//
// The LLVM Compiler Infrastructure
//
-// This file was developed by the LLVM research group and is distributed under
-// the University of Illinois Open Source License. See LICENSE.TXT for details.
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//
//===----------------------------------------------------------------------===//
-#include "llvm/Transforms/Utils/FunctionUtils.h"
-#include "llvm/Constants.h"
-#include "llvm/DerivedTypes.h"
-#include "llvm/Instructions.h"
-#include "llvm/Intrinsics.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
-#include "llvm/Analysis/Dominators.h"
+#include "llvm/Transforms/Utils/CodeExtractor.h"
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SetVector.h"
+#include "llvm/ADT/StringExtras.h"
#include "llvm/Analysis/LoopInfo.h"
-#include "llvm/Analysis/Verifier.h"
-#include "llvm/Transforms/Utils/BasicBlockUtils.h"
+#include "llvm/Analysis/RegionInfo.h"
+#include "llvm/Analysis/RegionIterator.h"
+#include "llvm/IR/Constants.h"
+#include "llvm/IR/DerivedTypes.h"
+#include "llvm/IR/Dominators.h"
+#include "llvm/IR/Instructions.h"
+#include "llvm/IR/Intrinsics.h"
+#include "llvm/IR/LLVMContext.h"
+#include "llvm/IR/Module.h"
+#include "llvm/IR/Verifier.h"
+#include "llvm/Pass.h"
#include "llvm/Support/CommandLine.h"
-#include "llvm/Support/Compiler.h"
#include "llvm/Support/Debug.h"
-#include "llvm/ADT/StringExtras.h"
+#include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/raw_ostream.h"
+#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include <algorithm>
#include <set>
using namespace llvm;
+#define DEBUG_TYPE "code-extractor"
+
// Provide a command-line option to aggregate function arguments into a struct
-// for functions produced by the code extrator. This is useful when converting
+// for functions produced by the code extractor. This is useful when converting
// extracted functions to pthread-based code, as only one argument (void*) can
// be passed in to pthread_create().
static cl::opt<bool>
AggregateArgsOpt("aggregate-extracted-args", cl::Hidden,
cl::desc("Aggregate arguments to code-extracted functions"));
-namespace {
- class VISIBILITY_HIDDEN CodeExtractor {
- typedef std::vector<Value*> Values;
- std::set<BasicBlock*> BlocksToExtract;
- DominatorSet *DS;
- bool AggregateArgs;
- unsigned NumExitBlocks;
- const Type *RetTy;
- public:
- CodeExtractor(DominatorSet *ds = 0, bool AggArgs = false)
- : DS(ds), AggregateArgs(AggArgs||AggregateArgsOpt), NumExitBlocks(~0U) {}
-
- Function *ExtractCodeRegion(const std::vector<BasicBlock*> &code);
-
- bool isEligible(const std::vector<BasicBlock*> &code);
-
- private:
- /// definedInRegion - Return true if the specified value is defined in the
- /// extracted region.
- bool definedInRegion(Value *V) const {
- if (Instruction *I = dyn_cast<Instruction>(V))
- if (BlocksToExtract.count(I->getParent()))
- return true;
- return false;
- }
+/// \brief Test whether a block is valid for extraction.
+static bool isBlockValidForExtraction(const BasicBlock &BB) {
+ // Landing pads must be in the function where they were inserted for cleanup.
+ if (BB.isLandingPad())
+ return false;
- /// definedInCaller - Return true if the specified value is defined in the
- /// function being code extracted, but not in the region being extracted.
- /// These values must be passed in as live-ins to the function.
- bool definedInCaller(Value *V) const {
- if (isa<Argument>(V)) return true;
- if (Instruction *I = dyn_cast<Instruction>(V))
- if (!BlocksToExtract.count(I->getParent()))
- return true;
+ // Don't hoist code containing allocas, invokes, or vastarts.
+ for (BasicBlock::const_iterator I = BB.begin(), E = BB.end(); I != E; ++I) {
+ if (isa<AllocaInst>(I) || isa<InvokeInst>(I))
return false;
+ if (const CallInst *CI = dyn_cast<CallInst>(I))
+ if (const Function *F = CI->getCalledFunction())
+ if (F->getIntrinsicID() == Intrinsic::vastart)
+ return false;
+ }
+
+ return true;
+}
+
+/// \brief Build a set of blocks to extract if the input blocks are viable.
+template <typename IteratorT>
+static SetVector<BasicBlock *> buildExtractionBlockSet(IteratorT BBBegin,
+ IteratorT BBEnd) {
+ SetVector<BasicBlock *> Result;
+
+ assert(BBBegin != BBEnd);
+
+ // Loop over the blocks, adding them to our set-vector, and aborting with an
+ // empty set if we encounter invalid blocks.
+ for (IteratorT I = BBBegin, E = BBEnd; I != E; ++I) {
+ if (!Result.insert(*I))
+ llvm_unreachable("Repeated basic blocks in extraction input");
+
+ if (!isBlockValidForExtraction(**I)) {
+ Result.clear();
+ return Result;
}
+ }
- void severSplitPHINodes(BasicBlock *&Header);
- void splitReturnBlocks();
- void findInputsOutputs(Values &inputs, Values &outputs);
+#ifndef NDEBUG
+ for (SetVector<BasicBlock *>::iterator I = std::next(Result.begin()),
+ E = Result.end();
+ I != E; ++I)
+ for (pred_iterator PI = pred_begin(*I), PE = pred_end(*I);
+ PI != PE; ++PI)
+ assert(Result.count(*PI) &&
+ "No blocks in this region may have entries from outside the region"
+ " except for the first block!");
+#endif
- Function *constructFunction(const Values &inputs,
- const Values &outputs,
- BasicBlock *header,
- BasicBlock *newRootNode, BasicBlock *newHeader,
- Function *oldFunction, Module *M);
+ return Result;
+}
+
+/// \brief Helper to call buildExtractionBlockSet with an ArrayRef.
+static SetVector<BasicBlock *>
+buildExtractionBlockSet(ArrayRef<BasicBlock *> BBs) {
+ return buildExtractionBlockSet(BBs.begin(), BBs.end());
+}
+
+/// \brief Helper to call buildExtractionBlockSet with a RegionNode.
+static SetVector<BasicBlock *>
+buildExtractionBlockSet(const RegionNode &RN) {
+ if (!RN.isSubRegion())
+ // Just a single BasicBlock.
+ return buildExtractionBlockSet(RN.getNodeAs<BasicBlock>());
+
+ const Region &R = *RN.getNodeAs<Region>();
- void moveCodeToFunction(Function *newFunction);
+ return buildExtractionBlockSet(R.block_begin(), R.block_end());
+}
- void emitCallAndSwitchStatement(Function *newFunction,
- BasicBlock *newHeader,
- Values &inputs,
- Values &outputs);
+CodeExtractor::CodeExtractor(BasicBlock *BB, bool AggregateArgs)
+ : DT(nullptr), AggregateArgs(AggregateArgs||AggregateArgsOpt),
+ Blocks(buildExtractionBlockSet(BB)), NumExitBlocks(~0U) {}
+
+CodeExtractor::CodeExtractor(ArrayRef<BasicBlock *> BBs, DominatorTree *DT,
+ bool AggregateArgs)
+ : DT(DT), AggregateArgs(AggregateArgs||AggregateArgsOpt),
+ Blocks(buildExtractionBlockSet(BBs)), NumExitBlocks(~0U) {}
+
+CodeExtractor::CodeExtractor(DominatorTree &DT, Loop &L, bool AggregateArgs)
+ : DT(&DT), AggregateArgs(AggregateArgs||AggregateArgsOpt),
+ Blocks(buildExtractionBlockSet(L.getBlocks())), NumExitBlocks(~0U) {}
+
+CodeExtractor::CodeExtractor(DominatorTree &DT, const RegionNode &RN,
+ bool AggregateArgs)
+ : DT(&DT), AggregateArgs(AggregateArgs||AggregateArgsOpt),
+ Blocks(buildExtractionBlockSet(RN)), NumExitBlocks(~0U) {}
+
+/// definedInRegion - Return true if the specified value is defined in the
+/// extracted region.
+static bool definedInRegion(const SetVector<BasicBlock *> &Blocks, Value *V) {
+ if (Instruction *I = dyn_cast<Instruction>(V))
+ if (Blocks.count(I->getParent()))
+ return true;
+ return false;
+}
- };
+/// definedInCaller - Return true if the specified value is defined in the
+/// function being code extracted, but not in the region being extracted.
+/// These values must be passed in as live-ins to the function.
+static bool definedInCaller(const SetVector<BasicBlock *> &Blocks, Value *V) {
+ if (isa<Argument>(V)) return true;
+ if (Instruction *I = dyn_cast<Instruction>(V))
+ if (!Blocks.count(I->getParent()))
+ return true;
+ return false;
+}
+
+void CodeExtractor::findInputsOutputs(ValueSet &Inputs,
+ ValueSet &Outputs) const {
+ for (SetVector<BasicBlock *>::const_iterator I = Blocks.begin(),
+ E = Blocks.end();
+ I != E; ++I) {
+ BasicBlock *BB = *I;
+
+ // If a used value is defined outside the region, it's an input. If an
+ // instruction is used outside the region, it's an output.
+ for (BasicBlock::iterator II = BB->begin(), IE = BB->end();
+ II != IE; ++II) {
+ for (User::op_iterator OI = II->op_begin(), OE = II->op_end();
+ OI != OE; ++OI)
+ if (definedInCaller(Blocks, *OI))
+ Inputs.insert(*OI);
+
+ for (User *U : II->users())
+ if (!definedInRegion(Blocks, U)) {
+ Outputs.insert(II);
+ break;
+ }
+ }
+ }
}
/// severSplitPHINodes - If a PHI node has multiple inputs from outside of the
/// region, we need to split the entry block of the region so that the PHI node
/// is easier to deal with.
void CodeExtractor::severSplitPHINodes(BasicBlock *&Header) {
- bool HasPredsFromRegion = false;
+ unsigned NumPredsFromRegion = 0;
unsigned NumPredsOutsideRegion = 0;
if (Header != &Header->getParent()->getEntryBlock()) {
// than one entry from outside the region. If so, we need to sever the
// header block into two.
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
- if (BlocksToExtract.count(PN->getIncomingBlock(i)))
- HasPredsFromRegion = true;
+ if (Blocks.count(PN->getIncomingBlock(i)))
+ ++NumPredsFromRegion;
else
++NumPredsOutsideRegion;
// containing PHI nodes merging values from outside of the region, and a
// second that contains all of the code for the block and merges back any
// incoming values from inside of the region.
- BasicBlock::iterator AfterPHIs = Header->begin();
- while (isa<PHINode>(AfterPHIs)) ++AfterPHIs;
+ BasicBlock::iterator AfterPHIs = Header->getFirstNonPHI();
BasicBlock *NewBB = Header->splitBasicBlock(AfterPHIs,
Header->getName()+".ce");
// We only want to code extract the second block now, and it becomes the new
// header of the region.
BasicBlock *OldPred = Header;
- BlocksToExtract.erase(OldPred);
- BlocksToExtract.insert(NewBB);
+ Blocks.remove(OldPred);
+ Blocks.insert(NewBB);
Header = NewBB;
// Okay, update dominator sets. The blocks that dominate the new one are the
// blocks that dominate TIBB plus the new block itself.
- if (DS) {
- DominatorSet::DomSetType DomSet = DS->getDominators(OldPred);
- DomSet.insert(NewBB); // A block always dominates itself.
- DS->addBasicBlock(NewBB, DomSet);
-
- // Additionally, NewBB dominates all blocks in the function that are
- // dominated by OldPred.
- Function *F = Header->getParent();
- for (Function::iterator I = F->begin(), E = F->end(); I != E; ++I)
- if (DS->properlyDominates(OldPred, I))
- DS->addDominator(I, NewBB);
- }
+ if (DT)
+ DT->splitBlock(NewBB);
// Okay, now we need to adjust the PHI nodes and any branches from within the
// region to go to the new header block instead of the old header block.
- if (HasPredsFromRegion) {
+ if (NumPredsFromRegion) {
PHINode *PN = cast<PHINode>(OldPred->begin());
// Loop over all of the predecessors of OldPred that are in the region,
// changing them to branch to NewBB instead.
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
- if (BlocksToExtract.count(PN->getIncomingBlock(i))) {
+ if (Blocks.count(PN->getIncomingBlock(i))) {
TerminatorInst *TI = PN->getIncomingBlock(i)->getTerminator();
TI->replaceUsesOfWith(OldPred, NewBB);
}
- // Okay, everthing within the region is now branching to the right block, we
+ // Okay, everything within the region is now branching to the right block, we
// just have to update the PHI nodes now, inserting PHI nodes into NewBB.
for (AfterPHIs = OldPred->begin(); isa<PHINode>(AfterPHIs); ++AfterPHIs) {
PHINode *PN = cast<PHINode>(AfterPHIs);
// Create a new PHI node in the new region, which has an incoming value
// from OldPred of PN.
- PHINode *NewPN = new PHINode(PN->getType(), PN->getName()+".ce",
- NewBB->begin());
+ PHINode *NewPN = PHINode::Create(PN->getType(), 1 + NumPredsFromRegion,
+ PN->getName()+".ce", NewBB->begin());
NewPN->addIncoming(PN, OldPred);
// Loop over all of the incoming value in PN, moving them to NewPN if they
// are from the extracted region.
for (unsigned i = 0; i != PN->getNumIncomingValues(); ++i) {
- if (BlocksToExtract.count(PN->getIncomingBlock(i))) {
+ if (Blocks.count(PN->getIncomingBlock(i))) {
NewPN->addIncoming(PN->getIncomingValue(i), PN->getIncomingBlock(i));
PN->removeIncomingValue(i);
--i;
}
void CodeExtractor::splitReturnBlocks() {
- for (std::set<BasicBlock*>::iterator I = BlocksToExtract.begin(),
- E = BlocksToExtract.end(); I != E; ++I)
- if (ReturnInst *RI = dyn_cast<ReturnInst>((*I)->getTerminator()))
- (*I)->splitBasicBlock(RI, (*I)->getName()+".ret");
-}
-
-// findInputsOutputs - Find inputs to, outputs from the code region.
-//
-void CodeExtractor::findInputsOutputs(Values &inputs, Values &outputs) {
- std::set<BasicBlock*> ExitBlocks;
- for (std::set<BasicBlock*>::const_iterator ci = BlocksToExtract.begin(),
- ce = BlocksToExtract.end(); ci != ce; ++ci) {
- BasicBlock *BB = *ci;
-
- for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I) {
- // If a used value is defined outside the region, it's an input. If an
- // instruction is used outside the region, it's an output.
- for (User::op_iterator O = I->op_begin(), E = I->op_end(); O != E; ++O)
- if (definedInCaller(*O))
- inputs.push_back(*O);
-
- // Consider uses of this instruction (outputs).
- for (Value::use_iterator UI = I->use_begin(), E = I->use_end();
- UI != E; ++UI)
- if (!definedInRegion(*UI)) {
- outputs.push_back(I);
- break;
- }
- } // for: insts
-
- // Keep track of the exit blocks from the region.
- TerminatorInst *TI = BB->getTerminator();
- for (unsigned i = 0, e = TI->getNumSuccessors(); i != e; ++i)
- if (!BlocksToExtract.count(TI->getSuccessor(i)))
- ExitBlocks.insert(TI->getSuccessor(i));
- } // for: basic blocks
-
- NumExitBlocks = ExitBlocks.size();
-
- // Eliminate duplicates.
- std::sort(inputs.begin(), inputs.end());
- inputs.erase(std::unique(inputs.begin(), inputs.end()), inputs.end());
- std::sort(outputs.begin(), outputs.end());
- outputs.erase(std::unique(outputs.begin(), outputs.end()), outputs.end());
+ for (SetVector<BasicBlock *>::iterator I = Blocks.begin(), E = Blocks.end();
+ I != E; ++I)
+ if (ReturnInst *RI = dyn_cast<ReturnInst>((*I)->getTerminator())) {
+ BasicBlock *New = (*I)->splitBasicBlock(RI, (*I)->getName()+".ret");
+ if (DT) {
+ // Old dominates New. New node dominates all other nodes dominated
+ // by Old.
+ DomTreeNode *OldNode = DT->getNode(*I);
+ SmallVector<DomTreeNode*, 8> Children;
+ for (DomTreeNode::iterator DI = OldNode->begin(), DE = OldNode->end();
+ DI != DE; ++DI)
+ Children.push_back(*DI);
+
+ DomTreeNode *NewNode = DT->addNewBlock(New, *I);
+
+ for (SmallVectorImpl<DomTreeNode *>::iterator I = Children.begin(),
+ E = Children.end(); I != E; ++I)
+ DT->changeImmediateDominator(*I, NewNode);
+ }
+ }
}
/// constructFunction - make a function based on inputs and outputs, as follows:
/// f(in0, ..., inN, out0, ..., outN)
///
-Function *CodeExtractor::constructFunction(const Values &inputs,
- const Values &outputs,
+Function *CodeExtractor::constructFunction(const ValueSet &inputs,
+ const ValueSet &outputs,
BasicBlock *header,
BasicBlock *newRootNode,
BasicBlock *newHeader,
Function *oldFunction,
Module *M) {
- DOUT << "inputs: " << inputs.size() << "\n";
- DOUT << "outputs: " << outputs.size() << "\n";
+ DEBUG(dbgs() << "inputs: " << inputs.size() << "\n");
+ DEBUG(dbgs() << "outputs: " << outputs.size() << "\n");
// This function returns unsigned, outputs will go back by reference.
switch (NumExitBlocks) {
case 0:
- case 1: RetTy = Type::VoidTy; break;
- case 2: RetTy = Type::Int1Ty; break;
- default: RetTy = Type::Int16Ty; break;
+ case 1: RetTy = Type::getVoidTy(header->getContext()); break;
+ case 2: RetTy = Type::getInt1Ty(header->getContext()); break;
+ default: RetTy = Type::getInt16Ty(header->getContext()); break;
}
- std::vector<const Type*> paramTy;
+ std::vector<Type*> paramTy;
// Add the types of the input values to the function's argument list
- for (Values::const_iterator i = inputs.begin(),
- e = inputs.end(); i != e; ++i) {
+ for (ValueSet::const_iterator i = inputs.begin(), e = inputs.end();
+ i != e; ++i) {
const Value *value = *i;
- DOUT << "value used in func: " << *value << "\n";
+ DEBUG(dbgs() << "value used in func: " << *value << "\n");
paramTy.push_back(value->getType());
}
// Add the types of the output values to the function's argument list.
- for (Values::const_iterator I = outputs.begin(), E = outputs.end();
+ for (ValueSet::const_iterator I = outputs.begin(), E = outputs.end();
I != E; ++I) {
- DOUT << "instr used in func: " << **I << "\n";
+ DEBUG(dbgs() << "instr used in func: " << **I << "\n");
if (AggregateArgs)
paramTy.push_back((*I)->getType());
else
- paramTy.push_back(PointerType::get((*I)->getType()));
+ paramTy.push_back(PointerType::getUnqual((*I)->getType()));
}
- DOUT << "Function type: " << *RetTy << " f(";
- for (std::vector<const Type*>::iterator i = paramTy.begin(),
+ DEBUG(dbgs() << "Function type: " << *RetTy << " f(");
+ for (std::vector<Type*>::iterator i = paramTy.begin(),
e = paramTy.end(); i != e; ++i)
- DOUT << **i << ", ";
- DOUT << ")\n";
+ DEBUG(dbgs() << **i << ", ");
+ DEBUG(dbgs() << ")\n");
if (AggregateArgs && (inputs.size() + outputs.size() > 0)) {
- PointerType *StructPtr = PointerType::get(StructType::get(paramTy));
+ PointerType *StructPtr =
+ PointerType::getUnqual(StructType::get(M->getContext(), paramTy));
paramTy.clear();
paramTy.push_back(StructPtr);
}
- const FunctionType *funcType = FunctionType::get(RetTy, paramTy, false);
+ FunctionType *funcType =
+ FunctionType::get(RetTy, paramTy, false);
// Create the new function
- Function *newFunction = new Function(funcType,
- GlobalValue::InternalLinkage,
- oldFunction->getName() + "_" +
- header->getName(), M);
+ Function *newFunction = Function::Create(funcType,
+ GlobalValue::InternalLinkage,
+ oldFunction->getName() + "_" +
+ header->getName(), M);
+ // If the old function is no-throw, so is the new one.
+ if (oldFunction->doesNotThrow())
+ newFunction->setDoesNotThrow();
+
newFunction->getBasicBlockList().push_back(newRootNode);
// Create an iterator to name all of the arguments we inserted.
for (unsigned i = 0, e = inputs.size(); i != e; ++i) {
Value *RewriteVal;
if (AggregateArgs) {
- Value *Idx0 = Constant::getNullValue(Type::Int32Ty);
- Value *Idx1 = ConstantInt::get(Type::Int32Ty, i);
- std::string GEPname = "gep_" + inputs[i]->getName();
+ Value *Idx[2];
+ Idx[0] = Constant::getNullValue(Type::getInt32Ty(header->getContext()));
+ Idx[1] = ConstantInt::get(Type::getInt32Ty(header->getContext()), i);
TerminatorInst *TI = newFunction->begin()->getTerminator();
- GetElementPtrInst *GEP = new GetElementPtrInst(AI, Idx0, Idx1,
- GEPname, TI);
- RewriteVal = new LoadInst(GEP, "load" + GEPname, TI);
+ GetElementPtrInst *GEP =
+ GetElementPtrInst::Create(AI, Idx, "gep_" + inputs[i]->getName(), TI);
+ RewriteVal = new LoadInst(GEP, "loadgep_" + inputs[i]->getName(), TI);
} else
RewriteVal = AI++;
- std::vector<User*> Users(inputs[i]->use_begin(), inputs[i]->use_end());
+ std::vector<User*> Users(inputs[i]->user_begin(), inputs[i]->user_end());
for (std::vector<User*>::iterator use = Users.begin(), useE = Users.end();
use != useE; ++use)
if (Instruction* inst = dyn_cast<Instruction>(*use))
- if (BlocksToExtract.count(inst->getParent()))
+ if (Blocks.count(inst->getParent()))
inst->replaceUsesOfWith(inputs[i], RewriteVal);
}
// Rewrite branches to basic blocks outside of the loop to new dummy blocks
// within the new function. This must be done before we lose track of which
// blocks were originally in the code region.
- std::vector<User*> Users(header->use_begin(), header->use_end());
+ std::vector<User*> Users(header->user_begin(), header->user_end());
for (unsigned i = 0, e = Users.size(); i != e; ++i)
// The BasicBlock which contains the branch is not in the region
// modify the branch target to a new block
if (TerminatorInst *TI = dyn_cast<TerminatorInst>(Users[i]))
- if (!BlocksToExtract.count(TI->getParent()) &&
+ if (!Blocks.count(TI->getParent()) &&
TI->getParent()->getParent() == oldFunction)
TI->replaceUsesOfWith(header, newHeader);
return newFunction;
}
+/// FindPhiPredForUseInBlock - Given a value and a basic block, find a PHI
+/// that uses the value within the basic block, and return the predecessor
+/// block associated with that use, or return 0 if none is found.
+static BasicBlock* FindPhiPredForUseInBlock(Value* Used, BasicBlock* BB) {
+ for (Use &U : Used->uses()) {
+ PHINode *P = dyn_cast<PHINode>(U.getUser());
+ if (P && P->getParent() == BB)
+ return P->getIncomingBlock(U);
+ }
+
+ return nullptr;
+}
+
/// emitCallAndSwitchStatement - This method sets up the caller side by adding
/// the call instruction, splitting any PHI nodes in the header block as
/// necessary.
void CodeExtractor::
emitCallAndSwitchStatement(Function *newFunction, BasicBlock *codeReplacer,
- Values &inputs, Values &outputs) {
+ ValueSet &inputs, ValueSet &outputs) {
// Emit a call to the new function, passing in: *pointer to struct (if
// aggregating parameters), or plan inputs and allocated memory for outputs
- std::vector<Value*> params, StructValues, ReloadOutputs;
+ std::vector<Value*> params, StructValues, ReloadOutputs, Reloads;
+
+ LLVMContext &Context = newFunction->getContext();
// Add inputs as params, or to be filled into the struct
- for (Values::iterator i = inputs.begin(), e = inputs.end(); i != e; ++i)
+ for (ValueSet::iterator i = inputs.begin(), e = inputs.end(); i != e; ++i)
if (AggregateArgs)
StructValues.push_back(*i);
else
params.push_back(*i);
// Create allocas for the outputs
- for (Values::iterator i = outputs.begin(), e = outputs.end(); i != e; ++i) {
+ for (ValueSet::iterator i = outputs.begin(), e = outputs.end(); i != e; ++i) {
if (AggregateArgs) {
StructValues.push_back(*i);
} else {
AllocaInst *alloca =
- new AllocaInst((*i)->getType(), 0, (*i)->getName()+".loc",
+ new AllocaInst((*i)->getType(), nullptr, (*i)->getName()+".loc",
codeReplacer->getParent()->begin()->begin());
ReloadOutputs.push_back(alloca);
params.push_back(alloca);
}
}
- AllocaInst *Struct = 0;
+ AllocaInst *Struct = nullptr;
if (AggregateArgs && (inputs.size() + outputs.size() > 0)) {
- std::vector<const Type*> ArgTypes;
- for (Values::iterator v = StructValues.begin(),
+ std::vector<Type*> ArgTypes;
+ for (ValueSet::iterator v = StructValues.begin(),
ve = StructValues.end(); v != ve; ++v)
ArgTypes.push_back((*v)->getType());
// Allocate a struct at the beginning of this function
- Type *StructArgTy = StructType::get(ArgTypes);
+ Type *StructArgTy = StructType::get(newFunction->getContext(), ArgTypes);
Struct =
- new AllocaInst(StructArgTy, 0, "structArg",
+ new AllocaInst(StructArgTy, nullptr, "structArg",
codeReplacer->getParent()->begin()->begin());
params.push_back(Struct);
for (unsigned i = 0, e = inputs.size(); i != e; ++i) {
- Value *Idx0 = Constant::getNullValue(Type::Int32Ty);
- Value *Idx1 = ConstantInt::get(Type::Int32Ty, i);
+ Value *Idx[2];
+ Idx[0] = Constant::getNullValue(Type::getInt32Ty(Context));
+ Idx[1] = ConstantInt::get(Type::getInt32Ty(Context), i);
GetElementPtrInst *GEP =
- new GetElementPtrInst(Struct, Idx0, Idx1,
- "gep_" + StructValues[i]->getName());
+ GetElementPtrInst::Create(Struct, Idx,
+ "gep_" + StructValues[i]->getName());
codeReplacer->getInstList().push_back(GEP);
StoreInst *SI = new StoreInst(StructValues[i], GEP);
codeReplacer->getInstList().push_back(SI);
}
// Emit the call to the function
- CallInst *call = new CallInst(newFunction, ¶ms[0], params.size(),
- NumExitBlocks > 1 ? "targetBlock" : "");
+ CallInst *call = CallInst::Create(newFunction, params,
+ NumExitBlocks > 1 ? "targetBlock" : "");
codeReplacer->getInstList().push_back(call);
Function::arg_iterator OutputArgBegin = newFunction->arg_begin();
// Reload the outputs passed in by reference
for (unsigned i = 0, e = outputs.size(); i != e; ++i) {
- Value *Output = 0;
+ Value *Output = nullptr;
if (AggregateArgs) {
- Value *Idx0 = Constant::getNullValue(Type::Int32Ty);
- Value *Idx1 = ConstantInt::get(Type::Int32Ty, FirstOut + i);
+ Value *Idx[2];
+ Idx[0] = Constant::getNullValue(Type::getInt32Ty(Context));
+ Idx[1] = ConstantInt::get(Type::getInt32Ty(Context), FirstOut + i);
GetElementPtrInst *GEP
- = new GetElementPtrInst(Struct, Idx0, Idx1,
- "gep_reload_" + outputs[i]->getName());
+ = GetElementPtrInst::Create(Struct, Idx,
+ "gep_reload_" + outputs[i]->getName());
codeReplacer->getInstList().push_back(GEP);
Output = GEP;
} else {
Output = ReloadOutputs[i];
}
LoadInst *load = new LoadInst(Output, outputs[i]->getName()+".reload");
+ Reloads.push_back(load);
codeReplacer->getInstList().push_back(load);
- std::vector<User*> Users(outputs[i]->use_begin(), outputs[i]->use_end());
+ std::vector<User*> Users(outputs[i]->user_begin(), outputs[i]->user_end());
for (unsigned u = 0, e = Users.size(); u != e; ++u) {
Instruction *inst = cast<Instruction>(Users[u]);
- if (!BlocksToExtract.count(inst->getParent()))
+ if (!Blocks.count(inst->getParent()))
inst->replaceUsesOfWith(outputs[i], load);
}
}
// Now we can emit a switch statement using the call as a value.
SwitchInst *TheSwitch =
- new SwitchInst(ConstantInt::getNullValue(Type::Int16Ty),
- codeReplacer, 0, codeReplacer);
+ SwitchInst::Create(Constant::getNullValue(Type::getInt16Ty(Context)),
+ codeReplacer, 0, codeReplacer);
// Since there may be multiple exits from the original region, make the new
// function return an unsigned, switch on that number. This loop iterates
std::map<BasicBlock*, BasicBlock*> ExitBlockMap;
unsigned switchVal = 0;
- for (std::set<BasicBlock*>::const_iterator i = BlocksToExtract.begin(),
- e = BlocksToExtract.end(); i != e; ++i) {
+ for (SetVector<BasicBlock*>::const_iterator i = Blocks.begin(),
+ e = Blocks.end(); i != e; ++i) {
TerminatorInst *TI = (*i)->getTerminator();
for (unsigned i = 0, e = TI->getNumSuccessors(); i != e; ++i)
- if (!BlocksToExtract.count(TI->getSuccessor(i))) {
+ if (!Blocks.count(TI->getSuccessor(i))) {
BasicBlock *OldTarget = TI->getSuccessor(i);
// add a new basic block which returns the appropriate value
BasicBlock *&NewTarget = ExitBlockMap[OldTarget];
if (!NewTarget) {
// If we don't already have an exit stub for this non-extracted
// destination, create one now!
- NewTarget = new BasicBlock(OldTarget->getName() + ".exitStub",
- newFunction);
+ NewTarget = BasicBlock::Create(Context,
+ OldTarget->getName() + ".exitStub",
+ newFunction);
unsigned SuccNum = switchVal++;
- Value *brVal = 0;
+ Value *brVal = nullptr;
switch (NumExitBlocks) {
case 0:
case 1: break; // No value needed.
case 2: // Conditional branch, return a bool
- brVal = ConstantInt::get(Type::Int1Ty, !SuccNum);
+ brVal = ConstantInt::get(Type::getInt1Ty(Context), !SuccNum);
break;
default:
- brVal = ConstantInt::get(Type::Int16Ty, SuccNum);
+ brVal = ConstantInt::get(Type::getInt16Ty(Context), SuccNum);
break;
}
- ReturnInst *NTRet = new ReturnInst(brVal, NewTarget);
+ ReturnInst *NTRet = ReturnInst::Create(Context, brVal, NewTarget);
// Update the switch instruction.
- TheSwitch->addCase(ConstantInt::get(Type::Int16Ty, SuccNum),
+ TheSwitch->addCase(ConstantInt::get(Type::getInt16Ty(Context),
+ SuccNum),
OldTarget);
// Restore values just before we exit
// In the extract block case, if the block we are extracting ends
// with an invoke instruction, make sure that we don't emit a
// store of the invoke value for the unwind block.
- if (!DS && DefBlock != OldTarget)
+ if (!DT && DefBlock != OldTarget)
DominatesDef = false;
}
- if (DS)
- DominatesDef = DS->dominates(DefBlock, OldTarget);
+ if (DT) {
+ DominatesDef = DT->dominates(DefBlock, OldTarget);
+
+ // If the output value is used by a phi in the target block,
+ // then we need to test for dominance of the phi's predecessor
+ // instead. Unfortunately, this a little complicated since we
+ // have already rewritten uses of the value to uses of the reload.
+ BasicBlock* pred = FindPhiPredForUseInBlock(Reloads[out],
+ OldTarget);
+ if (pred && DT && DT->dominates(DefBlock, pred))
+ DominatesDef = true;
+ }
if (DominatesDef) {
if (AggregateArgs) {
- Value *Idx0 = Constant::getNullValue(Type::Int32Ty);
- Value *Idx1 = ConstantInt::get(Type::Int32Ty,FirstOut+out);
+ Value *Idx[2];
+ Idx[0] = Constant::getNullValue(Type::getInt32Ty(Context));
+ Idx[1] = ConstantInt::get(Type::getInt32Ty(Context),
+ FirstOut+out);
GetElementPtrInst *GEP =
- new GetElementPtrInst(OAI, Idx0, Idx1,
- "gep_" + outputs[out]->getName(),
- NTRet);
+ GetElementPtrInst::Create(OAI, Idx,
+ "gep_" + outputs[out]->getName(),
+ NTRet);
new StoreInst(outputs[out], GEP, NTRet);
} else {
new StoreInst(outputs[out], OAI, NTRet);
}
// Now that we've done the deed, simplify the switch instruction.
- const Type *OldFnRetTy = TheSwitch->getParent()->getParent()->getReturnType();
+ Type *OldFnRetTy = TheSwitch->getParent()->getParent()->getReturnType();
switch (NumExitBlocks) {
case 0:
// There are no successors (the block containing the switch itself), which
// this should be rewritten as a `ret'
// Check if the function should return a value
- if (OldFnRetTy == Type::VoidTy) {
- new ReturnInst(0, TheSwitch); // Return void
+ if (OldFnRetTy->isVoidTy()) {
+ ReturnInst::Create(Context, nullptr, TheSwitch); // Return void
} else if (OldFnRetTy == TheSwitch->getCondition()->getType()) {
// return what we have
- new ReturnInst(TheSwitch->getCondition(), TheSwitch);
+ ReturnInst::Create(Context, TheSwitch->getCondition(), TheSwitch);
} else {
// Otherwise we must have code extracted an unwind or something, just
// return whatever we want.
- new ReturnInst(Constant::getNullValue(OldFnRetTy), TheSwitch);
+ ReturnInst::Create(Context,
+ Constant::getNullValue(OldFnRetTy), TheSwitch);
}
- TheSwitch->getParent()->getInstList().erase(TheSwitch);
+ TheSwitch->eraseFromParent();
break;
case 1:
// Only a single destination, change the switch into an unconditional
// branch.
- new BranchInst(TheSwitch->getSuccessor(1), TheSwitch);
- TheSwitch->getParent()->getInstList().erase(TheSwitch);
+ BranchInst::Create(TheSwitch->getSuccessor(1), TheSwitch);
+ TheSwitch->eraseFromParent();
break;
case 2:
- new BranchInst(TheSwitch->getSuccessor(1), TheSwitch->getSuccessor(2),
- call, TheSwitch);
- TheSwitch->getParent()->getInstList().erase(TheSwitch);
+ BranchInst::Create(TheSwitch->getSuccessor(1), TheSwitch->getSuccessor(2),
+ call, TheSwitch);
+ TheSwitch->eraseFromParent();
break;
default:
// Otherwise, make the default destination of the switch instruction be one
// of the other successors.
- TheSwitch->setOperand(0, call);
- TheSwitch->setSuccessor(0, TheSwitch->getSuccessor(NumExitBlocks));
- TheSwitch->removeCase(NumExitBlocks); // Remove redundant case
+ TheSwitch->setCondition(call);
+ TheSwitch->setDefaultDest(TheSwitch->getSuccessor(NumExitBlocks));
+ // Remove redundant case
+ TheSwitch->removeCase(SwitchInst::CaseIt(TheSwitch, NumExitBlocks-1));
break;
}
}
void CodeExtractor::moveCodeToFunction(Function *newFunction) {
- Function *oldFunc = (*BlocksToExtract.begin())->getParent();
+ Function *oldFunc = (*Blocks.begin())->getParent();
Function::BasicBlockListType &oldBlocks = oldFunc->getBasicBlockList();
Function::BasicBlockListType &newBlocks = newFunction->getBasicBlockList();
- for (std::set<BasicBlock*>::const_iterator i = BlocksToExtract.begin(),
- e = BlocksToExtract.end(); i != e; ++i) {
+ for (SetVector<BasicBlock*>::const_iterator i = Blocks.begin(),
+ e = Blocks.end(); i != e; ++i) {
// Delete the basic block from the old function, and the list of blocks
oldBlocks.remove(*i);
}
}
-/// ExtractRegion - Removes a loop from a function, replaces it with a call to
-/// new function. Returns pointer to the new function.
-///
-/// algorithm:
-///
-/// find inputs and outputs for the region
-///
-/// for inputs: add to function as args, map input instr* to arg#
-/// for outputs: add allocas for scalars,
-/// add to func as args, map output instr* to arg#
-///
-/// rewrite func to use argument #s instead of instr*
-///
-/// for each scalar output in the function: at every exit, store intermediate
-/// computed result back into memory.
-///
-Function *CodeExtractor::
-ExtractCodeRegion(const std::vector<BasicBlock*> &code) {
- if (!isEligible(code))
- return 0;
-
- // 1) Find inputs, outputs
- // 2) Construct new function
- // * Add allocas for defs, pass as args by reference
- // * Pass in uses as args
- // 3) Move code region, add call instr to func
- //
- BlocksToExtract.insert(code.begin(), code.end());
+Function *CodeExtractor::extractCodeRegion() {
+ if (!isEligible())
+ return nullptr;
- Values inputs, outputs;
+ ValueSet inputs, outputs;
// Assumption: this is a single-entry code region, and the header is the first
// block in the region.
- BasicBlock *header = code[0];
-
- for (unsigned i = 1, e = code.size(); i != e; ++i)
- for (pred_iterator PI = pred_begin(code[i]), E = pred_end(code[i]);
- PI != E; ++PI)
- assert(BlocksToExtract.count(*PI) &&
- "No blocks in this region may have entries from outside the region"
- " except for the first block!");
+ BasicBlock *header = *Blocks.begin();
// If we have to split PHI nodes or the entry block, do so now.
severSplitPHINodes(header);
Function *oldFunction = header->getParent();
// This takes place of the original loop
- BasicBlock *codeReplacer = new BasicBlock("codeRepl", oldFunction, header);
+ BasicBlock *codeReplacer = BasicBlock::Create(header->getContext(),
+ "codeRepl", oldFunction,
+ header);
// The new function needs a root node because other nodes can branch to the
// head of the region, but the entry node of a function cannot have preds.
- BasicBlock *newFuncRoot = new BasicBlock("newFuncRoot");
- newFuncRoot->getInstList().push_back(new BranchInst(header));
+ BasicBlock *newFuncRoot = BasicBlock::Create(header->getContext(),
+ "newFuncRoot");
+ newFuncRoot->getInstList().push_back(BranchInst::Create(header));
// Find inputs to, outputs from the code region.
findInputsOutputs(inputs, outputs);
+ SmallPtrSet<BasicBlock *, 1> ExitBlocks;
+ for (SetVector<BasicBlock *>::iterator I = Blocks.begin(), E = Blocks.end();
+ I != E; ++I)
+ for (succ_iterator SI = succ_begin(*I), SE = succ_end(*I); SI != SE; ++SI)
+ if (!Blocks.count(*SI))
+ ExitBlocks.insert(*SI);
+ NumExitBlocks = ExitBlocks.size();
+
// Construct new function based on inputs/outputs & add allocas for all defs.
Function *newFunction = constructFunction(inputs, outputs, header,
newFuncRoot,
for (BasicBlock::iterator I = header->begin(); isa<PHINode>(I); ++I) {
PHINode *PN = cast<PHINode>(I);
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
- if (!BlocksToExtract.count(PN->getIncomingBlock(i)))
+ if (!Blocks.count(PN->getIncomingBlock(i)))
PN->setIncomingBlock(i, newFuncRoot);
}
PHINode *PN = cast<PHINode>(I);
std::set<BasicBlock*> ProcessedPreds;
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
- if (BlocksToExtract.count(PN->getIncomingBlock(i)))
+ if (Blocks.count(PN->getIncomingBlock(i))) {
if (ProcessedPreds.insert(PN->getIncomingBlock(i)).second)
PN->setIncomingBlock(i, codeReplacer);
else {
PN->removeIncomingValue(i, false);
--i; --e;
}
+ }
}
//cerr << "NEW FUNCTION: " << *newFunction;
// cerr << "OLD FUNCTION: " << *oldFunction;
// verifyFunction(*oldFunction);
- DEBUG(if (verifyFunction(*newFunction)) abort());
+ DEBUG(if (verifyFunction(*newFunction))
+ report_fatal_error("verifyFunction failed!"));
return newFunction;
}
-
-bool CodeExtractor::isEligible(const std::vector<BasicBlock*> &code) {
- // Deny code region if it contains allocas or vastarts.
- for (std::vector<BasicBlock*>::const_iterator BB = code.begin(), e=code.end();
- BB != e; ++BB)
- for (BasicBlock::const_iterator I = (*BB)->begin(), Ie = (*BB)->end();
- I != Ie; ++I)
- if (isa<AllocaInst>(*I))
- return false;
- else if (const CallInst *CI = dyn_cast<CallInst>(I))
- if (const Function *F = CI->getCalledFunction())
- if (F->getIntrinsicID() == Intrinsic::vastart)
- return false;
- return true;
-}
-
-
-/// ExtractCodeRegion - slurp a sequence of basic blocks into a brand new
-/// function
-///
-Function* llvm::ExtractCodeRegion(DominatorSet &DS,
- const std::vector<BasicBlock*> &code,
- bool AggregateArgs) {
- return CodeExtractor(&DS, AggregateArgs).ExtractCodeRegion(code);
-}
-
-/// ExtractBasicBlock - slurp a natural loop into a brand new function
-///
-Function* llvm::ExtractLoop(DominatorSet &DS, Loop *L, bool AggregateArgs) {
- return CodeExtractor(&DS, AggregateArgs).ExtractCodeRegion(L->getBlocks());
-}
-
-/// ExtractBasicBlock - slurp a basic block into a brand new function
-///
-Function* llvm::ExtractBasicBlock(BasicBlock *BB, bool AggregateArgs) {
- std::vector<BasicBlock*> Blocks;
- Blocks.push_back(BB);
- return CodeExtractor(0, AggregateArgs).ExtractCodeRegion(Blocks);
-}