
continue;
}
MCSymbol *BeginLabel = MI->getOperand(0).getMCSymbol();
assert(BeginLabel && "Invalid label!");
if (BeginLabel == LastLabel)
MayThrow = false;
RangeMapType::iterator L = PadMap.find(BeginLabel);
if (L == PadMap.end())
continue;
PadRange P = L->second;
const LandingPadInfo *LandingPad = LandingPads[P.PadIndex];
assert(BeginLabel == LandingPad->BeginLabels[P.RangeIndex] &&
"Inconsistent landing pad map!");
// If some instruction between the previous try-range and this one may
// throw, create a call-site entry with no landing pad for the region
// between the try-ranges.
if (MayThrow) {
CallSiteEntry Site = {LastLabel, BeginLabel, 0, 0};
CallSites.push_back(Site);
}
LastLabel = LandingPad->EndLabels[P.RangeIndex];
CallSiteEntry Site = {BeginLabel, LastLabel,
LandingPad->LandingPadLabel, FirstActions[P.PadIndex]};
assert(Site.BeginLabel && Site.EndLabel && Site.PadLabel &&
"Invalid landing pad!");
// Try to merge with the previous call-site.
if (CallSites.size()) {
CallSiteEntry &Prev = CallSites.back();
if (Site.PadLabel == Prev.PadLabel && Site.Action == Prev.Action) {
// Extend the range of the previous entry.
Prev.EndLabel = Site.EndLabel;
continue;
}
}
// Otherwise, create a new call-site.
CallSites.push_back(Site);
}
}
// If some instruction between the previous try-range and the end of the
// function may throw, create a call-site entry with no landing pad for the
// region following the try-range.
if (MayThrow) {
CallSiteEntry Site = {LastLabel, 0, 0, 0};
CallSites.push_back(Site);
}
// Final tallies.
unsigned SizeSites = CallSites.size() * (sizeof(int32_t) + // Site start.
sizeof(int32_t) + // Site length.
sizeof(int32_t)); // Landing pad.
for (unsigned i = 0, e = CallSites.size(); i < e; ++i)
SizeSites += MCAsmInfo::getULEB128Size(CallSites[i].Action);
unsigned SizeTypes = TypeInfos.size() * TD->getPointerSize();
unsigned TypeOffset = sizeof(int8_t) + // Call site format
// Call-site table length
MCAsmInfo::getULEB128Size(SizeSites) +
SizeSites + SizeActions + SizeTypes;
// Begin the exception table.
JCE->emitAlignmentWithFill(4, 0);
// Asm->EOL("Padding");
unsigned char* DwarfExceptionTable = (unsigned char*)JCE->getCurrentPCValue();
// Emit the header.
JCE->emitByte(dwarf::DW_EH_PE_omit);
// Asm->EOL("LPStart format (DW_EH_PE_omit)");
JCE->emitByte(dwarf::DW_EH_PE_absptr);
// Asm->EOL("TType format (DW_EH_PE_absptr)");
JCE->emitULEB128Bytes(TypeOffset);
// Asm->EOL("TType base offset");
JCE->emitByte(dwarf::DW_EH_PE_udata4);
// Asm->EOL("Call site format (DW_EH_PE_udata4)");
JCE->emitULEB128Bytes(SizeSites);
// Asm->EOL("Call-site table length");
// Emit the landing pad site information.
for (unsigned i = 0; i < CallSites.size(); ++i) {
CallSiteEntry &S = CallSites[i];
intptr_t BeginLabelPtr = 0;
intptr_t EndLabelPtr = 0;
if (!S.BeginLabel) {
BeginLabelPtr = (intptr_t)StartFunction;
JCE->emitInt32(0);
} else {
BeginLabelPtr = JCE->getLabelAddress(S.BeginLabel);
JCE->emitInt32(BeginLabelPtr - (intptr_t)StartFunction);
}
// Asm->EOL("Region start");
if (!S.EndLabel)
EndLabelPtr = (intptr_t)EndFunction;
else
EndLabelPtr = JCE->getLabelAddress(S.EndLabel);
JCE->emitInt32(EndLabelPtr - BeginLabelPtr);
//Asm->EOL("Region length");
if (!S.PadLabel) {
JCE->emitInt32(0);
} else {
unsigned PadLabelPtr = JCE->getLabelAddress(S.PadLabel);
JCE->emitInt32(PadLabelPtr - (intptr_t)StartFunction);
}
// Asm->EOL("Landing pad");
JCE->emitULEB128Bytes(S.Action);
// Asm->EOL("Action");
}
// Emit the actions.
for (unsigned I = 0, N = Actions.size(); I != N; ++I) {
ActionEntry &Action = Actions[I];
JCE->emitSLEB128Bytes(Action.ValueForTypeID);
//Asm->EOL("TypeInfo index");
JCE->emitSLEB128Bytes(Action.NextAction);
//Asm->EOL("Next action");
}
// Emit the type ids.
for (unsigned M = TypeInfos.size(); M; --M) {
const GlobalVariable *GV = TypeInfos[M - 1];
if (GV) {
if (TD->getPointerSize() == sizeof(int32_t))
JCE->emitInt32((intptr_t)Jit.getOrEmitGlobalVariable(GV));
else
JCE->emitInt64((intptr_t)Jit.getOrEmitGlobalVariable(GV));
} else {
if (TD->getPointerSize() == sizeof(int32_t))
JCE->emitInt32(0);
else
JCE->emitInt64(0);
}
// Asm->EOL("TypeInfo");
}
// Emit the filter typeids.
for (unsigned j = 0, M = FilterIds.size(); j < M; ++j) {
unsigned TypeID = FilterIds[j];
JCE->emitULEB128Bytes(TypeID);
//Asm->EOL("Filter TypeInfo index");
}
JCE->emitAlignmentWithFill(4, 0);
return DwarfExceptionTable;
}
static void lookupInModule(ModuleDecl *module, ModuleDecl::AccessPathTy accessPath,
SmallVectorImpl<ValueDecl *> &decls,
ResolutionKind resolutionKind, bool canReturnEarly,
LazyResolver *typeResolver,
ModuleLookupCache &cache,
const DeclContext *moduleScopeContext,
bool respectAccessControl,
ArrayRef<ModuleDecl::ImportedModule> extraImports,
CallbackTy callback) {
assert(module);
assert(std::none_of(extraImports.begin(), extraImports.end(),
[](ModuleDecl::ImportedModule import) -> bool {
return !import.second;
}));
ModuleLookupCache::iterator iter;
bool isNew;
std::tie(iter, isNew) = cache.insert({{accessPath, module}, {}});
if (!isNew) {
decls.append(iter->second.begin(), iter->second.end());
return;
}
size_t initialCount = decls.size();
SmallVector<ValueDecl *, 4> localDecls;
callback(module, accessPath, localDecls);
if (respectAccessControl) {
auto newEndIter = std::remove_if(localDecls.begin(), localDecls.end(),
[=](ValueDecl *VD) {
return !VD->isAccessibleFrom(moduleScopeContext);
});
localDecls.erase(newEndIter, localDecls.end());
// This only applies to immediate imports of the top-level module.
if (moduleScopeContext && moduleScopeContext->getParentModule() != module)
moduleScopeContext = nullptr;
}
OverloadSetTy overloads;
resolutionKind = recordImportDecls(typeResolver, decls, localDecls,
overloads, resolutionKind);
bool foundDecls = decls.size() > initialCount;
if (!foundDecls || !canReturnEarly ||
resolutionKind == ResolutionKind::Overloadable) {
SmallVector<ModuleDecl::ImportedModule, 8> reexports;
module->getImportedModulesForLookup(reexports);
assert(std::none_of(reexports.begin(), reexports.end(),
[](ModuleDecl::ImportedModule import) -> bool {
return !import.second;
}));
reexports.append(extraImports.begin(), extraImports.end());
// Prefer scoped imports (import func Swift.max) to whole-module imports.
SmallVector<ValueDecl *, 8> unscopedValues;
SmallVector<ValueDecl *, 8> scopedValues;
for (auto next : reexports) {
// Filter any whole-module imports, and skip specific-decl imports if the
// import path doesn't match exactly.
ModuleDecl::AccessPathTy combinedAccessPath;
if (accessPath.empty()) {
combinedAccessPath = next.first;
} else if (!next.first.empty() &&
!ModuleDecl::isSameAccessPath(next.first, accessPath)) {
// If we ever allow importing non-top-level decls, it's possible the
// rule above isn't what we want.
assert(next.first.size() == 1 && "import of non-top-level decl");
continue;
} else {
combinedAccessPath = accessPath;
}
auto &resultSet = next.first.empty() ? unscopedValues : scopedValues;
lookupInModule<OverloadSetTy>(next.second, combinedAccessPath,
resultSet, resolutionKind, canReturnEarly,
typeResolver, cache, moduleScopeContext,
respectAccessControl, {}, callback);
}
// Add the results from scoped imports.
resolutionKind = recordImportDecls(typeResolver, decls, scopedValues,
overloads, resolutionKind);
// Add the results from unscoped imports.
foundDecls = decls.size() > initialCount;
if (!foundDecls || !canReturnEarly ||
resolutionKind == ResolutionKind::Overloadable) {
resolutionKind = recordImportDecls(typeResolver, decls, unscopedValues,
overloads, resolutionKind);
}
}
// Remove duplicated declarations.
llvm::SmallPtrSet<ValueDecl *, 4> knownDecls;
decls.erase(std::remove_if(decls.begin() + initialCount, decls.end(),
[&](ValueDecl *d) -> bool {
return !knownDecls.insert(d).second;
}),
decls.end());
auto &cachedValues = cache[{accessPath, module}];
cachedValues.insert(cachedValues.end(),
decls.begin() + initialCount,
decls.end());
}
/// Emit landing pads and actions.
///
/// The general organization of the table is complex, but the basic concepts are
/// easy. First there is a header which describes the location and organization
/// of the three components that follow.
///
/// 1. The landing pad site information describes the range of code covered by
/// the try. In our case it's an accumulation of the ranges covered by the
/// invokes in the try. There is also a reference to the landing pad that
/// handles the exception once processed. Finally an index into the actions
/// table.
/// 2. The action table, in our case, is composed of pairs of type IDs and next
/// action offset. Starting with the action index from the landing pad
/// site, each type ID is checked for a match to the current exception. If
/// it matches then the exception and type id are passed on to the landing
/// pad. Otherwise the next action is looked up. This chain is terminated
/// with a next action of zero. If no type id is found then the frame is
/// unwound and handling continues.
/// 3. Type ID table contains references to all the C++ typeinfo for all
/// catches in the function. This tables is reverse indexed base 1.
void EHStreamer::emitExceptionTable() {
const std::vector<const GlobalValue *> &TypeInfos = MMI->getTypeInfos();
const std::vector<unsigned> &FilterIds = MMI->getFilterIds();
const std::vector<LandingPadInfo> &PadInfos = MMI->getLandingPads();
// Sort the landing pads in order of their type ids. This is used to fold
// duplicate actions.
SmallVector<const LandingPadInfo *, 64> LandingPads;
LandingPads.reserve(PadInfos.size());
for (unsigned i = 0, N = PadInfos.size(); i != N; ++i)
LandingPads.push_back(&PadInfos[i]);
// Order landing pads lexicographically by type id.
std::sort(LandingPads.begin(), LandingPads.end(),
[](const LandingPadInfo *L,
const LandingPadInfo *R) { return L->TypeIds < R->TypeIds; });
// Compute the actions table and gather the first action index for each
// landing pad site.
SmallVector<ActionEntry, 32> Actions;
SmallVector<unsigned, 64> FirstActions;
unsigned SizeActions =
computeActionsTable(LandingPads, Actions, FirstActions);
// Compute the call-site table.
SmallVector<CallSiteEntry, 64> CallSites;
computeCallSiteTable(CallSites, LandingPads, FirstActions);
// Final tallies.
// Call sites.
bool IsSJLJ = Asm->MAI->getExceptionHandlingType() == ExceptionHandling::SjLj;
bool HaveTTData = IsSJLJ ? (!TypeInfos.empty() || !FilterIds.empty()) : true;
unsigned CallSiteTableLength;
if (IsSJLJ)
CallSiteTableLength = 0;
else {
unsigned SiteStartSize = 4; // dwarf::DW_EH_PE_udata4
unsigned SiteLengthSize = 4; // dwarf::DW_EH_PE_udata4
unsigned LandingPadSize = 4; // dwarf::DW_EH_PE_udata4
CallSiteTableLength =
CallSites.size() * (SiteStartSize + SiteLengthSize + LandingPadSize);
}
for (unsigned i = 0, e = CallSites.size(); i < e; ++i) {
CallSiteTableLength += getULEB128Size(CallSites[i].Action);
if (IsSJLJ)
CallSiteTableLength += getULEB128Size(i);
}
// Type infos.
const MCSection *LSDASection = Asm->getObjFileLowering().getLSDASection();
unsigned TTypeEncoding;
unsigned TypeFormatSize;
if (!HaveTTData) {
// For SjLj exceptions, if there is no TypeInfo, then we just explicitly say
// that we're omitting that bit.
TTypeEncoding = dwarf::DW_EH_PE_omit;
// dwarf::DW_EH_PE_absptr
TypeFormatSize = Asm->getDataLayout().getPointerSize();
} else {
// Okay, we have actual filters or typeinfos to emit. As such, we need to
// pick a type encoding for them. We're about to emit a list of pointers to
// typeinfo objects at the end of the LSDA. However, unless we're in static
// mode, this reference will require a relocation by the dynamic linker.
//
// Because of this, we have a couple of options:
//
// 1) If we are in -static mode, we can always use an absolute reference
// from the LSDA, because the static linker will resolve it.
//
// 2) Otherwise, if the LSDA section is writable, we can output the direct
// reference to the typeinfo and allow the dynamic linker to relocate
// it. Since it is in a writable section, the dynamic linker won't
// have a problem.
//
// 3) Finally, if we're in PIC mode and the LDSA section isn't writable,
// we need to use some form of indirection. For example, on Darwin,
// we can output a statically-relocatable reference to a dyld stub. The
// offset to the stub is constant, but the contents are in a section
// that is updated by the dynamic linker. This is easy enough, but we
// need to tell the personality function of the unwinder to indirect
// through the dyld stub.
//
// FIXME: When (3) is actually implemented, we'll have to emit the stubs
// somewhere. This predicate should be moved to a shared location that is
// in target-independent code.
//
TTypeEncoding = Asm->getObjFileLowering().getTTypeEncoding();
TypeFormatSize = Asm->GetSizeOfEncodedValue(TTypeEncoding);
}
// Begin the exception table.
// Sometimes we want not to emit the data into separate section (e.g. ARM
// EHABI). In this case LSDASection will be NULL.
if (LSDASection)
Asm->OutStreamer.SwitchSection(LSDASection);
Asm->EmitAlignment(2);
// Emit the LSDA.
MCSymbol *GCCETSym =
Asm->OutContext.GetOrCreateSymbol(Twine("GCC_except_table")+
Twine(Asm->getFunctionNumber()));
Asm->OutStreamer.EmitLabel(GCCETSym);
Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("exception",
Asm->getFunctionNumber()));
if (IsSJLJ)
Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("_LSDA_",
Asm->getFunctionNumber()));
// Emit the LSDA header.
Asm->EmitEncodingByte(dwarf::DW_EH_PE_omit, "@LPStart");
Asm->EmitEncodingByte(TTypeEncoding, "@TType");
// The type infos need to be aligned. GCC does this by inserting padding just
// before the type infos. However, this changes the size of the exception
// table, so you need to take this into account when you output the exception
// table size. However, the size is output using a variable length encoding.
// So by increasing the size by inserting padding, you may increase the number
// of bytes used for writing the size. If it increases, say by one byte, then
// you now need to output one less byte of padding to get the type infos
// aligned. However this decreases the size of the exception table. This
// changes the value you have to output for the exception table size. Due to
// the variable length encoding, the number of bytes used for writing the
// length may decrease. If so, you then have to increase the amount of
// padding. And so on. If you look carefully at the GCC code you will see that
// it indeed does this in a loop, going on and on until the values stabilize.
// We chose another solution: don't output padding inside the table like GCC
// does, instead output it before the table.
unsigned SizeTypes = TypeInfos.size() * TypeFormatSize;
unsigned CallSiteTableLengthSize = getULEB128Size(CallSiteTableLength);
unsigned TTypeBaseOffset =
sizeof(int8_t) + // Call site format
CallSiteTableLengthSize + // Call site table length size
CallSiteTableLength + // Call site table length
SizeActions + // Actions size
SizeTypes;
unsigned TTypeBaseOffsetSize = getULEB128Size(TTypeBaseOffset);
unsigned TotalSize =
sizeof(int8_t) + // LPStart format
sizeof(int8_t) + // TType format
(HaveTTData ? TTypeBaseOffsetSize : 0) + // TType base offset size
TTypeBaseOffset; // TType base offset
unsigned SizeAlign = (4 - TotalSize) & 3;
if (HaveTTData) {
// Account for any extra padding that will be added to the call site table
// length.
Asm->EmitULEB128(TTypeBaseOffset, "@TType base offset", SizeAlign);
SizeAlign = 0;
}
bool VerboseAsm = Asm->OutStreamer.isVerboseAsm();
// SjLj Exception handling
if (IsSJLJ) {
Asm->EmitEncodingByte(dwarf::DW_EH_PE_udata4, "Call site");
// Add extra padding if it wasn't added to the TType base offset.
Asm->EmitULEB128(CallSiteTableLength, "Call site table length", SizeAlign);
// Emit the landing pad site information.
unsigned idx = 0;
for (SmallVectorImpl<CallSiteEntry>::const_iterator
I = CallSites.begin(), E = CallSites.end(); I != E; ++I, ++idx) {
const CallSiteEntry &S = *I;
// Offset of the landing pad, counted in 16-byte bundles relative to the
// @LPStart address.
if (VerboseAsm) {
Asm->OutStreamer.AddComment(">> Call Site " + Twine(idx) + " <<");
Asm->OutStreamer.AddComment(" On exception at call site "+Twine(idx));
}
Asm->EmitULEB128(idx);
// Offset of the first associated action record, relative to the start of
// the action table. This value is biased by 1 (1 indicates the start of
// the action table), and 0 indicates that there are no actions.
if (VerboseAsm) {
if (S.Action == 0)
Asm->OutStreamer.AddComment(" Action: cleanup");
else
Asm->OutStreamer.AddComment(" Action: " +
Twine((S.Action - 1) / 2 + 1));
}
Asm->EmitULEB128(S.Action);
}
} else {
// Itanium LSDA exception handling
// The call-site table is a list of all call sites that may throw an
// exception (including C++ 'throw' statements) in the procedure
// fragment. It immediately follows the LSDA header. Each entry indicates,
// for a given call, the first corresponding action record and corresponding
// landing pad.
//
// The table begins with the number of bytes, stored as an LEB128
// compressed, unsigned integer. The records immediately follow the record
// count. They are sorted in increasing call-site address. Each record
// indicates:
//
// * The position of the call-site.
// * The position of the landing pad.
// * The first action record for that call site.
//
// A missing entry in the call-site table indicates that a call is not
// supposed to throw.
// Emit the landing pad call site table.
Asm->EmitEncodingByte(dwarf::DW_EH_PE_udata4, "Call site");
// Add extra padding if it wasn't added to the TType base offset.
Asm->EmitULEB128(CallSiteTableLength, "Call site table length", SizeAlign);
unsigned Entry = 0;
for (SmallVectorImpl<CallSiteEntry>::const_iterator
I = CallSites.begin(), E = CallSites.end(); I != E; ++I) {
const CallSiteEntry &S = *I;
MCSymbol *EHFuncBeginSym =
Asm->GetTempSymbol("eh_func_begin", Asm->getFunctionNumber());
MCSymbol *BeginLabel = S.BeginLabel;
if (!BeginLabel)
BeginLabel = EHFuncBeginSym;
MCSymbol *EndLabel = S.EndLabel;
if (!EndLabel)
EndLabel = Asm->GetTempSymbol("eh_func_end", Asm->getFunctionNumber());
// Offset of the call site relative to the previous call site, counted in
// number of 16-byte bundles. The first call site is counted relative to
// the start of the procedure fragment.
if (VerboseAsm)
Asm->OutStreamer.AddComment(">> Call Site " + Twine(++Entry) + " <<");
Asm->EmitLabelDifference(BeginLabel, EHFuncBeginSym, 4);
if (VerboseAsm)
Asm->OutStreamer.AddComment(Twine(" Call between ") +
BeginLabel->getName() + " and " +
EndLabel->getName());
Asm->EmitLabelDifference(EndLabel, BeginLabel, 4);
// Offset of the landing pad, counted in 16-byte bundles relative to the
// @LPStart address.
if (!S.PadLabel) {
if (VerboseAsm)
Asm->OutStreamer.AddComment(" has no landing pad");
Asm->OutStreamer.EmitIntValue(0, 4/*size*/);
} else {
if (VerboseAsm)
Asm->OutStreamer.AddComment(Twine(" jumps to ") +
S.PadLabel->getName());
Asm->EmitLabelDifference(S.PadLabel, EHFuncBeginSym, 4);
}
// Offset of the first associated action record, relative to the start of
// the action table. This value is biased by 1 (1 indicates the start of
// the action table), and 0 indicates that there are no actions.
if (VerboseAsm) {
if (S.Action == 0)
Asm->OutStreamer.AddComment(" On action: cleanup");
else
Asm->OutStreamer.AddComment(" On action: " +
Twine((S.Action - 1) / 2 + 1));
}
Asm->EmitULEB128(S.Action);
}
}
// Emit the Action Table.
int Entry = 0;
for (SmallVectorImpl<ActionEntry>::const_iterator
I = Actions.begin(), E = Actions.end(); I != E; ++I) {
const ActionEntry &Action = *I;
if (VerboseAsm) {
// Emit comments that decode the action table.
Asm->OutStreamer.AddComment(">> Action Record " + Twine(++Entry) + " <<");
}
// Type Filter
//
// Used by the runtime to match the type of the thrown exception to the
// type of the catch clauses or the types in the exception specification.
if (VerboseAsm) {
if (Action.ValueForTypeID > 0)
Asm->OutStreamer.AddComment(" Catch TypeInfo " +
Twine(Action.ValueForTypeID));
else if (Action.ValueForTypeID < 0)
Asm->OutStreamer.AddComment(" Filter TypeInfo " +
Twine(Action.ValueForTypeID));
else
Asm->OutStreamer.AddComment(" Cleanup");
}
Asm->EmitSLEB128(Action.ValueForTypeID);
// Action Record
//
// Self-relative signed displacement in bytes of the next action record,
// or 0 if there is no next action record.
if (VerboseAsm) {
if (Action.NextAction == 0) {
Asm->OutStreamer.AddComment(" No further actions");
} else {
unsigned NextAction = Entry + (Action.NextAction + 1) / 2;
Asm->OutStreamer.AddComment(" Continue to action "+Twine(NextAction));
}
}
Asm->EmitSLEB128(Action.NextAction);
}
emitTypeInfos(TTypeEncoding);
Asm->EmitAlignment(2);
}
static void emitImplicitValueConstructor(SILGenFunction &gen,
ConstructorDecl *ctor) {
RegularLocation Loc(ctor);
Loc.markAutoGenerated();
// FIXME: Handle 'self' along with the other arguments.
auto *paramList = ctor->getParameterList(1);
auto selfTyCan = ctor->getImplicitSelfDecl()->getType()->getInOutObjectType();
SILType selfTy = gen.getLoweredType(selfTyCan);
// Emit the indirect return argument, if any.
SILValue resultSlot;
if (selfTy.isAddressOnly(gen.SGM.M)) {
auto &AC = gen.getASTContext();
auto VD = new (AC) ParamDecl(/*IsLet*/ false, SourceLoc(),
AC.getIdentifier("$return_value"),
SourceLoc(),
AC.getIdentifier("$return_value"), selfTyCan,
ctor);
resultSlot = new (gen.F.getModule()) SILArgument(gen.F.begin(), selfTy, VD);
}
// Emit the elementwise arguments.
SmallVector<RValue, 4> elements;
for (size_t i = 0, size = paramList->size(); i < size; ++i) {
auto ¶m = paramList->get(i);
elements.push_back(
emitImplicitValueConstructorArg(gen, Loc,
param.decl->getType()->getCanonicalType(),
ctor));
}
emitConstructorMetatypeArg(gen, ctor);
auto *decl = selfTy.getStructOrBoundGenericStruct();
assert(decl && "not a struct?!");
// If we have an indirect return slot, initialize it in-place.
if (resultSlot) {
auto elti = elements.begin(), eltEnd = elements.end();
for (VarDecl *field : decl->getStoredProperties()) {
auto fieldTy = selfTy.getFieldType(field, gen.SGM.M);
auto &fieldTL = gen.getTypeLowering(fieldTy);
SILValue slot = gen.B.createStructElementAddr(Loc, resultSlot, field,
fieldTL.getLoweredType().getAddressType());
InitializationPtr init(new KnownAddressInitialization(slot));
// An initialized 'let' property has a single value specified by the
// initializer - it doesn't come from an argument.
if (!field->isStatic() && field->isLet() &&
field->getParentInitializer()) {
assert(field->getType()->isEqual(field->getParentInitializer()
->getType()) && "Checked by sema");
// Cleanup after this initialization.
FullExpr scope(gen.Cleanups, field->getParentPatternBinding());
gen.emitRValue(field->getParentInitializer())
.forwardInto(gen, init.get(), Loc);
continue;
}
assert(elti != eltEnd && "number of args does not match number of fields");
(void)eltEnd;
std::move(*elti).forwardInto(gen, init.get(), Loc);
++elti;
}
gen.B.createReturn(ImplicitReturnLocation::getImplicitReturnLoc(Loc),
gen.emitEmptyTuple(Loc));
return;
}
// Otherwise, build a struct value directly from the elements.
SmallVector<SILValue, 4> eltValues;
auto elti = elements.begin(), eltEnd = elements.end();
for (VarDecl *field : decl->getStoredProperties()) {
auto fieldTy = selfTy.getFieldType(field, gen.SGM.M);
SILValue v;
// An initialized 'let' property has a single value specified by the
// initializer - it doesn't come from an argument.
if (!field->isStatic() && field->isLet() && field->getParentInitializer()) {
// Cleanup after this initialization.
FullExpr scope(gen.Cleanups, field->getParentPatternBinding());
v = gen.emitRValue(field->getParentInitializer())
.forwardAsSingleStorageValue(gen, fieldTy, Loc);
} else {
assert(elti != eltEnd && "number of args does not match number of fields");
(void)eltEnd;
v = std::move(*elti).forwardAsSingleStorageValue(gen, fieldTy, Loc);
++elti;
}
eltValues.push_back(v);
}
SILValue selfValue = gen.B.createStruct(Loc, selfTy, eltValues);
gen.B.createReturn(ImplicitReturnLocation::getImplicitReturnLoc(Loc),
selfValue);
return;
}
bool RecurrenceDescriptor::AddReductionVar(PHINode *Phi, RecurrenceKind Kind,
Loop *TheLoop, bool HasFunNoNaNAttr,
RecurrenceDescriptor &RedDes) {
if (Phi->getNumIncomingValues() != 2)
return false;
// Reduction variables are only found in the loop header block.
if (Phi->getParent() != TheLoop->getHeader())
return false;
// Obtain the reduction start value from the value that comes from the loop
// preheader.
Value *RdxStart = Phi->getIncomingValueForBlock(TheLoop->getLoopPreheader());
// ExitInstruction is the single value which is used outside the loop.
// We only allow for a single reduction value to be used outside the loop.
// This includes users of the reduction, variables (which form a cycle
// which ends in the phi node).
Instruction *ExitInstruction = nullptr;
// Indicates that we found a reduction operation in our scan.
bool FoundReduxOp = false;
// We start with the PHI node and scan for all of the users of this
// instruction. All users must be instructions that can be used as reduction
// variables (such as ADD). We must have a single out-of-block user. The cycle
// must include the original PHI.
bool FoundStartPHI = false;
// To recognize min/max patterns formed by a icmp select sequence, we store
// the number of instruction we saw from the recognized min/max pattern,
// to make sure we only see exactly the two instructions.
unsigned NumCmpSelectPatternInst = 0;
InstDesc ReduxDesc(false, nullptr);
// Data used for determining if the recurrence has been type-promoted.
Type *RecurrenceType = Phi->getType();
SmallPtrSet<Instruction *, 4> CastInsts;
Instruction *Start = Phi;
bool IsSigned = false;
SmallPtrSet<Instruction *, 8> VisitedInsts;
SmallVector<Instruction *, 8> Worklist;
// Return early if the recurrence kind does not match the type of Phi. If the
// recurrence kind is arithmetic, we attempt to look through AND operations
// resulting from the type promotion performed by InstCombine. Vector
// operations are not limited to the legal integer widths, so we may be able
// to evaluate the reduction in the narrower width.
if (RecurrenceType->isFloatingPointTy()) {
if (!isFloatingPointRecurrenceKind(Kind))
return false;
} else {
if (!isIntegerRecurrenceKind(Kind))
return false;
if (isArithmeticRecurrenceKind(Kind))
Start = lookThroughAnd(Phi, RecurrenceType, VisitedInsts, CastInsts);
}
Worklist.push_back(Start);
VisitedInsts.insert(Start);
// A value in the reduction can be used:
// - By the reduction:
// - Reduction operation:
// - One use of reduction value (safe).
// - Multiple use of reduction value (not safe).
// - PHI:
// - All uses of the PHI must be the reduction (safe).
// - Otherwise, not safe.
// - By one instruction outside of the loop (safe).
// - By further instructions outside of the loop (not safe).
// - By an instruction that is not part of the reduction (not safe).
// This is either:
// * An instruction type other than PHI or the reduction operation.
// * A PHI in the header other than the initial PHI.
while (!Worklist.empty()) {
Instruction *Cur = Worklist.back();
Worklist.pop_back();
// No Users.
// If the instruction has no users then this is a broken chain and can't be
// a reduction variable.
if (Cur->use_empty())
return false;
bool IsAPhi = isa<PHINode>(Cur);
// A header PHI use other than the original PHI.
if (Cur != Phi && IsAPhi && Cur->getParent() == Phi->getParent())
return false;
// Reductions of instructions such as Div, and Sub is only possible if the
// LHS is the reduction variable.
if (!Cur->isCommutative() && !IsAPhi && !isa<SelectInst>(Cur) &&
!isa<ICmpInst>(Cur) && !isa<FCmpInst>(Cur) &&
!VisitedInsts.count(dyn_cast<Instruction>(Cur->getOperand(0))))
return false;
// Any reduction instruction must be of one of the allowed kinds. We ignore
// the starting value (the Phi or an AND instruction if the Phi has been
// type-promoted).
if (Cur != Start) {
ReduxDesc = isRecurrenceInstr(Cur, Kind, ReduxDesc, HasFunNoNaNAttr);
if (!ReduxDesc.isRecurrence())
return false;
}
// A reduction operation must only have one use of the reduction value.
if (!IsAPhi && Kind != RK_IntegerMinMax && Kind != RK_FloatMinMax &&
hasMultipleUsesOf(Cur, VisitedInsts))
return false;
// All inputs to a PHI node must be a reduction value.
if (IsAPhi && Cur != Phi && !areAllUsesIn(Cur, VisitedInsts))
return false;
if (Kind == RK_IntegerMinMax &&
(isa<ICmpInst>(Cur) || isa<SelectInst>(Cur)))
++NumCmpSelectPatternInst;
if (Kind == RK_FloatMinMax && (isa<FCmpInst>(Cur) || isa<SelectInst>(Cur)))
++NumCmpSelectPatternInst;
// Check whether we found a reduction operator.
FoundReduxOp |= !IsAPhi && Cur != Start;
// Process users of current instruction. Push non-PHI nodes after PHI nodes
// onto the stack. This way we are going to have seen all inputs to PHI
// nodes once we get to them.
SmallVector<Instruction *, 8> NonPHIs;
SmallVector<Instruction *, 8> PHIs;
for (User *U : Cur->users()) {
Instruction *UI = cast<Instruction>(U);
// Check if we found the exit user.
BasicBlock *Parent = UI->getParent();
if (!TheLoop->contains(Parent)) {
// Exit if you find multiple outside users or if the header phi node is
// being used. In this case the user uses the value of the previous
// iteration, in which case we would loose "VF-1" iterations of the
// reduction operation if we vectorize.
if (ExitInstruction != nullptr || Cur == Phi)
return false;
// The instruction used by an outside user must be the last instruction
// before we feed back to the reduction phi. Otherwise, we loose VF-1
// operations on the value.
if (!is_contained(Phi->operands(), Cur))
return false;
ExitInstruction = Cur;
continue;
}
// Process instructions only once (termination). Each reduction cycle
// value must only be used once, except by phi nodes and min/max
// reductions which are represented as a cmp followed by a select.
InstDesc IgnoredVal(false, nullptr);
if (VisitedInsts.insert(UI).second) {
if (isa<PHINode>(UI))
PHIs.push_back(UI);
else
NonPHIs.push_back(UI);
} else if (!isa<PHINode>(UI) &&
((!isa<FCmpInst>(UI) && !isa<ICmpInst>(UI) &&
!isa<SelectInst>(UI)) ||
!isMinMaxSelectCmpPattern(UI, IgnoredVal).isRecurrence()))
return false;
// Remember that we completed the cycle.
if (UI == Phi)
FoundStartPHI = true;
}
Worklist.append(PHIs.begin(), PHIs.end());
Worklist.append(NonPHIs.begin(), NonPHIs.end());
}
// This means we have seen one but not the other instruction of the
// pattern or more than just a select and cmp.
if ((Kind == RK_IntegerMinMax || Kind == RK_FloatMinMax) &&
NumCmpSelectPatternInst != 2)
return false;
if (!FoundStartPHI || !FoundReduxOp || !ExitInstruction)
return false;
// If we think Phi may have been type-promoted, we also need to ensure that
// all source operands of the reduction are either SExtInsts or ZEstInsts. If
// so, we will be able to evaluate the reduction in the narrower bit width.
if (Start != Phi)
if (!getSourceExtensionKind(Start, ExitInstruction, RecurrenceType,
IsSigned, VisitedInsts, CastInsts))
return false;
// We found a reduction var if we have reached the original phi node and we
// only have a single instruction with out-of-loop users.
// The ExitInstruction(Instruction which is allowed to have out-of-loop users)
// is saved as part of the RecurrenceDescriptor.
// Save the description of this reduction variable.
RecurrenceDescriptor RD(
RdxStart, ExitInstruction, Kind, ReduxDesc.getMinMaxKind(),
ReduxDesc.getUnsafeAlgebraInst(), RecurrenceType, IsSigned, CastInsts);
RedDes = RD;
return true;
}
int main(int argc_, const char **argv_) {
llvm::sys::PrintStackTraceOnErrorSignal();
llvm::PrettyStackTraceProgram X(argc_, argv_);
std::set<std::string> SavedStrings;
SmallVector<const char*, 256> argv;
ExpandArgv(argc_, argv_, argv, SavedStrings);
// Handle -cc1 integrated tools.
if (argv.size() > 1 && StringRef(argv[1]).startswith("-cc1")) {
StringRef Tool = argv[1] + 4;
if (Tool == "")
return cc1_main(argv.data()+2, argv.data()+argv.size(), argv[0],
(void*) (intptr_t) GetExecutablePath);
if (Tool == "as")
return cc1as_main(argv.data()+2, argv.data()+argv.size(), argv[0],
(void*) (intptr_t) GetExecutablePath);
// Reject unknown tools.
llvm::errs() << "error: unknown integrated tool '" << Tool << "'\n";
return 1;
}
bool CanonicalPrefixes = true;
for (int i = 1, size = argv.size(); i < size; ++i) {
if (StringRef(argv[i]) == "-no-canonical-prefixes") {
CanonicalPrefixes = false;
break;
}
}
llvm::sys::Path Path = GetExecutablePath(argv[0], CanonicalPrefixes);
IntrusiveRefCntPtr<DiagnosticOptions> DiagOpts = new DiagnosticOptions;
{
// Note that ParseDiagnosticArgs() uses the cc1 option table.
OwningPtr<OptTable> CC1Opts(createDriverOptTable());
unsigned MissingArgIndex, MissingArgCount;
OwningPtr<InputArgList> Args(CC1Opts->ParseArgs(argv.begin()+1, argv.end(),
MissingArgIndex, MissingArgCount));
// We ignore MissingArgCount and the return value of ParseDiagnosticArgs.
// Any errors that would be diagnosed here will also be diagnosed later,
// when the DiagnosticsEngine actually exists.
(void) ParseDiagnosticArgs(*DiagOpts, *Args);
}
// Now we can create the DiagnosticsEngine with a properly-filled-out
// DiagnosticOptions instance.
TextDiagnosticPrinter *DiagClient
= new TextDiagnosticPrinter(llvm::errs(), &*DiagOpts);
DiagClient->setPrefix(llvm::sys::path::stem(Path.str()));
IntrusiveRefCntPtr<DiagnosticIDs> DiagID(new DiagnosticIDs());
DiagnosticsEngine Diags(DiagID, &*DiagOpts, DiagClient);
ProcessWarningOptions(Diags, *DiagOpts);
#ifdef CLANG_IS_PRODUCTION
const bool IsProduction = true;
#else
const bool IsProduction = false;
#endif
Driver TheDriver(Path.str(), llvm::sys::getDefaultTargetTriple(),
"a.out", IsProduction, Diags);
// Attempt to find the original path used to invoke the driver, to determine
// the installed path. We do this manually, because we want to support that
// path being a symlink.
{
SmallString<128> InstalledPath(argv[0]);
// Do a PATH lookup, if there are no directory components.
if (llvm::sys::path::filename(InstalledPath) == InstalledPath) {
llvm::sys::Path Tmp = llvm::sys::Program::FindProgramByName(
llvm::sys::path::filename(InstalledPath.str()));
if (!Tmp.empty())
InstalledPath = Tmp.str();
}
llvm::sys::fs::make_absolute(InstalledPath);
InstalledPath = llvm::sys::path::parent_path(InstalledPath);
bool exists;
if (!llvm::sys::fs::exists(InstalledPath.str(), exists) && exists)
TheDriver.setInstalledDir(InstalledPath);
}
llvm::InitializeAllTargets();
ParseProgName(argv, SavedStrings, TheDriver);
// Handle CC_PRINT_OPTIONS and CC_PRINT_OPTIONS_FILE.
TheDriver.CCPrintOptions = !!::getenv("CC_PRINT_OPTIONS");
if (TheDriver.CCPrintOptions)
TheDriver.CCPrintOptionsFilename = ::getenv("CC_PRINT_OPTIONS_FILE");
// Handle CC_PRINT_HEADERS and CC_PRINT_HEADERS_FILE.
TheDriver.CCPrintHeaders = !!::getenv("CC_PRINT_HEADERS");
if (TheDriver.CCPrintHeaders)
TheDriver.CCPrintHeadersFilename = ::getenv("CC_PRINT_HEADERS_FILE");
// Handle CC_LOG_DIAGNOSTICS and CC_LOG_DIAGNOSTICS_FILE.
TheDriver.CCLogDiagnostics = !!::getenv("CC_LOG_DIAGNOSTICS");
if (TheDriver.CCLogDiagnostics)
TheDriver.CCLogDiagnosticsFilename = ::getenv("CC_LOG_DIAGNOSTICS_FILE");
// Handle QA_OVERRIDE_GCC3_OPTIONS and CCC_ADD_ARGS, used for editing a
// command line behind the scenes.
if (const char *OverrideStr = ::getenv("QA_OVERRIDE_GCC3_OPTIONS")) {
// FIXME: Driver shouldn't take extra initial argument.
ApplyQAOverride(argv, OverrideStr, SavedStrings);
} else if (const char *Cur = ::getenv("CCC_ADD_ARGS")) {
// FIXME: Driver shouldn't take extra initial argument.
std::vector<const char*> ExtraArgs;
for (;;) {
const char *Next = strchr(Cur, ',');
if (Next) {
ExtraArgs.push_back(SaveStringInSet(SavedStrings,
std::string(Cur, Next)));
Cur = Next + 1;
} else {
if (*Cur != '\0')
ExtraArgs.push_back(SaveStringInSet(SavedStrings, Cur));
break;
}
}
argv.insert(&argv[1], ExtraArgs.begin(), ExtraArgs.end());
}
OwningPtr<Compilation> C(TheDriver.BuildCompilation(argv));
int Res = 0;
const Command *FailingCommand = 0;
if (C.get())
Res = TheDriver.ExecuteCompilation(*C, FailingCommand);
// Force a crash to test the diagnostics.
if(::getenv("FORCE_CLANG_DIAGNOSTICS_CRASH"))
Res = -1;
// If result status is < 0, then the driver command signalled an error.
// In this case, generate additional diagnostic information if possible.
if (Res < 0)
TheDriver.generateCompilationDiagnostics(*C, FailingCommand);
// If any timers were active but haven't been destroyed yet, print their
// results now. This happens in -disable-free mode.
llvm::TimerGroup::printAll(llvm::errs());
llvm::llvm_shutdown();
#ifdef _WIN32
// Exit status should not be negative on Win32, unless abnormal termination.
// Once abnormal termiation was caught, negative status should not be
// propagated.
if (Res < 0)
Res = 1;
#endif
return Res;
}
/// optimizeExtInstr - If instruction is a copy-like instruction, i.e. it reads
/// a single register and writes a single register and it does not modify the
/// source, and if the source value is preserved as a sub-register of the
/// result, then replace all reachable uses of the source with the subreg of the
/// result.
///
/// Do not generate an EXTRACT that is used only in a debug use, as this changes
/// the code. Since this code does not currently share EXTRACTs, just ignore all
/// debug uses.
bool PeepholeOptimizer::
optimizeExtInstr(MachineInstr *MI, MachineBasicBlock *MBB,
SmallPtrSet<MachineInstr*, 8> &LocalMIs) {
unsigned SrcReg, DstReg, SubIdx;
if (!TII->isCoalescableExtInstr(*MI, SrcReg, DstReg, SubIdx))
return false;
if (TargetRegisterInfo::isPhysicalRegister(DstReg) ||
TargetRegisterInfo::isPhysicalRegister(SrcReg))
return false;
MachineRegisterInfo::use_nodbg_iterator UI = MRI->use_nodbg_begin(SrcReg);
if (++UI == MRI->use_nodbg_end())
// No other uses.
return false;
// The source has other uses. See if we can replace the other uses with use of
// the result of the extension.
SmallPtrSet<MachineBasicBlock*, 4> ReachedBBs;
UI = MRI->use_nodbg_begin(DstReg);
for (MachineRegisterInfo::use_nodbg_iterator UE = MRI->use_nodbg_end();
UI != UE; ++UI)
ReachedBBs.insert(UI->getParent());
// Uses that are in the same BB of uses of the result of the instruction.
SmallVector<MachineOperand*, 8> Uses;
// Uses that the result of the instruction can reach.
SmallVector<MachineOperand*, 8> ExtendedUses;
bool ExtendLife = true;
UI = MRI->use_nodbg_begin(SrcReg);
for (MachineRegisterInfo::use_nodbg_iterator UE = MRI->use_nodbg_end();
UI != UE; ++UI) {
MachineOperand &UseMO = UI.getOperand();
MachineInstr *UseMI = &*UI;
if (UseMI == MI)
continue;
if (UseMI->isPHI()) {
ExtendLife = false;
continue;
}
// It's an error to translate this:
//
// %reg1025 = <sext> %reg1024
// ...
// %reg1026 = SUBREG_TO_REG 0, %reg1024, 4
//
// into this:
//
// %reg1025 = <sext> %reg1024
// ...
// %reg1027 = COPY %reg1025:4
// %reg1026 = SUBREG_TO_REG 0, %reg1027, 4
//
// The problem here is that SUBREG_TO_REG is there to assert that an
// implicit zext occurs. It doesn't insert a zext instruction. If we allow
// the COPY here, it will give us the value after the <sext>, not the
// original value of %reg1024 before <sext>.
if (UseMI->getOpcode() == TargetOpcode::SUBREG_TO_REG)
continue;
MachineBasicBlock *UseMBB = UseMI->getParent();
if (UseMBB == MBB) {
// Local uses that come after the extension.
if (!LocalMIs.count(UseMI))
Uses.push_back(&UseMO);
} else if (ReachedBBs.count(UseMBB)) {
// Non-local uses where the result of the extension is used. Always
// replace these unless it's a PHI.
Uses.push_back(&UseMO);
} else if (Aggressive && DT->dominates(MBB, UseMBB)) {
// We may want to extend the live range of the extension result in order
// to replace these uses.
ExtendedUses.push_back(&UseMO);
} else {
// Both will be live out of the def MBB anyway. Don't extend live range of
// the extension result.
ExtendLife = false;
break;
}
}
if (ExtendLife && !ExtendedUses.empty())
// Extend the liveness of the extension result.
std::copy(ExtendedUses.begin(), ExtendedUses.end(),
std::back_inserter(Uses));
// Now replace all uses.
bool Changed = false;
if (!Uses.empty()) {
SmallPtrSet<MachineBasicBlock*, 4> PHIBBs;
// Look for PHI uses of the extended result, we don't want to extend the
// liveness of a PHI input. It breaks all kinds of assumptions down
// stream. A PHI use is expected to be the kill of its source values.
UI = MRI->use_nodbg_begin(DstReg);
for (MachineRegisterInfo::use_nodbg_iterator
UE = MRI->use_nodbg_end(); UI != UE; ++UI)
if (UI->isPHI())
PHIBBs.insert(UI->getParent());
const TargetRegisterClass *RC = MRI->getRegClass(SrcReg);
for (unsigned i = 0, e = Uses.size(); i != e; ++i) {
MachineOperand *UseMO = Uses[i];
MachineInstr *UseMI = UseMO->getParent();
MachineBasicBlock *UseMBB = UseMI->getParent();
if (PHIBBs.count(UseMBB))
continue;
// About to add uses of DstReg, clear DstReg's kill flags.
if (!Changed)
MRI->clearKillFlags(DstReg);
unsigned NewVR = MRI->createVirtualRegister(RC);
BuildMI(*UseMBB, UseMI, UseMI->getDebugLoc(),
TII->get(TargetOpcode::COPY), NewVR)
.addReg(DstReg, 0, SubIdx);
UseMO->setReg(NewVR);
++NumReuse;
Changed = true;
}
}
return Changed;
}
/// DetermineInsertionPoint - 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::priority_queue<DomTreeNodePair, SmallVector<DomTreeNodePair, 32>,
DomTreeNodeCompare> 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);
}
/// Unroll the given loop by Count. The loop must be in LCSSA form. Returns true
/// if unrolling was successful, or false if the loop was unmodified. Unrolling
/// can only fail when the loop's latch block is not terminated by a conditional
/// branch instruction. However, if the trip count (and multiple) are not known,
/// loop unrolling will mostly produce more code that is no faster.
///
/// TripCount is generally defined as the number of times the loop header
/// executes. UnrollLoop relaxes the definition to permit early exits: here
/// TripCount is the iteration on which control exits LatchBlock if no early
/// exits were taken. Note that UnrollLoop assumes that the loop counter test
/// terminates LatchBlock in order to remove unnecesssary instances of the
/// test. In other words, control may exit the loop prior to TripCount
/// iterations via an early branch, but control may not exit the loop from the
/// LatchBlock's terminator prior to TripCount iterations.
///
/// Similarly, TripMultiple divides the number of times that the LatchBlock may
/// execute without exiting the loop.
///
/// If AllowRuntime is true then UnrollLoop will consider unrolling loops that
/// have a runtime (i.e. not compile time constant) trip count. Unrolling these
/// loops require a unroll "prologue" that runs "RuntimeTripCount % Count"
/// iterations before branching into the unrolled loop. UnrollLoop will not
/// runtime-unroll the loop if computing RuntimeTripCount will be expensive and
/// AllowExpensiveTripCount is false.
///
/// The LoopInfo Analysis that is passed will be kept consistent.
///
/// This utility preserves LoopInfo. It will also preserve ScalarEvolution and
/// DominatorTree if they are non-null.
bool llvm::UnrollLoop(Loop *L, unsigned Count, unsigned TripCount, bool Force,
bool AllowRuntime, bool AllowExpensiveTripCount,
unsigned TripMultiple, LoopInfo *LI, ScalarEvolution *SE,
DominatorTree *DT, AssumptionCache *AC,
bool PreserveLCSSA) {
BasicBlock *Preheader = L->getLoopPreheader();
if (!Preheader) {
DEBUG(dbgs() << " Can't unroll; loop preheader-insertion failed.\n");
return false;
}
BasicBlock *LatchBlock = L->getLoopLatch();
if (!LatchBlock) {
DEBUG(dbgs() << " Can't unroll; loop exit-block-insertion failed.\n");
return false;
}
// Loops with indirectbr cannot be cloned.
if (!L->isSafeToClone()) {
DEBUG(dbgs() << " Can't unroll; Loop body cannot be cloned.\n");
return false;
}
BasicBlock *Header = L->getHeader();
BranchInst *BI = dyn_cast<BranchInst>(LatchBlock->getTerminator());
if (!BI || BI->isUnconditional()) {
// The loop-rotate pass can be helpful to avoid this in many cases.
DEBUG(dbgs() <<
" Can't unroll; loop not terminated by a conditional branch.\n");
return false;
}
if (Header->hasAddressTaken()) {
// The loop-rotate pass can be helpful to avoid this in many cases.
DEBUG(dbgs() <<
" Won't unroll loop: address of header block is taken.\n");
return false;
}
if (TripCount != 0)
DEBUG(dbgs() << " Trip Count = " << TripCount << "\n");
if (TripMultiple != 1)
DEBUG(dbgs() << " Trip Multiple = " << TripMultiple << "\n");
// Effectively "DCE" unrolled iterations that are beyond the tripcount
// and will never be executed.
if (TripCount != 0 && Count > TripCount)
Count = TripCount;
// Don't enter the unroll code if there is nothing to do. This way we don't
// need to support "partial unrolling by 1".
if (TripCount == 0 && Count < 2)
return false;
assert(Count > 0);
assert(TripMultiple > 0);
assert(TripCount == 0 || TripCount % TripMultiple == 0);
// Are we eliminating the loop control altogether?
bool CompletelyUnroll = Count == TripCount;
SmallVector<BasicBlock *, 4> ExitBlocks;
L->getExitBlocks(ExitBlocks);
std::vector<BasicBlock*> OriginalLoopBlocks = L->getBlocks();
// Go through all exits of L and see if there are any phi-nodes there. We just
// conservatively assume that they're inserted to preserve LCSSA form, which
// means that complete unrolling might break this form. We need to either fix
// it in-place after the transformation, or entirely rebuild LCSSA. TODO: For
// now we just recompute LCSSA for the outer loop, but it should be possible
// to fix it in-place.
bool NeedToFixLCSSA = PreserveLCSSA && CompletelyUnroll &&
std::any_of(ExitBlocks.begin(), ExitBlocks.end(),
[&](BasicBlock *BB) { return isa<PHINode>(BB->begin()); });
// We assume a run-time trip count if the compiler cannot
// figure out the loop trip count and the unroll-runtime
// flag is specified.
bool RuntimeTripCount = (TripCount == 0 && Count > 0 && AllowRuntime);
// Loops containing convergent instructions must have a count that divides
// their TripMultiple.
DEBUG(
{
bool HasConvergent = false;
for (auto &BB : L->blocks())
for (auto &I : *BB)
if (auto CS = CallSite(&I))
HasConvergent |= CS.isConvergent();
assert((!HasConvergent || TripMultiple % Count == 0) &&
"Unroll count must divide trip multiple if loop contains a "
"convergent operation.");
});
void SelectionDAGBuilder::LowerStatepoint(
ImmutableStatepoint ISP, MachineBasicBlock *LandingPad /*=nullptr*/) {
// The basic scheme here is that information about both the original call and
// the safepoint is encoded in the CallInst. We create a temporary call and
// lower it, then reverse engineer the calling sequence.
NumOfStatepoints++;
// Clear state
StatepointLowering.startNewStatepoint(*this);
ImmutableCallSite CS(ISP.getCallSite());
#ifndef NDEBUG
// Consistency check. Don't do this for invokes. It would be too
// expensive to preserve this information across different basic blocks
if (!CS.isInvoke()) {
for (const User *U : CS->users()) {
const CallInst *Call = cast<CallInst>(U);
if (isGCRelocate(Call))
StatepointLowering.scheduleRelocCall(*Call);
}
}
#endif
#ifndef NDEBUG
// If this is a malformed statepoint, report it early to simplify debugging.
// This should catch any IR level mistake that's made when constructing or
// transforming statepoints.
ISP.verify();
// Check that the associated GCStrategy expects to encounter statepoints.
assert(GFI->getStrategy().useStatepoints() &&
"GCStrategy does not expect to encounter statepoints");
#endif
// Lower statepoint vmstate and gcstate arguments
SmallVector<SDValue, 10> LoweredMetaArgs;
lowerStatepointMetaArgs(LoweredMetaArgs, ISP, *this);
// Get call node, we will replace it later with statepoint
SDNode *CallNode =
lowerCallFromStatepoint(ISP, LandingPad, *this, PendingExports);
// Construct the actual GC_TRANSITION_START, STATEPOINT, and GC_TRANSITION_END
// nodes with all the appropriate arguments and return values.
// Call Node: Chain, Target, {Args}, RegMask, [Glue]
SDValue Chain = CallNode->getOperand(0);
SDValue Glue;
bool CallHasIncomingGlue = CallNode->getGluedNode();
if (CallHasIncomingGlue) {
// Glue is always last operand
Glue = CallNode->getOperand(CallNode->getNumOperands() - 1);
}
// Build the GC_TRANSITION_START node if necessary.
//
// The operands to the GC_TRANSITION_{START,END} nodes are laid out in the
// order in which they appear in the call to the statepoint intrinsic. If
// any of the operands is a pointer-typed, that operand is immediately
// followed by a SRCVALUE for the pointer that may be used during lowering
// (e.g. to form MachinePointerInfo values for loads/stores).
const bool IsGCTransition =
(ISP.getFlags() & (uint64_t)StatepointFlags::GCTransition) ==
(uint64_t)StatepointFlags::GCTransition;
if (IsGCTransition) {
SmallVector<SDValue, 8> TSOps;
// Add chain
TSOps.push_back(Chain);
// Add GC transition arguments
for (const Value *V : ISP.gc_transition_args()) {
TSOps.push_back(getValue(V));
if (V->getType()->isPointerTy())
TSOps.push_back(DAG.getSrcValue(V));
}
// Add glue if necessary
if (CallHasIncomingGlue)
TSOps.push_back(Glue);
SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
SDValue GCTransitionStart =
DAG.getNode(ISD::GC_TRANSITION_START, getCurSDLoc(), NodeTys, TSOps);
Chain = GCTransitionStart.getValue(0);
Glue = GCTransitionStart.getValue(1);
}
// TODO: Currently, all of these operands are being marked as read/write in
// PrologEpilougeInserter.cpp, we should special case the VMState arguments
// and flags to be read-only.
SmallVector<SDValue, 40> Ops;
// Add the <id> and <numBytes> constants.
Ops.push_back(DAG.getTargetConstant(ISP.getID(), getCurSDLoc(), MVT::i64));
Ops.push_back(
DAG.getTargetConstant(ISP.getNumPatchBytes(), getCurSDLoc(), MVT::i32));
// Calculate and push starting position of vmstate arguments
// Get number of arguments incoming directly into call node
unsigned NumCallRegArgs =
CallNode->getNumOperands() - (CallHasIncomingGlue ? 4 : 3);
Ops.push_back(DAG.getTargetConstant(NumCallRegArgs, getCurSDLoc(), MVT::i32));
// Add call target
SDValue CallTarget = SDValue(CallNode->getOperand(1).getNode(), 0);
Ops.push_back(CallTarget);
// Add call arguments
// Get position of register mask in the call
SDNode::op_iterator RegMaskIt;
if (CallHasIncomingGlue)
RegMaskIt = CallNode->op_end() - 2;
else
RegMaskIt = CallNode->op_end() - 1;
Ops.insert(Ops.end(), CallNode->op_begin() + 2, RegMaskIt);
// Add a constant argument for the calling convention
pushStackMapConstant(Ops, *this, CS.getCallingConv());
// Add a constant argument for the flags
uint64_t Flags = ISP.getFlags();
assert(
((Flags & ~(uint64_t)StatepointFlags::MaskAll) == 0)
&& "unknown flag used");
pushStackMapConstant(Ops, *this, Flags);
// Insert all vmstate and gcstate arguments
Ops.insert(Ops.end(), LoweredMetaArgs.begin(), LoweredMetaArgs.end());
// Add register mask from call node
Ops.push_back(*RegMaskIt);
// Add chain
Ops.push_back(Chain);
// Same for the glue, but we add it only if original call had it
if (Glue.getNode())
Ops.push_back(Glue);
// Compute return values. Provide a glue output since we consume one as
// input. This allows someone else to chain off us as needed.
SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
SDNode *StatepointMCNode =
DAG.getMachineNode(TargetOpcode::STATEPOINT, getCurSDLoc(), NodeTys, Ops);
SDNode *SinkNode = StatepointMCNode;
// Build the GC_TRANSITION_END node if necessary.
//
// See the comment above regarding GC_TRANSITION_START for the layout of
// the operands to the GC_TRANSITION_END node.
if (IsGCTransition) {
SmallVector<SDValue, 8> TEOps;
// Add chain
TEOps.push_back(SDValue(StatepointMCNode, 0));
// Add GC transition arguments
for (const Value *V : ISP.gc_transition_args()) {
TEOps.push_back(getValue(V));
if (V->getType()->isPointerTy())
TEOps.push_back(DAG.getSrcValue(V));
}
// Add glue
TEOps.push_back(SDValue(StatepointMCNode, 1));
SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
SDValue GCTransitionStart =
DAG.getNode(ISD::GC_TRANSITION_END, getCurSDLoc(), NodeTys, TEOps);
SinkNode = GCTransitionStart.getNode();
}
// Replace original call
DAG.ReplaceAllUsesWith(CallNode, SinkNode); // This may update Root
// Remove originall call node
DAG.DeleteNode(CallNode);
// DON'T set the root - under the assumption that it's already set past the
// inserted node we created.
// TODO: A better future implementation would be to emit a single variable
// argument, variable return value STATEPOINT node here and then hookup the
// return value of each gc.relocate to the respective output of the
// previously emitted STATEPOINT value. Unfortunately, this doesn't appear
// to actually be possible today.
}
// A soft instruction can be changed to work in other domains given by mask.
void ExeDepsFix::visitSoftInstr(MachineInstr *mi, unsigned mask) {
// Bitmask of available domains for this instruction after taking collapsed
// operands into account.
unsigned available = mask;
// Scan the explicit use operands for incoming domains.
SmallVector<int, 4> used;
if (LiveRegs)
for (unsigned i = mi->getDesc().getNumDefs(),
e = mi->getDesc().getNumOperands(); i != e; ++i) {
MachineOperand &mo = mi->getOperand(i);
if (!mo.isReg()) continue;
for (int rx : regIndices(mo.getReg())) {
DomainValue *dv = LiveRegs[rx].Value;
if (dv == nullptr)
continue;
// Bitmask of domains that dv and available have in common.
unsigned common = dv->getCommonDomains(available);
// Is it possible to use this collapsed register for free?
if (dv->isCollapsed()) {
// Restrict available domains to the ones in common with the operand.
// If there are no common domains, we must pay the cross-domain
// penalty for this operand.
if (common) available = common;
} else if (common)
// Open DomainValue is compatible, save it for merging.
used.push_back(rx);
else
// Open DomainValue is not compatible with instruction. It is useless
// now.
kill(rx);
}
}
// If the collapsed operands force a single domain, propagate the collapse.
if (isPowerOf2_32(available)) {
unsigned domain = countTrailingZeros(available);
TII->setExecutionDomain(*mi, domain);
visitHardInstr(mi, domain);
return;
}
// Kill off any remaining uses that don't match available, and build a list of
// incoming DomainValues that we want to merge.
SmallVector<LiveReg, 4> Regs;
for (SmallVectorImpl<int>::iterator i=used.begin(), e=used.end(); i!=e; ++i) {
int rx = *i;
assert(LiveRegs && "no space allocated for live registers");
const LiveReg &LR = LiveRegs[rx];
// This useless DomainValue could have been missed above.
if (!LR.Value->getCommonDomains(available)) {
kill(rx);
continue;
}
// Sorted insertion.
bool Inserted = false;
for (SmallVectorImpl<LiveReg>::iterator i = Regs.begin(), e = Regs.end();
i != e && !Inserted; ++i) {
if (LR.Def < i->Def) {
Inserted = true;
Regs.insert(i, LR);
}
}
if (!Inserted)
Regs.push_back(LR);
}
// doms are now sorted in order of appearance. Try to merge them all, giving
// priority to the latest ones.
DomainValue *dv = nullptr;
while (!Regs.empty()) {
if (!dv) {
dv = Regs.pop_back_val().Value;
// Force the first dv to match the current instruction.
dv->AvailableDomains = dv->getCommonDomains(available);
assert(dv->AvailableDomains && "Domain should have been filtered");
continue;
}
DomainValue *Latest = Regs.pop_back_val().Value;
// Skip already merged values.
if (Latest == dv || Latest->Next)
continue;
if (merge(dv, Latest))
continue;
// If latest didn't merge, it is useless now. Kill all registers using it.
for (int i : used) {
assert(LiveRegs && "no space allocated for live registers");
if (LiveRegs[i].Value == Latest)
kill(i);
}
}
// dv is the DomainValue we are going to use for this instruction.
if (!dv) {
dv = alloc();
dv->AvailableDomains = available;
}
dv->Instrs.push_back(mi);
// Finally set all defs and non-collapsed uses to dv. We must iterate through
// all the operators, including imp-def ones.
for (MachineInstr::mop_iterator ii = mi->operands_begin(),
ee = mi->operands_end();
ii != ee; ++ii) {
MachineOperand &mo = *ii;
if (!mo.isReg()) continue;
for (int rx : regIndices(mo.getReg())) {
if (!LiveRegs[rx].Value || (mo.isDef() && LiveRegs[rx].Value != dv)) {
kill(rx);
setLiveReg(rx, dv);
}
}
}
}
/// findLoopBackEdges - Do a DFS walk to find loop back edges.
///
void CodeGenPrepare::findLoopBackEdges(const Function &F) {
SmallVector<std::pair<const BasicBlock*,const BasicBlock*>, 32> Edges;
FindFunctionBackedges(F, Edges);
BackEdges.insert(Edges.begin(), Edges.end());
}
void PlistDiagnostics::FlushDiagnosticsImpl(
std::vector<const PathDiagnostic *> &Diags,
FilesMade *filesMade) {
// Build up a set of FIDs that we use by scanning the locations and
// ranges of the diagnostics.
FIDMap FM;
SmallVector<FileID, 10> Fids;
const SourceManager* SM = nullptr;
if (!Diags.empty())
SM = &(*(*Diags.begin())->path.begin())->getLocation().getManager();
for (std::vector<const PathDiagnostic*>::iterator DI = Diags.begin(),
DE = Diags.end(); DI != DE; ++DI) {
const PathDiagnostic *D = *DI;
SmallVector<const PathPieces *, 5> WorkList;
WorkList.push_back(&D->path);
while (!WorkList.empty()) {
const PathPieces &path = *WorkList.pop_back_val();
for (PathPieces::const_iterator I = path.begin(), E = path.end(); I != E;
++I) {
const PathDiagnosticPiece *piece = I->getPtr();
AddFID(FM, Fids, *SM, piece->getLocation().asLocation());
ArrayRef<SourceRange> Ranges = piece->getRanges();
for (ArrayRef<SourceRange>::iterator I = Ranges.begin(),
E = Ranges.end(); I != E; ++I) {
AddFID(FM, Fids, *SM, I->getBegin());
AddFID(FM, Fids, *SM, I->getEnd());
}
if (const PathDiagnosticCallPiece *call =
dyn_cast<PathDiagnosticCallPiece>(piece)) {
IntrusiveRefCntPtr<PathDiagnosticEventPiece>
callEnterWithin = call->getCallEnterWithinCallerEvent();
if (callEnterWithin)
AddFID(FM, Fids, *SM, callEnterWithin->getLocation().asLocation());
WorkList.push_back(&call->path);
}
else if (const PathDiagnosticMacroPiece *macro =
dyn_cast<PathDiagnosticMacroPiece>(piece)) {
WorkList.push_back(¯o->subPieces);
}
}
}
}
// Open the file.
std::string ErrMsg;
llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg, llvm::sys::fs::F_Text);
if (!ErrMsg.empty()) {
llvm::errs() << "warning: could not create file: " << OutputFile << '\n';
return;
}
// Write the plist header.
o << PlistHeader;
// Write the root object: a <dict> containing...
// - "clang_version", the string representation of clang version
// - "files", an <array> mapping from FIDs to file names
// - "diagnostics", an <array> containing the path diagnostics
o << "<dict>\n" <<
" <key>clang_version</key>\n";
EmitString(o, getClangFullVersion()) << '\n';
o << " <key>files</key>\n"
" <array>\n";
for (SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
I!=E; ++I) {
o << " ";
EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
}
o << " </array>\n"
" <key>diagnostics</key>\n"
" <array>\n";
for (std::vector<const PathDiagnostic*>::iterator DI=Diags.begin(),
DE = Diags.end(); DI!=DE; ++DI) {
o << " <dict>\n"
" <key>path</key>\n";
const PathDiagnostic *D = *DI;
o << " <array>\n";
for (PathPieces::const_iterator I = D->path.begin(), E = D->path.end();
I != E; ++I)
ReportDiag(o, **I, FM, *SM, LangOpts);
o << " </array>\n";
// Output the bug type and bug category.
o << " <key>description</key>";
EmitString(o, D->getShortDescription()) << '\n';
o << " <key>category</key>";
EmitString(o, D->getCategory()) << '\n';
o << " <key>type</key>";
EmitString(o, D->getBugType()) << '\n';
// Output information about the semantic context where
// the issue occurred.
if (const Decl *DeclWithIssue = D->getDeclWithIssue()) {
// FIXME: handle blocks, which have no name.
if (const NamedDecl *ND = dyn_cast<NamedDecl>(DeclWithIssue)) {
StringRef declKind;
switch (ND->getKind()) {
case Decl::CXXRecord:
declKind = "C++ class";
break;
case Decl::CXXMethod:
declKind = "C++ method";
break;
case Decl::ObjCMethod:
declKind = "Objective-C method";
break;
case Decl::Function:
declKind = "function";
break;
default:
break;
}
if (!declKind.empty()) {
const std::string &declName = ND->getDeclName().getAsString();
o << " <key>issue_context_kind</key>";
EmitString(o, declKind) << '\n';
o << " <key>issue_context</key>";
EmitString(o, declName) << '\n';
}
// Output the bug hash for issue unique-ing. Currently, it's just an
// offset from the beginning of the function.
if (const Stmt *Body = DeclWithIssue->getBody()) {
// If the bug uniqueing location exists, use it for the hash.
// For example, this ensures that two leaks reported on the same line
// will have different issue_hashes and that the hash will identify
// the leak location even after code is added between the allocation
// site and the end of scope (leak report location).
PathDiagnosticLocation UPDLoc = D->getUniqueingLoc();
if (UPDLoc.isValid()) {
FullSourceLoc UL(SM->getExpansionLoc(UPDLoc.asLocation()),
*SM);
FullSourceLoc UFunL(SM->getExpansionLoc(
D->getUniqueingDecl()->getBody()->getLocStart()), *SM);
o << " <key>issue_hash</key><string>"
<< UL.getExpansionLineNumber() - UFunL.getExpansionLineNumber()
<< "</string>\n";
// Otherwise, use the location on which the bug is reported.
} else {
FullSourceLoc L(SM->getExpansionLoc(D->getLocation().asLocation()),
*SM);
FullSourceLoc FunL(SM->getExpansionLoc(Body->getLocStart()), *SM);
o << " <key>issue_hash</key><string>"
<< L.getExpansionLineNumber() - FunL.getExpansionLineNumber()
<< "</string>\n";
}
}
}
}
// Output the location of the bug.
o << " <key>location</key>\n";
EmitLocation(o, *SM, LangOpts, D->getLocation().asLocation(), FM, 2);
// Output the diagnostic to the sub-diagnostic client, if any.
if (!filesMade->empty()) {
StringRef lastName;
PDFileEntry::ConsumerFiles *files = filesMade->getFiles(*D);
if (files) {
for (PDFileEntry::ConsumerFiles::const_iterator CI = files->begin(),
CE = files->end(); CI != CE; ++CI) {
StringRef newName = CI->first;
if (newName != lastName) {
if (!lastName.empty()) {
o << " </array>\n";
}
lastName = newName;
o << " <key>" << lastName << "_files</key>\n";
o << " <array>\n";
}
o << " <string>" << CI->second << "</string>\n";
}
o << " </array>\n";
}
}
// Close up the entry.
o << " </dict>\n";
}
o << " </array>\n";
// Finish.
o << "</dict>\n</plist>";
}
/// handleEndBlock - Remove dead stores to stack-allocated locations in the
/// function end block. Ex:
/// %A = alloca i32
/// ...
/// store i32 1, i32* %A
/// ret void
bool DSE::handleEndBlock(BasicBlock &BB) {
bool MadeChange = false;
// Keep track of all of the stack objects that are dead at the end of the
// function.
SmallSetVector<Value*, 16> DeadStackObjects;
// Find all of the alloca'd pointers in the entry block.
BasicBlock *Entry = BB.getParent()->begin();
for (BasicBlock::iterator I = Entry->begin(), E = Entry->end(); I != E; ++I) {
if (isa<AllocaInst>(I))
DeadStackObjects.insert(I);
// Okay, so these are dead heap objects, but if the pointer never escapes
// then it's leaked by this function anyways.
else if (isAllocLikeFn(I, TLI) && !PointerMayBeCaptured(I, true, true))
DeadStackObjects.insert(I);
}
// Treat byval arguments the same, stores to them are dead at the end of the
// function.
for (Function::arg_iterator AI = BB.getParent()->arg_begin(),
AE = BB.getParent()->arg_end(); AI != AE; ++AI)
if (AI->hasByValAttr())
DeadStackObjects.insert(AI);
// Scan the basic block backwards
for (BasicBlock::iterator BBI = BB.end(); BBI != BB.begin(); ){
--BBI;
// If we find a store, check to see if it points into a dead stack value.
if (hasMemoryWrite(BBI, TLI) && isRemovable(BBI)) {
// See through pointer-to-pointer bitcasts
SmallVector<Value *, 4> Pointers;
GetUnderlyingObjects(getStoredPointerOperand(BBI), Pointers);
// Stores to stack values are valid candidates for removal.
bool AllDead = true;
for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(),
E = Pointers.end(); I != E; ++I)
if (!DeadStackObjects.count(*I)) {
AllDead = false;
break;
}
if (AllDead) {
Instruction *Dead = BBI++;
DEBUG(dbgs() << "DSE: Dead Store at End of Block:\n DEAD: "
<< *Dead << "\n Objects: ";
for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(),
E = Pointers.end(); I != E; ++I) {
dbgs() << **I;
if (llvm::next(I) != E)
dbgs() << ", ";
}
dbgs() << '\n');
// DCE instructions only used to calculate that store.
DeleteDeadInstruction(Dead, *MD, TLI, &DeadStackObjects);
++NumFastStores;
MadeChange = true;
continue;
}
}
// Remove any dead non-memory-mutating instructions.
if (isInstructionTriviallyDead(BBI, TLI)) {
Instruction *Inst = BBI++;
DeleteDeadInstruction(Inst, *MD, TLI, &DeadStackObjects);
++NumFastOther;
MadeChange = true;
continue;
}
if (isa<AllocaInst>(BBI)) {
// Remove allocas from the list of dead stack objects; there can't be
// any references before the definition.
DeadStackObjects.remove(BBI);
continue;
}
if (CallSite CS = cast<Value>(BBI)) {
// Remove allocation function calls from the list of dead stack objects;
// there can't be any references before the definition.
if (isAllocLikeFn(BBI, TLI))
DeadStackObjects.remove(BBI);
// If this call does not access memory, it can't be loading any of our
// pointers.
if (AA->doesNotAccessMemory(CS))
continue;
// If the call might load from any of our allocas, then any store above
// the call is live.
CouldRef Pred = { CS, AA };
DeadStackObjects.remove_if(Pred);
// If all of the allocas were clobbered by the call then we're not going
// to find anything else to process.
if (DeadStackObjects.empty())
break;
continue;
}
AliasAnalysis::Location LoadedLoc;
// If we encounter a use of the pointer, it is no longer considered dead
if (LoadInst *L = dyn_cast<LoadInst>(BBI)) {
if (!L->isUnordered()) // Be conservative with atomic/volatile load
break;
LoadedLoc = AA->getLocation(L);
} else if (VAArgInst *V = dyn_cast<VAArgInst>(BBI)) {
LoadedLoc = AA->getLocation(V);
} else if (MemTransferInst *MTI = dyn_cast<MemTransferInst>(BBI)) {
LoadedLoc = AA->getLocationForSource(MTI);
} else if (!BBI->mayReadFromMemory()) {
// Instruction doesn't read memory. Note that stores that weren't removed
// above will hit this case.
continue;
} else {
// Unknown inst; assume it clobbers everything.
break;
}
// Remove any allocas from the DeadPointer set that are loaded, as this
// makes any stores above the access live.
RemoveAccessedObjects(LoadedLoc, DeadStackObjects);
// If all of the allocas were clobbered by the access then we're not going
// to find anything else to process.
if (DeadStackObjects.empty())
break;
}
bool MemDepPrinter::runOnFunction(Function &F) {
this->F = &F;
AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
MemoryDependenceAnalysis &MDA = getAnalysis<MemoryDependenceAnalysis>();
// All this code uses non-const interfaces because MemDep is not
// const-friendly, though nothing is actually modified.
for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) {
Instruction *Inst = &*I;
if (!Inst->mayReadFromMemory() && !Inst->mayWriteToMemory())
continue;
MemDepResult Res = MDA.getDependency(Inst);
if (!Res.isNonLocal()) {
Deps[Inst].insert(std::make_pair(getInstTypePair(Res),
static_cast<BasicBlock *>(nullptr)));
} else if (CallSite CS = cast<Value>(Inst)) {
const MemoryDependenceAnalysis::NonLocalDepInfo &NLDI =
MDA.getNonLocalCallDependency(CS);
DepSet &InstDeps = Deps[Inst];
for (MemoryDependenceAnalysis::NonLocalDepInfo::const_iterator
I = NLDI.begin(), E = NLDI.end(); I != E; ++I) {
const MemDepResult &Res = I->getResult();
InstDeps.insert(std::make_pair(getInstTypePair(Res), I->getBB()));
}
} else {
SmallVector<NonLocalDepResult, 4> NLDI;
if (LoadInst *LI = dyn_cast<LoadInst>(Inst)) {
if (!LI->isUnordered()) {
// FIXME: Handle atomic/volatile loads.
Deps[Inst].insert(std::make_pair(getInstTypePair(nullptr, Unknown),
static_cast<BasicBlock *>(nullptr)));
continue;
}
AliasAnalysis::Location Loc = AA.getLocation(LI);
MDA.getNonLocalPointerDependency(Loc, true, LI->getParent(), NLDI);
} else if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
if (!SI->isUnordered()) {
// FIXME: Handle atomic/volatile stores.
Deps[Inst].insert(std::make_pair(getInstTypePair(nullptr, Unknown),
static_cast<BasicBlock *>(nullptr)));
continue;
}
AliasAnalysis::Location Loc = AA.getLocation(SI);
MDA.getNonLocalPointerDependency(Loc, false, SI->getParent(), NLDI);
} else if (VAArgInst *VI = dyn_cast<VAArgInst>(Inst)) {
AliasAnalysis::Location Loc = AA.getLocation(VI);
MDA.getNonLocalPointerDependency(Loc, false, VI->getParent(), NLDI);
} else {
llvm_unreachable("Unknown memory instruction!");
}
DepSet &InstDeps = Deps[Inst];
for (SmallVectorImpl<NonLocalDepResult>::const_iterator
I = NLDI.begin(), E = NLDI.end(); I != E; ++I) {
const MemDepResult &Res = I->getResult();
InstDeps.insert(std::make_pair(getInstTypePair(Res), I->getBB()));
}
}
}
return false;
}
/// IsProfitableToFoldIntoAddressingMode - It is possible for the addressing
/// mode of the machine to fold the specified instruction into a load or store
/// that ultimately uses it. However, the specified instruction has multiple
/// uses. Given this, it may actually increase register pressure to fold it
/// into the load. For example, consider this code:
///
/// X = ...
/// Y = X+1
/// use(Y) -> nonload/store
/// Z = Y+1
/// load Z
///
/// In this case, Y has multiple uses, and can be folded into the load of Z
/// (yielding load [X+2]). However, doing this will cause both "X" and "X+1" to
/// be live at the use(Y) line. If we don't fold Y into load Z, we use one
/// fewer register. Since Y can't be folded into "use(Y)" we don't increase the
/// number of computations either.
///
/// Note that this (like most of CodeGenPrepare) is just a rough heuristic. If
/// X was live across 'load Z' for other reasons, we actually *would* want to
/// fold the addressing mode in the Z case. This would make Y die earlier.
bool AddressingModeMatcher::
IsProfitableToFoldIntoAddressingMode(Instruction *I, ExtAddrMode &AMBefore,
ExtAddrMode &AMAfter) {
if (IgnoreProfitability) return true;
// AMBefore is the addressing mode before this instruction was folded into it,
// and AMAfter is the addressing mode after the instruction was folded. Get
// the set of registers referenced by AMAfter and subtract out those
// referenced by AMBefore: this is the set of values which folding in this
// address extends the lifetime of.
//
// Note that there are only two potential values being referenced here,
// BaseReg and ScaleReg (global addresses are always available, as are any
// folded immediates).
Value *BaseReg = AMAfter.BaseReg, *ScaledReg = AMAfter.ScaledReg;
// If the BaseReg or ScaledReg was referenced by the previous addrmode, their
// lifetime wasn't extended by adding this instruction.
if (ValueAlreadyLiveAtInst(BaseReg, AMBefore.BaseReg, AMBefore.ScaledReg))
BaseReg = 0;
if (ValueAlreadyLiveAtInst(ScaledReg, AMBefore.BaseReg, AMBefore.ScaledReg))
ScaledReg = 0;
// If folding this instruction (and it's subexprs) didn't extend any live
// ranges, we're ok with it.
if (BaseReg == 0 && ScaledReg == 0)
return true;
// If all uses of this instruction are ultimately load/store/inlineasm's,
// check to see if their addressing modes will include this instruction. If
// so, we can fold it into all uses, so it doesn't matter if it has multiple
// uses.
SmallVector<std::pair<Instruction*,unsigned>, 16> MemoryUses;
SmallPtrSet<Instruction*, 16> ConsideredInsts;
if (FindAllMemoryUses(I, MemoryUses, ConsideredInsts, TLI))
return false; // Has a non-memory, non-foldable use!
// Now that we know that all uses of this instruction are part of a chain of
// computation involving only operations that could theoretically be folded
// into a memory use, loop over each of these uses and see if they could
// *actually* fold the instruction.
SmallVector<Instruction*, 32> MatchedAddrModeInsts;
for (unsigned i = 0, e = MemoryUses.size(); i != e; ++i) {
Instruction *User = MemoryUses[i].first;
unsigned OpNo = MemoryUses[i].second;
// Get the access type of this use. If the use isn't a pointer, we don't
// know what it accesses.
Value *Address = User->getOperand(OpNo);
if (!Address->getType()->isPointerTy())
return false;
const Type *AddressAccessTy =
cast<PointerType>(Address->getType())->getElementType();
// Do a match against the root of this address, ignoring profitability. This
// will tell us if the addressing mode for the memory operation will
// *actually* cover the shared instruction.
ExtAddrMode Result;
AddressingModeMatcher Matcher(MatchedAddrModeInsts, TLI, AddressAccessTy,
MemoryInst, Result);
Matcher.IgnoreProfitability = true;
bool Success = Matcher.MatchAddr(Address, 0);
Success = Success; assert(Success && "Couldn't select *anything*?");
// If the match didn't cover I, then it won't be shared by it.
if (std::find(MatchedAddrModeInsts.begin(), MatchedAddrModeInsts.end(),
I) == MatchedAddrModeInsts.end())
return false;
MatchedAddrModeInsts.clear();
}
return true;
}
bool AArch64FrameLowering::restoreCalleeSavedRegisters(
MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
const std::vector<CalleeSavedInfo> &CSI,
const TargetRegisterInfo *TRI) const {
MachineFunction &MF = *MBB.getParent();
const TargetInstrInfo &TII = *MF.getSubtarget().getInstrInfo();
DebugLoc DL;
SmallVector<RegPairInfo, 8> RegPairs;
if (MI != MBB.end())
DL = MI->getDebugLoc();
computeCalleeSaveRegisterPairs(MF, CSI, TRI, RegPairs);
for (auto RPII = RegPairs.begin(), RPIE = RegPairs.end(); RPII != RPIE;
++RPII) {
RegPairInfo RPI = *RPII;
unsigned Reg1 = RPI.Reg1;
unsigned Reg2 = RPI.Reg2;
// Issue sequence of non-sp increment and sp-pi restores for cs regs. Only
// the last load is sp-pi post-increment and de-allocates the stack:
// For example:
// ldp fp, lr, [sp, #32] // addImm(+4)
// ldp x20, x19, [sp, #16] // addImm(+2)
// ldp x22, x21, [sp], #48 // addImm(+6)
// Note: see comment in spillCalleeSavedRegisters()
unsigned LdrOpc;
bool BumpSP = RPII == std::prev(RegPairs.end());
if (RPI.IsGPR) {
if (BumpSP)
LdrOpc = RPI.isPaired() ? AArch64::LDPXpost : AArch64::LDRXpost;
else
LdrOpc = RPI.isPaired() ? AArch64::LDPXi : AArch64::LDRXui;
} else {
if (BumpSP)
LdrOpc = RPI.isPaired() ? AArch64::LDPDpost : AArch64::LDRDpost;
else
LdrOpc = RPI.isPaired() ? AArch64::LDPDi : AArch64::LDRDui;
}
DEBUG(dbgs() << "CSR restore: (" << TRI->getName(Reg1);
if (RPI.isPaired())
dbgs() << ", " << TRI->getName(Reg2);
dbgs() << ") -> fi#(" << RPI.FrameIdx;
if (RPI.isPaired())
dbgs() << ", " << RPI.FrameIdx+1;
dbgs() << ")\n");
const int Offset = RPI.Offset;
MachineInstrBuilder MIB = BuildMI(MBB, MI, DL, TII.get(LdrOpc));
if (BumpSP)
MIB.addReg(AArch64::SP, RegState::Define);
if (RPI.isPaired())
MIB.addReg(Reg2, getDefRegState(true))
.addReg(Reg1, getDefRegState(true))
.addReg(AArch64::SP)
.addImm(Offset) // [sp], #offset * 8 or [sp, #offset * 8]
// where the factor * 8 is implicit
.setMIFlag(MachineInstr::FrameDestroy);
else
MIB.addReg(Reg1, getDefRegState(true))
.addReg(AArch64::SP)
.addImm(BumpSP ? Offset * 8 : Offset) // post-dec version is unscaled
.setMIFlag(MachineInstr::FrameDestroy);
}
return true;
}
/// Split - Splits a string of comma separated items in to a vector of strings.
///
static void Split(std::vector<std::string> &V, StringRef S) {
SmallVector<StringRef, 3> Tmp;
S.split(Tmp, ',', -1, false /* KeepEmpty */);
V.assign(Tmp.begin(), Tmp.end());
}
/// calculateFrameObjectOffsets - Calculate actual frame offsets for all of the
/// abstract stack objects.
///
void PEI::calculateFrameObjectOffsets(MachineFunction &Fn) {
const TargetFrameLowering &TFI = *Fn.getSubtarget().getFrameLowering();
StackProtector *SP = &getAnalysis<StackProtector>();
bool StackGrowsDown =
TFI.getStackGrowthDirection() == TargetFrameLowering::StackGrowsDown;
// Loop over all of the stack objects, assigning sequential addresses...
MachineFrameInfo *MFI = Fn.getFrameInfo();
// Start at the beginning of the local area.
// The Offset is the distance from the stack top in the direction
// of stack growth -- so it's always nonnegative.
int LocalAreaOffset = TFI.getOffsetOfLocalArea();
if (StackGrowsDown)
LocalAreaOffset = -LocalAreaOffset;
assert(LocalAreaOffset >= 0
&& "Local area offset should be in direction of stack growth");
int64_t Offset = LocalAreaOffset;
// If there are fixed sized objects that are preallocated in the local area,
// non-fixed objects can't be allocated right at the start of local area.
// We currently don't support filling in holes in between fixed sized
// objects, so we adjust 'Offset' to point to the end of last fixed sized
// preallocated object.
for (int i = MFI->getObjectIndexBegin(); i != 0; ++i) {
int64_t FixedOff;
if (StackGrowsDown) {
// The maximum distance from the stack pointer is at lower address of
// the object -- which is given by offset. For down growing stack
// the offset is negative, so we negate the offset to get the distance.
FixedOff = -MFI->getObjectOffset(i);
} else {
// The maximum distance from the start pointer is at the upper
// address of the object.
FixedOff = MFI->getObjectOffset(i) + MFI->getObjectSize(i);
}
if (FixedOff > Offset) Offset = FixedOff;
}
// First assign frame offsets to stack objects that are used to spill
// callee saved registers.
if (StackGrowsDown) {
for (unsigned i = MinCSFrameIndex; i <= MaxCSFrameIndex; ++i) {
// If the stack grows down, we need to add the size to find the lowest
// address of the object.
Offset += MFI->getObjectSize(i);
unsigned Align = MFI->getObjectAlignment(i);
// Adjust to alignment boundary
Offset = RoundUpToAlignment(Offset, Align);
MFI->setObjectOffset(i, -Offset); // Set the computed offset
}
} else {
int MaxCSFI = MaxCSFrameIndex, MinCSFI = MinCSFrameIndex;
for (int i = MaxCSFI; i >= MinCSFI ; --i) {
unsigned Align = MFI->getObjectAlignment(i);
// Adjust to alignment boundary
Offset = RoundUpToAlignment(Offset, Align);
MFI->setObjectOffset(i, Offset);
Offset += MFI->getObjectSize(i);
}
}
unsigned MaxAlign = MFI->getMaxAlignment();
// Make sure the special register scavenging spill slot is closest to the
// incoming stack pointer if a frame pointer is required and is closer
// to the incoming rather than the final stack pointer.
const TargetRegisterInfo *RegInfo = Fn.getSubtarget().getRegisterInfo();
bool EarlyScavengingSlots = (TFI.hasFP(Fn) &&
TFI.isFPCloseToIncomingSP() &&
RegInfo->useFPForScavengingIndex(Fn) &&
!RegInfo->needsStackRealignment(Fn));
if (RS && EarlyScavengingSlots) {
SmallVector<int, 2> SFIs;
RS->getScavengingFrameIndices(SFIs);
for (SmallVectorImpl<int>::iterator I = SFIs.begin(),
IE = SFIs.end(); I != IE; ++I)
AdjustStackOffset(MFI, *I, StackGrowsDown, Offset, MaxAlign);
}
// FIXME: Once this is working, then enable flag will change to a target
// check for whether the frame is large enough to want to use virtual
// frame index registers. Functions which don't want/need this optimization
// will continue to use the existing code path.
if (MFI->getUseLocalStackAllocationBlock()) {
unsigned Align = MFI->getLocalFrameMaxAlign();
// Adjust to alignment boundary.
Offset = RoundUpToAlignment(Offset, Align);
DEBUG(dbgs() << "Local frame base offset: " << Offset << "\n");
// Resolve offsets for objects in the local block.
for (unsigned i = 0, e = MFI->getLocalFrameObjectCount(); i != e; ++i) {
std::pair<int, int64_t> Entry = MFI->getLocalFrameObjectMap(i);
int64_t FIOffset = (StackGrowsDown ? -Offset : Offset) + Entry.second;
DEBUG(dbgs() << "alloc FI(" << Entry.first << ") at SP[" <<
FIOffset << "]\n");
MFI->setObjectOffset(Entry.first, FIOffset);
}
// Allocate the local block
Offset += MFI->getLocalFrameSize();
MaxAlign = std::max(Align, MaxAlign);
}
// Make sure that the stack protector comes before the local variables on the
// stack.
SmallSet<int, 16> ProtectedObjs;
if (MFI->getStackProtectorIndex() >= 0) {
StackObjSet LargeArrayObjs;
StackObjSet SmallArrayObjs;
StackObjSet AddrOfObjs;
AdjustStackOffset(MFI, MFI->getStackProtectorIndex(), StackGrowsDown,
Offset, MaxAlign);
// Assign large stack objects first.
for (unsigned i = 0, e = MFI->getObjectIndexEnd(); i != e; ++i) {
if (MFI->isObjectPreAllocated(i) &&
MFI->getUseLocalStackAllocationBlock())
continue;
if (i >= MinCSFrameIndex && i <= MaxCSFrameIndex)
continue;
if (RS && RS->isScavengingFrameIndex((int)i))
continue;
if (MFI->isDeadObjectIndex(i))
continue;
if (MFI->getStackProtectorIndex() == (int)i)
continue;
switch (SP->getSSPLayout(MFI->getObjectAllocation(i))) {
case StackProtector::SSPLK_None:
continue;
case StackProtector::SSPLK_SmallArray:
SmallArrayObjs.insert(i);
continue;
case StackProtector::SSPLK_AddrOf:
AddrOfObjs.insert(i);
continue;
case StackProtector::SSPLK_LargeArray:
LargeArrayObjs.insert(i);
continue;
}
llvm_unreachable("Unexpected SSPLayoutKind.");
}
AssignProtectedObjSet(LargeArrayObjs, ProtectedObjs, MFI, StackGrowsDown,
Offset, MaxAlign);
AssignProtectedObjSet(SmallArrayObjs, ProtectedObjs, MFI, StackGrowsDown,
Offset, MaxAlign);
AssignProtectedObjSet(AddrOfObjs, ProtectedObjs, MFI, StackGrowsDown,
Offset, MaxAlign);
}
// Then assign frame offsets to stack objects that are not used to spill
// callee saved registers.
for (unsigned i = 0, e = MFI->getObjectIndexEnd(); i != e; ++i) {
if (MFI->isObjectPreAllocated(i) &&
MFI->getUseLocalStackAllocationBlock())
continue;
if (i >= MinCSFrameIndex && i <= MaxCSFrameIndex)
continue;
if (RS && RS->isScavengingFrameIndex((int)i))
continue;
if (MFI->isDeadObjectIndex(i))
continue;
if (MFI->getStackProtectorIndex() == (int)i)
continue;
if (ProtectedObjs.count(i))
continue;
AdjustStackOffset(MFI, i, StackGrowsDown, Offset, MaxAlign);
}
// Make sure the special register scavenging spill slot is closest to the
// stack pointer.
if (RS && !EarlyScavengingSlots) {
SmallVector<int, 2> SFIs;
RS->getScavengingFrameIndices(SFIs);
for (SmallVectorImpl<int>::iterator I = SFIs.begin(),
IE = SFIs.end(); I != IE; ++I)
AdjustStackOffset(MFI, *I, StackGrowsDown, Offset, MaxAlign);
}
if (!TFI.targetHandlesStackFrameRounding()) {
// If we have reserved argument space for call sites in the function
// immediately on entry to the current function, count it as part of the
// overall stack size.
if (MFI->adjustsStack() && TFI.hasReservedCallFrame(Fn))
Offset += MFI->getMaxCallFrameSize();
// Round up the size to a multiple of the alignment. If the function has
// any calls or alloca's, align to the target's StackAlignment value to
// ensure that the callee's frame or the alloca data is suitably aligned;
// otherwise, for leaf functions, align to the TransientStackAlignment
// value.
unsigned StackAlign;
if (MFI->adjustsStack() || MFI->hasVarSizedObjects() ||
(RegInfo->needsStackRealignment(Fn) && MFI->getObjectIndexEnd() != 0))
StackAlign = TFI.getStackAlignment();
else
StackAlign = TFI.getTransientStackAlignment();
// If the frame pointer is eliminated, all frame offsets will be relative to
// SP not FP. Align to MaxAlign so this works.
StackAlign = std::max(StackAlign, MaxAlign);
Offset = RoundUpToAlignment(Offset, StackAlign);
}
// Update frame info to pretend that this is part of the stack...
int64_t StackSize = Offset - LocalAreaOffset;
MFI->setStackSize(StackSize);
NumBytesStackSpace += StackSize;
}
Value *SSAUpdater::GetValueInMiddleOfBlock(BasicBlock *BB) {
// If there is no definition of the renamed variable in this block, just use
// GetValueAtEndOfBlock to do our work.
if (!HasValueForBlock(BB))
return GetValueAtEndOfBlock(BB);
// Otherwise, we have the hard case. Get the live-in values for each
// predecessor.
SmallVector<std::pair<BasicBlock*, Value*>, 8> PredValues;
Value *SingularValue = nullptr;
// We can get our predecessor info by walking the pred_iterator list, but it
// is relatively slow. If we already have PHI nodes in this block, walk one
// of them to get the predecessor list instead.
if (PHINode *SomePhi = dyn_cast<PHINode>(BB->begin())) {
for (unsigned i = 0, e = SomePhi->getNumIncomingValues(); i != e; ++i) {
BasicBlock *PredBB = SomePhi->getIncomingBlock(i);
Value *PredVal = GetValueAtEndOfBlock(PredBB);
PredValues.push_back(std::make_pair(PredBB, PredVal));
// Compute SingularValue.
if (i == 0)
SingularValue = PredVal;
else if (PredVal != SingularValue)
SingularValue = nullptr;
}
} else {
bool isFirstPred = true;
for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
BasicBlock *PredBB = *PI;
Value *PredVal = GetValueAtEndOfBlock(PredBB);
PredValues.push_back(std::make_pair(PredBB, PredVal));
// Compute SingularValue.
if (isFirstPred) {
SingularValue = PredVal;
isFirstPred = false;
} else if (PredVal != SingularValue)
SingularValue = nullptr;
}
}
// If there are no predecessors, just return undef.
if (PredValues.empty())
return UndefValue::get(ProtoType);
// Otherwise, if all the merged values are the same, just use it.
if (SingularValue)
return SingularValue;
// Otherwise, we do need a PHI: check to see if we already have one available
// in this block that produces the right value.
if (isa<PHINode>(BB->begin())) {
SmallDenseMap<BasicBlock*, Value*, 8> ValueMapping(PredValues.begin(),
PredValues.end());
PHINode *SomePHI;
for (BasicBlock::iterator It = BB->begin();
(SomePHI = dyn_cast<PHINode>(It)); ++It) {
if (IsEquivalentPHI(SomePHI, ValueMapping))
return SomePHI;
}
}
// Ok, we have no way out, insert a new one now.
PHINode *InsertedPHI = PHINode::Create(ProtoType, PredValues.size(),
ProtoName, &BB->front());
// Fill in all the predecessors of the PHI.
for (unsigned i = 0, e = PredValues.size(); i != e; ++i)
InsertedPHI->addIncoming(PredValues[i].second, PredValues[i].first);
// See if the PHI node can be merged to a single value. This can happen in
// loop cases when we get a PHI of itself and one other value.
if (Value *V =
SimplifyInstruction(InsertedPHI, BB->getModule()->getDataLayout())) {
InsertedPHI->eraseFromParent();
return V;
}
// Set the DebugLoc of the inserted PHI, if available.
DebugLoc DL;
if (const Instruction *I = BB->getFirstNonPHI())
DL = I->getDebugLoc();
InsertedPHI->setDebugLoc(DL);
// If the client wants to know about all new instructions, tell it.
if (InsertedPHIs) InsertedPHIs->push_back(InsertedPHI);
DEBUG(dbgs() << " Inserted PHI: " << *InsertedPHI << "\n");
return InsertedPHI;
}
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;
}
bool StackColoring::runOnMachineFunction(MachineFunction &Func) {
DEBUG(dbgs() << "********** Stack Coloring **********\n"
<< "********** Function: "
<< ((const Value*)Func.getFunction())->getName() << '\n');
MF = &Func;
MFI = MF->getFrameInfo();
Indexes = &getAnalysis<SlotIndexes>();
SP = &getAnalysis<StackProtector>();
BlockLiveness.clear();
BasicBlocks.clear();
BasicBlockNumbering.clear();
Markers.clear();
Intervals.clear();
VNInfoAllocator.Reset();
unsigned NumSlots = MFI->getObjectIndexEnd();
// If there are no stack slots then there are no markers to remove.
if (!NumSlots)
return false;
SmallVector<int, 8> SortedSlots;
SortedSlots.reserve(NumSlots);
Intervals.reserve(NumSlots);
unsigned NumMarkers = collectMarkers(NumSlots);
unsigned TotalSize = 0;
DEBUG(dbgs()<<"Found "<<NumMarkers<<" markers and "<<NumSlots<<" slots\n");
DEBUG(dbgs()<<"Slot structure:\n");
for (int i=0; i < MFI->getObjectIndexEnd(); ++i) {
DEBUG(dbgs()<<"Slot #"<<i<<" - "<<MFI->getObjectSize(i)<<" bytes.\n");
TotalSize += MFI->getObjectSize(i);
}
DEBUG(dbgs()<<"Total Stack size: "<<TotalSize<<" bytes\n\n");
// Don't continue because there are not enough lifetime markers, or the
// stack is too small, or we are told not to optimize the slots.
if (NumMarkers < 2 || TotalSize < 16 || DisableColoring ||
skipFunction(*Func.getFunction())) {
DEBUG(dbgs()<<"Will not try to merge slots.\n");
return removeAllMarkers();
}
for (unsigned i=0; i < NumSlots; ++i) {
std::unique_ptr<LiveInterval> LI(new LiveInterval(i, 0));
LI->getNextValue(Indexes->getZeroIndex(), VNInfoAllocator);
Intervals.push_back(std::move(LI));
SortedSlots.push_back(i);
}
// Calculate the liveness of each block.
calculateLocalLiveness();
DEBUG(dbgs() << "Dataflow iterations: " << NumIterations << "\n");
DEBUG(dump());
// Propagate the liveness information.
calculateLiveIntervals(NumSlots);
DEBUG(dumpIntervals());
// Search for allocas which are used outside of the declared lifetime
// markers.
if (ProtectFromEscapedAllocas)
removeInvalidSlotRanges();
// Maps old slots to new slots.
DenseMap<int, int> SlotRemap;
unsigned RemovedSlots = 0;
unsigned ReducedSize = 0;
// Do not bother looking at empty intervals.
for (unsigned I = 0; I < NumSlots; ++I) {
if (Intervals[SortedSlots[I]]->empty())
SortedSlots[I] = -1;
}
// This is a simple greedy algorithm for merging allocas. First, sort the
// slots, placing the largest slots first. Next, perform an n^2 scan and look
// for disjoint slots. When you find disjoint slots, merge the samller one
// into the bigger one and update the live interval. Remove the small alloca
// and continue.
// Sort the slots according to their size. Place unused slots at the end.
// Use stable sort to guarantee deterministic code generation.
std::stable_sort(SortedSlots.begin(), SortedSlots.end(),
[this](int LHS, int RHS) {
// We use -1 to denote a uninteresting slot. Place these slots at the end.
if (LHS == -1) return false;
if (RHS == -1) return true;
// Sort according to size.
return MFI->getObjectSize(LHS) > MFI->getObjectSize(RHS);
});
bool Changed = true;
while (Changed) {
Changed = false;
for (unsigned I = 0; I < NumSlots; ++I) {
if (SortedSlots[I] == -1)
continue;
for (unsigned J=I+1; J < NumSlots; ++J) {
if (SortedSlots[J] == -1)
continue;
int FirstSlot = SortedSlots[I];
int SecondSlot = SortedSlots[J];
LiveInterval *First = &*Intervals[FirstSlot];
LiveInterval *Second = &*Intervals[SecondSlot];
assert (!First->empty() && !Second->empty() && "Found an empty range");
// Merge disjoint slots.
if (!First->overlaps(*Second)) {
Changed = true;
First->MergeSegmentsInAsValue(*Second, First->getValNumInfo(0));
SlotRemap[SecondSlot] = FirstSlot;
SortedSlots[J] = -1;
DEBUG(dbgs()<<"Merging #"<<FirstSlot<<" and slots #"<<
SecondSlot<<" together.\n");
unsigned MaxAlignment = std::max(MFI->getObjectAlignment(FirstSlot),
MFI->getObjectAlignment(SecondSlot));
assert(MFI->getObjectSize(FirstSlot) >=
MFI->getObjectSize(SecondSlot) &&
"Merging a small object into a larger one");
RemovedSlots+=1;
ReducedSize += MFI->getObjectSize(SecondSlot);
MFI->setObjectAlignment(FirstSlot, MaxAlignment);
MFI->RemoveStackObject(SecondSlot);
}
}
}
}// While changed.
// Record statistics.
StackSpaceSaved += ReducedSize;
StackSlotMerged += RemovedSlots;
DEBUG(dbgs()<<"Merge "<<RemovedSlots<<" slots. Saved "<<
ReducedSize<<" bytes\n");
// Scan the entire function and update all machine operands that use frame
// indices to use the remapped frame index.
expungeSlotMap(SlotRemap, NumSlots);
remapInstructions(SlotRemap);
return removeAllMarkers();
}
std::vector<std::string> ArgList::getAllArgValues(OptSpecifier Id) const {
SmallVector<const char *, 16> Values;
AddAllArgValues(Values, Id);
return std::vector<std::string>(Values.begin(), Values.end());
}
// NewBB is split and now it has one successor. Update dominance frontier to
// reflect this change.
void DominanceFrontier::splitBlock(BasicBlock *NewBB) {
assert(NewBB->getTerminator()->getNumSuccessors() == 1
&& "NewBB should have a single successor!");
BasicBlock *NewBBSucc = NewBB->getTerminator()->getSuccessor(0);
SmallVector<BasicBlock*, 8> PredBlocks;
for (pred_iterator PI = pred_begin(NewBB), PE = pred_end(NewBB);
PI != PE; ++PI)
PredBlocks.push_back(*PI);
if (PredBlocks.empty())
// If NewBB does not have any predecessors then it is a entry block.
// In this case, NewBB and its successor NewBBSucc dominates all
// other blocks.
return;
// NewBBSucc inherits original NewBB frontier.
DominanceFrontier::iterator NewBBI = find(NewBB);
if (NewBBI != end()) {
DominanceFrontier::DomSetType NewBBSet = NewBBI->second;
DominanceFrontier::DomSetType NewBBSuccSet;
NewBBSuccSet.insert(NewBBSet.begin(), NewBBSet.end());
addBasicBlock(NewBBSucc, NewBBSuccSet);
}
// If NewBB dominates NewBBSucc, then DF(NewBB) is now going to be the
// DF(NewBBSucc) without the stuff that the new block does not dominate
// a predecessor of.
DominatorTree &DT = getAnalysis<DominatorTree>();
if (DT.dominates(NewBB, NewBBSucc)) {
DominanceFrontier::iterator DFI = find(NewBBSucc);
if (DFI != end()) {
DominanceFrontier::DomSetType Set = DFI->second;
// Filter out stuff in Set that we do not dominate a predecessor of.
for (DominanceFrontier::DomSetType::iterator SetI = Set.begin(),
E = Set.end(); SetI != E;) {
bool DominatesPred = false;
for (pred_iterator PI = pred_begin(*SetI), E = pred_end(*SetI);
PI != E; ++PI)
if (DT.dominates(NewBB, *PI))
DominatesPred = true;
if (!DominatesPred)
Set.erase(SetI++);
else
++SetI;
}
if (NewBBI != end()) {
for (DominanceFrontier::DomSetType::iterator SetI = Set.begin(),
E = Set.end(); SetI != E; ++SetI) {
BasicBlock *SB = *SetI;
addToFrontier(NewBBI, SB);
}
} else
addBasicBlock(NewBB, Set);
}
} else {
// DF(NewBB) is {NewBBSucc} because NewBB does not strictly dominate
// NewBBSucc, but it does dominate itself (and there is an edge (NewBB ->
// NewBBSucc)). NewBBSucc is the single successor of NewBB.
DominanceFrontier::DomSetType NewDFSet;
NewDFSet.insert(NewBBSucc);
addBasicBlock(NewBB, NewDFSet);
}
// Now we must loop over all of the dominance frontiers in the function,
// replacing occurrences of NewBBSucc with NewBB in some cases. All
// blocks that dominate a block in PredBlocks and contained NewBBSucc in
// their dominance frontier must be updated to contain NewBB instead.
//
for (Function::iterator FI = NewBB->getParent()->begin(),
FE = NewBB->getParent()->end(); FI != FE; ++FI) {
DominanceFrontier::iterator DFI = find(FI);
if (DFI == end()) continue; // unreachable block.
// Only consider nodes that have NewBBSucc in their dominator frontier.
if (!DFI->second.count(NewBBSucc)) continue;
// Verify whether this block dominates a block in predblocks. If not, do
// not update it.
bool BlockDominatesAny = false;
for (SmallVectorImpl<BasicBlock*>::const_iterator BI = PredBlocks.begin(),
BE = PredBlocks.end(); BI != BE; ++BI) {
if (DT.dominates(FI, *BI)) {
BlockDominatesAny = true;
break;
}
}
// If NewBBSucc should not stay in our dominator frontier, remove it.
// We remove it unless there is a predecessor of NewBBSucc that we
// dominate, but we don't strictly dominate NewBBSucc.
bool ShouldRemove = true;
if ((BasicBlock*)FI == NewBBSucc || !DT.dominates(FI, NewBBSucc)) {
// Okay, we know that PredDom does not strictly dominate NewBBSucc.
// Check to see if it dominates any predecessors of NewBBSucc.
for (pred_iterator PI = pred_begin(NewBBSucc),
E = pred_end(NewBBSucc); PI != E; ++PI)
if (DT.dominates(FI, *PI)) {
ShouldRemove = false;
break;
}
}
if (ShouldRemove)
removeFromFrontier(DFI, NewBBSucc);
if (BlockDominatesAny && (&*FI == NewBB || !DT.dominates(FI, NewBB)))
addToFrontier(DFI, NewBB);
}
}
/// TailDuplicate - If it is profitable, duplicate TailBB's contents in each
/// of its predecessors.
bool
TailDuplicatePass::TailDuplicate(MachineBasicBlock *TailBB,
bool IsSimple,
MachineFunction &MF,
SmallVector<MachineBasicBlock*, 8> &TDBBs,
SmallVector<MachineInstr*, 16> &Copies) {
DEBUG(dbgs() << "\n*** Tail-duplicating BB#" << TailBB->getNumber() << '\n');
DenseSet<unsigned> UsedByPhi;
getRegsUsedByPHIs(*TailBB, &UsedByPhi);
if (IsSimple)
return duplicateSimpleBB(TailBB, TDBBs, UsedByPhi, Copies);
// Iterate through all the unique predecessors and tail-duplicate this
// block into them, if possible. Copying the list ahead of time also
// avoids trouble with the predecessor list reallocating.
bool Changed = false;
SmallSetVector<MachineBasicBlock*, 8> Preds(TailBB->pred_begin(),
TailBB->pred_end());
for (SmallSetVector<MachineBasicBlock *, 8>::iterator PI = Preds.begin(),
PE = Preds.end(); PI != PE; ++PI) {
MachineBasicBlock *PredBB = *PI;
assert(TailBB != PredBB &&
"Single-block loop should have been rejected earlier!");
// EH edges are ignored by AnalyzeBranch.
if (PredBB->succ_size() > 1)
continue;
MachineBasicBlock *PredTBB, *PredFBB;
SmallVector<MachineOperand, 4> PredCond;
if (TII->AnalyzeBranch(*PredBB, PredTBB, PredFBB, PredCond, true))
continue;
if (!PredCond.empty())
continue;
// Don't duplicate into a fall-through predecessor (at least for now).
if (PredBB->isLayoutSuccessor(TailBB) && PredBB->canFallThrough())
continue;
DEBUG(dbgs() << "\nTail-duplicating into PredBB: " << *PredBB
<< "From Succ: " << *TailBB);
TDBBs.push_back(PredBB);
// Remove PredBB's unconditional branch.
TII->RemoveBranch(*PredBB);
if (RS && !TailBB->livein_empty()) {
// Update PredBB livein.
RS->enterBasicBlock(PredBB);
if (!PredBB->empty())
RS->forward(prior(PredBB->end()));
BitVector RegsLiveAtExit(TRI->getNumRegs());
RS->getRegsUsed(RegsLiveAtExit, false);
for (MachineBasicBlock::livein_iterator I = TailBB->livein_begin(),
E = TailBB->livein_end(); I != E; ++I) {
if (!RegsLiveAtExit[*I])
// If a register is previously livein to the tail but it's not live
// at the end of predecessor BB, then it should be added to its
// livein list.
PredBB->addLiveIn(*I);
}
}
// Clone the contents of TailBB into PredBB.
DenseMap<unsigned, unsigned> LocalVRMap;
SmallVector<std::pair<unsigned,unsigned>, 4> CopyInfos;
// Use instr_iterator here to properly handle bundles, e.g.
// ARM Thumb2 IT block.
MachineBasicBlock::instr_iterator I = TailBB->instr_begin();
while (I != TailBB->instr_end()) {
MachineInstr *MI = &*I;
++I;
if (MI->isPHI()) {
// Replace the uses of the def of the PHI with the register coming
// from PredBB.
ProcessPHI(MI, TailBB, PredBB, LocalVRMap, CopyInfos, UsedByPhi, true);
} else {
// Replace def of virtual registers with new registers, and update
// uses with PHI source register or the new registers.
DuplicateInstruction(MI, TailBB, PredBB, MF, LocalVRMap, UsedByPhi);
}
}
MachineBasicBlock::iterator Loc = PredBB->getFirstTerminator();
for (unsigned i = 0, e = CopyInfos.size(); i != e; ++i) {
Copies.push_back(BuildMI(*PredBB, Loc, DebugLoc(),
TII->get(TargetOpcode::COPY),
CopyInfos[i].first).addReg(CopyInfos[i].second));
}
// Simplify
TII->AnalyzeBranch(*PredBB, PredTBB, PredFBB, PredCond, true);
NumInstrDups += TailBB->size() - 1; // subtract one for removed branch
// Update the CFG.
PredBB->removeSuccessor(PredBB->succ_begin());
assert(PredBB->succ_empty() &&
"TailDuplicate called on block with multiple successors!");
for (MachineBasicBlock::succ_iterator I = TailBB->succ_begin(),
E = TailBB->succ_end(); I != E; ++I)
PredBB->addSuccessor(*I);
Changed = true;
++NumTailDups;
}
// If TailBB was duplicated into all its predecessors except for the prior
// block, which falls through unconditionally, move the contents of this
// block into the prior block.
MachineBasicBlock *PrevBB = prior(MachineFunction::iterator(TailBB));
MachineBasicBlock *PriorTBB = 0, *PriorFBB = 0;
SmallVector<MachineOperand, 4> PriorCond;
// This has to check PrevBB->succ_size() because EH edges are ignored by
// AnalyzeBranch.
if (PrevBB->succ_size() == 1 &&
!TII->AnalyzeBranch(*PrevBB, PriorTBB, PriorFBB, PriorCond, true) &&
PriorCond.empty() && !PriorTBB && TailBB->pred_size() == 1 &&
!TailBB->hasAddressTaken()) {
DEBUG(dbgs() << "\nMerging into block: " << *PrevBB
<< "From MBB: " << *TailBB);
if (PreRegAlloc) {
DenseMap<unsigned, unsigned> LocalVRMap;
SmallVector<std::pair<unsigned,unsigned>, 4> CopyInfos;
MachineBasicBlock::iterator I = TailBB->begin();
// Process PHI instructions first.
while (I != TailBB->end() && I->isPHI()) {
// Replace the uses of the def of the PHI with the register coming
// from PredBB.
MachineInstr *MI = &*I++;
ProcessPHI(MI, TailBB, PrevBB, LocalVRMap, CopyInfos, UsedByPhi, true);
if (MI->getParent())
MI->eraseFromParent();
}
// Now copy the non-PHI instructions.
while (I != TailBB->end()) {
// Replace def of virtual registers with new registers, and update
// uses with PHI source register or the new registers.
MachineInstr *MI = &*I++;
assert(!MI->isBundle() && "Not expecting bundles before regalloc!");
DuplicateInstruction(MI, TailBB, PrevBB, MF, LocalVRMap, UsedByPhi);
MI->eraseFromParent();
}
MachineBasicBlock::iterator Loc = PrevBB->getFirstTerminator();
for (unsigned i = 0, e = CopyInfos.size(); i != e; ++i) {
Copies.push_back(BuildMI(*PrevBB, Loc, DebugLoc(),
TII->get(TargetOpcode::COPY),
CopyInfos[i].first)
.addReg(CopyInfos[i].second));
}
} else {
// No PHIs to worry about, just splice the instructions over.
PrevBB->splice(PrevBB->end(), TailBB, TailBB->begin(), TailBB->end());
}
PrevBB->removeSuccessor(PrevBB->succ_begin());
assert(PrevBB->succ_empty());
PrevBB->transferSuccessors(TailBB);
TDBBs.push_back(PrevBB);
Changed = true;
}
// If this is after register allocation, there are no phis to fix.
if (!PreRegAlloc)
return Changed;
// If we made no changes so far, we are safe.
if (!Changed)
return Changed;
// Handle the nasty case in that we duplicated a block that is part of a loop
// into some but not all of its predecessors. For example:
// 1 -> 2 <-> 3 |
// \ |
// \---> rest |
// if we duplicate 2 into 1 but not into 3, we end up with
// 12 -> 3 <-> 2 -> rest |
// \ / |
// \----->-----/ |
// If there was a "var = phi(1, 3)" in 2, it has to be ultimately replaced
// with a phi in 3 (which now dominates 2).
// What we do here is introduce a copy in 3 of the register defined by the
// phi, just like when we are duplicating 2 into 3, but we don't copy any
// real instructions or remove the 3 -> 2 edge from the phi in 2.
for (SmallSetVector<MachineBasicBlock *, 8>::iterator PI = Preds.begin(),
PE = Preds.end(); PI != PE; ++PI) {
MachineBasicBlock *PredBB = *PI;
if (std::find(TDBBs.begin(), TDBBs.end(), PredBB) != TDBBs.end())
continue;
// EH edges
if (PredBB->succ_size() != 1)
continue;
DenseMap<unsigned, unsigned> LocalVRMap;
SmallVector<std::pair<unsigned,unsigned>, 4> CopyInfos;
MachineBasicBlock::iterator I = TailBB->begin();
// Process PHI instructions first.
while (I != TailBB->end() && I->isPHI()) {
// Replace the uses of the def of the PHI with the register coming
// from PredBB.
MachineInstr *MI = &*I++;
ProcessPHI(MI, TailBB, PredBB, LocalVRMap, CopyInfos, UsedByPhi, false);
}
MachineBasicBlock::iterator Loc = PredBB->getFirstTerminator();
for (unsigned i = 0, e = CopyInfos.size(); i != e; ++i) {
Copies.push_back(BuildMI(*PredBB, Loc, DebugLoc(),
TII->get(TargetOpcode::COPY),
CopyInfos[i].first).addReg(CopyInfos[i].second));
}
}
return Changed;
}
void Diagnostic::
FormatDiagnostic(const char *DiagStr, const char *DiagEnd,
SmallVectorImpl<char> &OutStr) const {
/// FormattedArgs - Keep track of all of the arguments formatted by
/// ConvertArgToString and pass them into subsequent calls to
/// ConvertArgToString, allowing the implementation to avoid redundancies in
/// obvious cases.
SmallVector<DiagnosticsEngine::ArgumentValue, 8> FormattedArgs;
/// QualTypeVals - Pass a vector of arrays so that QualType names can be
/// compared to see if more information is needed to be printed.
SmallVector<intptr_t, 2> QualTypeVals;
SmallVector<char, 64> Tree;
for (unsigned i = 0, e = getNumArgs(); i < e; ++i)
if (getArgKind(i) == DiagnosticsEngine::ak_qualtype)
QualTypeVals.push_back(getRawArg(i));
while (DiagStr != DiagEnd) {
if (DiagStr[0] != '%') {
// Append non-%0 substrings to Str if we have one.
const char *StrEnd = std::find(DiagStr, DiagEnd, '%');
OutStr.append(DiagStr, StrEnd);
DiagStr = StrEnd;
continue;
} else if (ispunct(DiagStr[1])) {
OutStr.push_back(DiagStr[1]); // %% -> %.
DiagStr += 2;
continue;
}
// Skip the %.
++DiagStr;
// This must be a placeholder for a diagnostic argument. The format for a
// placeholder is one of "%0", "%modifier0", or "%modifier{arguments}0".
// The digit is a number from 0-9 indicating which argument this comes from.
// The modifier is a string of digits from the set [-a-z]+, arguments is a
// brace enclosed string.
const char *Modifier = 0, *Argument = 0;
unsigned ModifierLen = 0, ArgumentLen = 0;
// Check to see if we have a modifier. If so eat it.
if (!isdigit(DiagStr[0])) {
Modifier = DiagStr;
while (DiagStr[0] == '-' ||
(DiagStr[0] >= 'a' && DiagStr[0] <= 'z'))
++DiagStr;
ModifierLen = DiagStr-Modifier;
// If we have an argument, get it next.
if (DiagStr[0] == '{') {
++DiagStr; // Skip {.
Argument = DiagStr;
DiagStr = ScanFormat(DiagStr, DiagEnd, '}');
assert(DiagStr != DiagEnd && "Mismatched {}'s in diagnostic string!");
ArgumentLen = DiagStr-Argument;
++DiagStr; // Skip }.
}
}
assert(isdigit(*DiagStr) && "Invalid format for argument in diagnostic");
unsigned ArgNo = *DiagStr++ - '0';
// Only used for type diffing.
unsigned ArgNo2 = ArgNo;
DiagnosticsEngine::ArgumentKind Kind = getArgKind(ArgNo);
if (Kind == DiagnosticsEngine::ak_qualtype &&
ModifierIs(Modifier, ModifierLen, "diff")) {
Kind = DiagnosticsEngine::ak_qualtype_pair;
assert(*DiagStr == ',' && isdigit(*(DiagStr + 1)) &&
"Invalid format for diff modifier");
++DiagStr; // Comma.
ArgNo2 = *DiagStr++ - '0';
assert(getArgKind(ArgNo2) == DiagnosticsEngine::ak_qualtype &&
"Second value of type diff must be a qualtype");
}
switch (Kind) {
// ---- STRINGS ----
case DiagnosticsEngine::ak_std_string: {
const std::string &S = getArgStdStr(ArgNo);
assert(ModifierLen == 0 && "No modifiers for strings yet");
OutStr.append(S.begin(), S.end());
break;
}
case DiagnosticsEngine::ak_c_string: {
const char *S = getArgCStr(ArgNo);
assert(ModifierLen == 0 && "No modifiers for strings yet");
// Don't crash if get passed a null pointer by accident.
if (!S)
S = "(null)";
OutStr.append(S, S + strlen(S));
break;
}
// ---- INTEGERS ----
case DiagnosticsEngine::ak_sint: {
int Val = getArgSInt(ArgNo);
if (ModifierIs(Modifier, ModifierLen, "select")) {
HandleSelectModifier(*this, (unsigned)Val, Argument, ArgumentLen,
OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "s")) {
HandleIntegerSModifier(Val, OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "plural")) {
HandlePluralModifier(*this, (unsigned)Val, Argument, ArgumentLen,
OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "ordinal")) {
HandleOrdinalModifier((unsigned)Val, OutStr);
} else {
assert(ModifierLen == 0 && "Unknown integer modifier");
llvm::raw_svector_ostream(OutStr) << Val;
}
break;
}
case DiagnosticsEngine::ak_uint: {
unsigned Val = getArgUInt(ArgNo);
if (ModifierIs(Modifier, ModifierLen, "select")) {
HandleSelectModifier(*this, Val, Argument, ArgumentLen, OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "s")) {
HandleIntegerSModifier(Val, OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "plural")) {
HandlePluralModifier(*this, (unsigned)Val, Argument, ArgumentLen,
OutStr);
} else if (ModifierIs(Modifier, ModifierLen, "ordinal")) {
HandleOrdinalModifier(Val, OutStr);
} else {
assert(ModifierLen == 0 && "Unknown integer modifier");
llvm::raw_svector_ostream(OutStr) << Val;
}
break;
}
// ---- NAMES and TYPES ----
case DiagnosticsEngine::ak_identifierinfo: {
const IdentifierInfo *II = getArgIdentifier(ArgNo);
assert(ModifierLen == 0 && "No modifiers for strings yet");
// Don't crash if get passed a null pointer by accident.
if (!II) {
const char *S = "(null)";
OutStr.append(S, S + strlen(S));
continue;
}
llvm::raw_svector_ostream(OutStr) << '\'' << II->getName() << '\'';
break;
}
case DiagnosticsEngine::ak_qualtype:
case DiagnosticsEngine::ak_declarationname:
case DiagnosticsEngine::ak_nameddecl:
case DiagnosticsEngine::ak_nestednamespec:
case DiagnosticsEngine::ak_declcontext:
getDiags()->ConvertArgToString(Kind, getRawArg(ArgNo),
Modifier, ModifierLen,
Argument, ArgumentLen,
FormattedArgs.data(), FormattedArgs.size(),
OutStr, QualTypeVals);
break;
case DiagnosticsEngine::ak_qualtype_pair:
// Create a struct with all the info needed for printing.
TemplateDiffTypes TDT;
TDT.FromType = getRawArg(ArgNo);
TDT.ToType = getRawArg(ArgNo2);
TDT.ElideType = getDiags()->ElideType;
TDT.ShowColors = getDiags()->ShowColors;
intptr_t val = reinterpret_cast<intptr_t>(&TDT);
const char *ArgumentEnd = Argument + ArgumentLen;
const char *Pipe = ScanFormat(Argument, ArgumentEnd, '|');
// Print the tree.
if (getDiags()->PrintTemplateTree) {
TDT.PrintFromType = true;
TDT.PrintTree = true;
getDiags()->ConvertArgToString(Kind, val,
Modifier, ModifierLen,
Argument, ArgumentLen,
FormattedArgs.data(),
FormattedArgs.size(),
Tree, QualTypeVals);
// If there is no tree information, fall back to regular printing.
if (!Tree.empty()) {
FormatDiagnostic(Pipe + 1, ArgumentEnd, OutStr);
break;
}
}
// Non-tree printing, also the fall-back when tree printing fails.
// The fall-back is triggered when the types compared are not templates.
const char *FirstDollar = ScanFormat(Argument, ArgumentEnd, '$');
const char *SecondDollar = ScanFormat(FirstDollar + 1, ArgumentEnd, '$');
// Append before text
FormatDiagnostic(Argument, FirstDollar, OutStr);
// Append first type
TDT.PrintTree = false;
TDT.PrintFromType = true;
getDiags()->ConvertArgToString(Kind, val,
Modifier, ModifierLen,
Argument, ArgumentLen,
FormattedArgs.data(), FormattedArgs.size(),
OutStr, QualTypeVals);
// Append middle text
FormatDiagnostic(FirstDollar + 1, SecondDollar, OutStr);
// Append second type
TDT.PrintFromType = false;
getDiags()->ConvertArgToString(Kind, val,
Modifier, ModifierLen,
Argument, ArgumentLen,
FormattedArgs.data(), FormattedArgs.size(),
OutStr, QualTypeVals);
// Append end text
FormatDiagnostic(SecondDollar + 1, Pipe, OutStr);
break;
}
// Remember this argument info for subsequent formatting operations. Turn
// std::strings into a null terminated string to make it be the same case as
// all the other ones.
if (Kind == DiagnosticsEngine::ak_qualtype_pair)
continue;
else if (Kind != DiagnosticsEngine::ak_std_string)
FormattedArgs.push_back(std::make_pair(Kind, getRawArg(ArgNo)));
else
FormattedArgs.push_back(std::make_pair(DiagnosticsEngine::ak_c_string,
(intptr_t)getArgStdStr(ArgNo).c_str()));
}
// Append the type tree to the end of the diagnostics.
OutStr.append(Tree.begin(), Tree.end());
}