llvm-project/llvm/lib/Transforms/IPO/IROutliner.cpp
2023-03-15 18:06:32 -07:00

3047 lines
125 KiB
C++

//===- IROutliner.cpp -- Outline Similar Regions ----------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
///
/// \file
// Implementation for the IROutliner which is used by the IROutliner Pass.
//
//===----------------------------------------------------------------------===//
#include "llvm/Transforms/IPO/IROutliner.h"
#include "llvm/Analysis/IRSimilarityIdentifier.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/IR/Attributes.h"
#include "llvm/IR/DIBuilder.h"
#include "llvm/IR/DebugInfo.h"
#include "llvm/IR/DebugInfoMetadata.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Mangler.h"
#include "llvm/IR/PassManager.h"
#include "llvm/InitializePasses.h"
#include "llvm/Pass.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Transforms/IPO.h"
#include <optional>
#include <vector>
#define DEBUG_TYPE "iroutliner"
using namespace llvm;
using namespace IRSimilarity;
// A command flag to be used for debugging to exclude branches from similarity
// matching and outlining.
namespace llvm {
extern cl::opt<bool> DisableBranches;
// A command flag to be used for debugging to indirect calls from similarity
// matching and outlining.
extern cl::opt<bool> DisableIndirectCalls;
// A command flag to be used for debugging to exclude intrinsics from similarity
// matching and outlining.
extern cl::opt<bool> DisableIntrinsics;
} // namespace llvm
// Set to true if the user wants the ir outliner to run on linkonceodr linkage
// functions. This is false by default because the linker can dedupe linkonceodr
// functions. Since the outliner is confined to a single module (modulo LTO),
// this is off by default. It should, however, be the default behavior in
// LTO.
static cl::opt<bool> EnableLinkOnceODRIROutlining(
"enable-linkonceodr-ir-outlining", cl::Hidden,
cl::desc("Enable the IR outliner on linkonceodr functions"),
cl::init(false));
// This is a debug option to test small pieces of code to ensure that outlining
// works correctly.
static cl::opt<bool> NoCostModel(
"ir-outlining-no-cost", cl::init(false), cl::ReallyHidden,
cl::desc("Debug option to outline greedily, without restriction that "
"calculated benefit outweighs cost"));
/// The OutlinableGroup holds all the overarching information for outlining
/// a set of regions that are structurally similar to one another, such as the
/// types of the overall function, the output blocks, the sets of stores needed
/// and a list of the different regions. This information is used in the
/// deduplication of extracted regions with the same structure.
struct OutlinableGroup {
/// The sections that could be outlined
std::vector<OutlinableRegion *> Regions;
/// The argument types for the function created as the overall function to
/// replace the extracted function for each region.
std::vector<Type *> ArgumentTypes;
/// The FunctionType for the overall function.
FunctionType *OutlinedFunctionType = nullptr;
/// The Function for the collective overall function.
Function *OutlinedFunction = nullptr;
/// Flag for whether we should not consider this group of OutlinableRegions
/// for extraction.
bool IgnoreGroup = false;
/// The return blocks for the overall function.
DenseMap<Value *, BasicBlock *> EndBBs;
/// The PHIBlocks with their corresponding return block based on the return
/// value as the key.
DenseMap<Value *, BasicBlock *> PHIBlocks;
/// A set containing the different GVN store sets needed. Each array contains
/// a sorted list of the different values that need to be stored into output
/// registers.
DenseSet<ArrayRef<unsigned>> OutputGVNCombinations;
/// Flag for whether the \ref ArgumentTypes have been defined after the
/// extraction of the first region.
bool InputTypesSet = false;
/// The number of input values in \ref ArgumentTypes. Anything after this
/// index in ArgumentTypes is an output argument.
unsigned NumAggregateInputs = 0;
/// The mapping of the canonical numbering of the values in outlined sections
/// to specific arguments.
DenseMap<unsigned, unsigned> CanonicalNumberToAggArg;
/// The number of branches in the region target a basic block that is outside
/// of the region.
unsigned BranchesToOutside = 0;
/// Tracker counting backwards from the highest unsigned value possible to
/// avoid conflicting with the GVNs of assigned values. We start at -3 since
/// -2 and -1 are assigned by the DenseMap.
unsigned PHINodeGVNTracker = -3;
DenseMap<unsigned,
std::pair<std::pair<unsigned, unsigned>, SmallVector<unsigned, 2>>>
PHINodeGVNToGVNs;
DenseMap<hash_code, unsigned> GVNsToPHINodeGVN;
/// The number of instructions that will be outlined by extracting \ref
/// Regions.
InstructionCost Benefit = 0;
/// The number of added instructions needed for the outlining of the \ref
/// Regions.
InstructionCost Cost = 0;
/// The argument that needs to be marked with the swifterr attribute. If not
/// needed, there is no value.
std::optional<unsigned> SwiftErrorArgument;
/// For the \ref Regions, we look at every Value. If it is a constant,
/// we check whether it is the same in Region.
///
/// \param [in,out] NotSame contains the global value numbers where the
/// constant is not always the same, and must be passed in as an argument.
void findSameConstants(DenseSet<unsigned> &NotSame);
/// For the regions, look at each set of GVN stores needed and account for
/// each combination. Add an argument to the argument types if there is
/// more than one combination.
///
/// \param [in] M - The module we are outlining from.
void collectGVNStoreSets(Module &M);
};
/// Move the contents of \p SourceBB to before the last instruction of \p
/// TargetBB.
/// \param SourceBB - the BasicBlock to pull Instructions from.
/// \param TargetBB - the BasicBlock to put Instruction into.
static void moveBBContents(BasicBlock &SourceBB, BasicBlock &TargetBB) {
for (Instruction &I : llvm::make_early_inc_range(SourceBB))
I.moveBefore(TargetBB, TargetBB.end());
}
/// A function to sort the keys of \p Map, which must be a mapping of constant
/// values to basic blocks and return it in \p SortedKeys
///
/// \param SortedKeys - The vector the keys will be return in and sorted.
/// \param Map - The DenseMap containing keys to sort.
static void getSortedConstantKeys(std::vector<Value *> &SortedKeys,
DenseMap<Value *, BasicBlock *> &Map) {
for (auto &VtoBB : Map)
SortedKeys.push_back(VtoBB.first);
// Here we expect to have either 1 value that is void (nullptr) or multiple
// values that are all constant integers.
if (SortedKeys.size() == 1) {
assert(!SortedKeys[0] && "Expected a single void value.");
return;
}
stable_sort(SortedKeys, [](const Value *LHS, const Value *RHS) {
assert(LHS && RHS && "Expected non void values.");
const ConstantInt *LHSC = cast<ConstantInt>(LHS);
const ConstantInt *RHSC = cast<ConstantInt>(RHS);
return LHSC->getLimitedValue() < RHSC->getLimitedValue();
});
}
Value *OutlinableRegion::findCorrespondingValueIn(const OutlinableRegion &Other,
Value *V) {
std::optional<unsigned> GVN = Candidate->getGVN(V);
assert(GVN && "No GVN for incoming value");
std::optional<unsigned> CanonNum = Candidate->getCanonicalNum(*GVN);
std::optional<unsigned> FirstGVN =
Other.Candidate->fromCanonicalNum(*CanonNum);
std::optional<Value *> FoundValueOpt = Other.Candidate->fromGVN(*FirstGVN);
return FoundValueOpt.value_or(nullptr);
}
BasicBlock *
OutlinableRegion::findCorrespondingBlockIn(const OutlinableRegion &Other,
BasicBlock *BB) {
Instruction *FirstNonPHI = BB->getFirstNonPHI();
assert(FirstNonPHI && "block is empty?");
Value *CorrespondingVal = findCorrespondingValueIn(Other, FirstNonPHI);
if (!CorrespondingVal)
return nullptr;
BasicBlock *CorrespondingBlock =
cast<Instruction>(CorrespondingVal)->getParent();
return CorrespondingBlock;
}
/// Rewrite the BranchInsts in the incoming blocks to \p PHIBlock that are found
/// in \p Included to branch to BasicBlock \p Replace if they currently branch
/// to the BasicBlock \p Find. This is used to fix up the incoming basic blocks
/// when PHINodes are included in outlined regions.
///
/// \param PHIBlock - The BasicBlock containing the PHINodes that need to be
/// checked.
/// \param Find - The successor block to be replaced.
/// \param Replace - The new succesor block to branch to.
/// \param Included - The set of blocks about to be outlined.
static void replaceTargetsFromPHINode(BasicBlock *PHIBlock, BasicBlock *Find,
BasicBlock *Replace,
DenseSet<BasicBlock *> &Included) {
for (PHINode &PN : PHIBlock->phis()) {
for (unsigned Idx = 0, PNEnd = PN.getNumIncomingValues(); Idx != PNEnd;
++Idx) {
// Check if the incoming block is included in the set of blocks being
// outlined.
BasicBlock *Incoming = PN.getIncomingBlock(Idx);
if (!Included.contains(Incoming))
continue;
BranchInst *BI = dyn_cast<BranchInst>(Incoming->getTerminator());
assert(BI && "Not a branch instruction?");
// Look over the branching instructions into this block to see if we
// used to branch to Find in this outlined block.
for (unsigned Succ = 0, End = BI->getNumSuccessors(); Succ != End;
Succ++) {
// If we have found the block to replace, we do so here.
if (BI->getSuccessor(Succ) != Find)
continue;
BI->setSuccessor(Succ, Replace);
}
}
}
}
void OutlinableRegion::splitCandidate() {
assert(!CandidateSplit && "Candidate already split!");
Instruction *BackInst = Candidate->backInstruction();
Instruction *EndInst = nullptr;
// Check whether the last instruction is a terminator, if it is, we do
// not split on the following instruction. We leave the block as it is. We
// also check that this is not the last instruction in the Module, otherwise
// the check for whether the current following instruction matches the
// previously recorded instruction will be incorrect.
if (!BackInst->isTerminator() ||
BackInst->getParent() != &BackInst->getFunction()->back()) {
EndInst = Candidate->end()->Inst;
assert(EndInst && "Expected an end instruction?");
}
// We check if the current instruction following the last instruction in the
// region is the same as the recorded instruction following the last
// instruction. If they do not match, there could be problems in rewriting
// the program after outlining, so we ignore it.
if (!BackInst->isTerminator() &&
EndInst != BackInst->getNextNonDebugInstruction())
return;
Instruction *StartInst = (*Candidate->begin()).Inst;
assert(StartInst && "Expected a start instruction?");
StartBB = StartInst->getParent();
PrevBB = StartBB;
DenseSet<BasicBlock *> BBSet;
Candidate->getBasicBlocks(BBSet);
// We iterate over the instructions in the region, if we find a PHINode, we
// check if there are predecessors outside of the region, if there are,
// we ignore this region since we are unable to handle the severing of the
// phi node right now.
// TODO: Handle extraneous inputs for PHINodes through variable number of
// inputs, similar to how outputs are handled.
BasicBlock::iterator It = StartInst->getIterator();
EndBB = BackInst->getParent();
BasicBlock *IBlock;
BasicBlock *PHIPredBlock = nullptr;
bool EndBBTermAndBackInstDifferent = EndBB->getTerminator() != BackInst;
while (PHINode *PN = dyn_cast<PHINode>(&*It)) {
unsigned NumPredsOutsideRegion = 0;
for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
if (!BBSet.contains(PN->getIncomingBlock(i))) {
PHIPredBlock = PN->getIncomingBlock(i);
++NumPredsOutsideRegion;
continue;
}
// We must consider the case there the incoming block to the PHINode is
// the same as the final block of the OutlinableRegion. If this is the
// case, the branch from this block must also be outlined to be valid.
IBlock = PN->getIncomingBlock(i);
if (IBlock == EndBB && EndBBTermAndBackInstDifferent) {
PHIPredBlock = PN->getIncomingBlock(i);
++NumPredsOutsideRegion;
}
}
if (NumPredsOutsideRegion > 1)
return;
It++;
}
// If the region starts with a PHINode, but is not the initial instruction of
// the BasicBlock, we ignore this region for now.
if (isa<PHINode>(StartInst) && StartInst != &*StartBB->begin())
return;
// If the region ends with a PHINode, but does not contain all of the phi node
// instructions of the region, we ignore it for now.
if (isa<PHINode>(BackInst) &&
BackInst != &*std::prev(EndBB->getFirstInsertionPt()))
return;
// The basic block gets split like so:
// block: block:
// inst1 inst1
// inst2 inst2
// region1 br block_to_outline
// region2 block_to_outline:
// region3 -> region1
// region4 region2
// inst3 region3
// inst4 region4
// br block_after_outline
// block_after_outline:
// inst3
// inst4
std::string OriginalName = PrevBB->getName().str();
StartBB = PrevBB->splitBasicBlock(StartInst, OriginalName + "_to_outline");
PrevBB->replaceSuccessorsPhiUsesWith(PrevBB, StartBB);
// If there was a PHINode with an incoming block outside the region,
// make sure is correctly updated in the newly split block.
if (PHIPredBlock)
PrevBB->replaceSuccessorsPhiUsesWith(PHIPredBlock, PrevBB);
CandidateSplit = true;
if (!BackInst->isTerminator()) {
EndBB = EndInst->getParent();
FollowBB = EndBB->splitBasicBlock(EndInst, OriginalName + "_after_outline");
EndBB->replaceSuccessorsPhiUsesWith(EndBB, FollowBB);
FollowBB->replaceSuccessorsPhiUsesWith(PrevBB, FollowBB);
} else {
EndBB = BackInst->getParent();
EndsInBranch = true;
FollowBB = nullptr;
}
// Refind the basic block set.
BBSet.clear();
Candidate->getBasicBlocks(BBSet);
// For the phi nodes in the new starting basic block of the region, we
// reassign the targets of the basic blocks branching instructions.
replaceTargetsFromPHINode(StartBB, PrevBB, StartBB, BBSet);
if (FollowBB)
replaceTargetsFromPHINode(FollowBB, EndBB, FollowBB, BBSet);
}
void OutlinableRegion::reattachCandidate() {
assert(CandidateSplit && "Candidate is not split!");
// The basic block gets reattached like so:
// block: block:
// inst1 inst1
// inst2 inst2
// br block_to_outline region1
// block_to_outline: -> region2
// region1 region3
// region2 region4
// region3 inst3
// region4 inst4
// br block_after_outline
// block_after_outline:
// inst3
// inst4
assert(StartBB != nullptr && "StartBB for Candidate is not defined!");
assert(PrevBB->getTerminator() && "Terminator removed from PrevBB!");
// Make sure PHINode references to the block we are merging into are
// updated to be incoming blocks from the predecessor to the current block.
// NOTE: If this is updated such that the outlined block can have more than
// one incoming block to a PHINode, this logic will have to updated
// to handle multiple precessors instead.
// We only need to update this if the outlined section contains a PHINode, if
// it does not, then the incoming block was never changed in the first place.
// On the other hand, if PrevBB has no predecessors, it means that all
// incoming blocks to the first block are contained in the region, and there
// will be nothing to update.
Instruction *StartInst = (*Candidate->begin()).Inst;
if (isa<PHINode>(StartInst) && !PrevBB->hasNPredecessors(0)) {
assert(!PrevBB->hasNPredecessorsOrMore(2) &&
"PrevBB has more than one predecessor. Should be 0 or 1.");
BasicBlock *BeforePrevBB = PrevBB->getSinglePredecessor();
PrevBB->replaceSuccessorsPhiUsesWith(PrevBB, BeforePrevBB);
}
PrevBB->getTerminator()->eraseFromParent();
// If we reattaching after outlining, we iterate over the phi nodes to
// the initial block, and reassign the branch instructions of the incoming
// blocks to the block we are remerging into.
if (!ExtractedFunction) {
DenseSet<BasicBlock *> BBSet;
Candidate->getBasicBlocks(BBSet);
replaceTargetsFromPHINode(StartBB, StartBB, PrevBB, BBSet);
if (!EndsInBranch)
replaceTargetsFromPHINode(FollowBB, FollowBB, EndBB, BBSet);
}
moveBBContents(*StartBB, *PrevBB);
BasicBlock *PlacementBB = PrevBB;
if (StartBB != EndBB)
PlacementBB = EndBB;
if (!EndsInBranch && PlacementBB->getUniqueSuccessor() != nullptr) {
assert(FollowBB != nullptr && "FollowBB for Candidate is not defined!");
assert(PlacementBB->getTerminator() && "Terminator removed from EndBB!");
PlacementBB->getTerminator()->eraseFromParent();
moveBBContents(*FollowBB, *PlacementBB);
PlacementBB->replaceSuccessorsPhiUsesWith(FollowBB, PlacementBB);
FollowBB->eraseFromParent();
}
PrevBB->replaceSuccessorsPhiUsesWith(StartBB, PrevBB);
StartBB->eraseFromParent();
// Make sure to save changes back to the StartBB.
StartBB = PrevBB;
EndBB = nullptr;
PrevBB = nullptr;
FollowBB = nullptr;
CandidateSplit = false;
}
/// Find whether \p V matches the Constants previously found for the \p GVN.
///
/// \param V - The value to check for consistency.
/// \param GVN - The global value number assigned to \p V.
/// \param GVNToConstant - The mapping of global value number to Constants.
/// \returns true if the Value matches the Constant mapped to by V and false if
/// it \p V is a Constant but does not match.
/// \returns std::nullopt if \p V is not a Constant.
static std::optional<bool>
constantMatches(Value *V, unsigned GVN,
DenseMap<unsigned, Constant *> &GVNToConstant) {
// See if we have a constants
Constant *CST = dyn_cast<Constant>(V);
if (!CST)
return std::nullopt;
// Holds a mapping from a global value number to a Constant.
DenseMap<unsigned, Constant *>::iterator GVNToConstantIt;
bool Inserted;
// If we have a constant, try to make a new entry in the GVNToConstant.
std::tie(GVNToConstantIt, Inserted) =
GVNToConstant.insert(std::make_pair(GVN, CST));
// If it was found and is not equal, it is not the same. We do not
// handle this case yet, and exit early.
if (Inserted || (GVNToConstantIt->second == CST))
return true;
return false;
}
InstructionCost OutlinableRegion::getBenefit(TargetTransformInfo &TTI) {
InstructionCost Benefit = 0;
// Estimate the benefit of outlining a specific sections of the program. We
// delegate mostly this task to the TargetTransformInfo so that if the target
// has specific changes, we can have a more accurate estimate.
// However, getInstructionCost delegates the code size calculation for
// arithmetic instructions to getArithmeticInstrCost in
// include/Analysis/TargetTransformImpl.h, where it always estimates that the
// code size for a division and remainder instruction to be equal to 4, and
// everything else to 1. This is not an accurate representation of the
// division instruction for targets that have a native division instruction.
// To be overly conservative, we only add 1 to the number of instructions for
// each division instruction.
for (IRInstructionData &ID : *Candidate) {
Instruction *I = ID.Inst;
switch (I->getOpcode()) {
case Instruction::FDiv:
case Instruction::FRem:
case Instruction::SDiv:
case Instruction::SRem:
case Instruction::UDiv:
case Instruction::URem:
Benefit += 1;
break;
default:
Benefit += TTI.getInstructionCost(I, TargetTransformInfo::TCK_CodeSize);
break;
}
}
return Benefit;
}
/// Check the \p OutputMappings structure for value \p Input, if it exists
/// it has been used as an output for outlining, and has been renamed, and we
/// return the new value, otherwise, we return the same value.
///
/// \param OutputMappings [in] - The mapping of values to their renamed value
/// after being used as an output for an outlined region.
/// \param Input [in] - The value to find the remapped value of, if it exists.
/// \return The remapped value if it has been renamed, and the same value if has
/// not.
static Value *findOutputMapping(const DenseMap<Value *, Value *> OutputMappings,
Value *Input) {
DenseMap<Value *, Value *>::const_iterator OutputMapping =
OutputMappings.find(Input);
if (OutputMapping != OutputMappings.end())
return OutputMapping->second;
return Input;
}
/// Find whether \p Region matches the global value numbering to Constant
/// mapping found so far.
///
/// \param Region - The OutlinableRegion we are checking for constants
/// \param GVNToConstant - The mapping of global value number to Constants.
/// \param NotSame - The set of global value numbers that do not have the same
/// constant in each region.
/// \returns true if all Constants are the same in every use of a Constant in \p
/// Region and false if not
static bool
collectRegionsConstants(OutlinableRegion &Region,
DenseMap<unsigned, Constant *> &GVNToConstant,
DenseSet<unsigned> &NotSame) {
bool ConstantsTheSame = true;
IRSimilarityCandidate &C = *Region.Candidate;
for (IRInstructionData &ID : C) {
// Iterate over the operands in an instruction. If the global value number,
// assigned by the IRSimilarityCandidate, has been seen before, we check if
// the the number has been found to be not the same value in each instance.
for (Value *V : ID.OperVals) {
std::optional<unsigned> GVNOpt = C.getGVN(V);
assert(GVNOpt && "Expected a GVN for operand?");
unsigned GVN = *GVNOpt;
// Check if this global value has been found to not be the same already.
if (NotSame.contains(GVN)) {
if (isa<Constant>(V))
ConstantsTheSame = false;
continue;
}
// If it has been the same so far, we check the value for if the
// associated Constant value match the previous instances of the same
// global value number. If the global value does not map to a Constant,
// it is considered to not be the same value.
std::optional<bool> ConstantMatches =
constantMatches(V, GVN, GVNToConstant);
if (ConstantMatches) {
if (*ConstantMatches)
continue;
else
ConstantsTheSame = false;
}
// While this value is a register, it might not have been previously,
// make sure we don't already have a constant mapped to this global value
// number.
if (GVNToConstant.contains(GVN))
ConstantsTheSame = false;
NotSame.insert(GVN);
}
}
return ConstantsTheSame;
}
void OutlinableGroup::findSameConstants(DenseSet<unsigned> &NotSame) {
DenseMap<unsigned, Constant *> GVNToConstant;
for (OutlinableRegion *Region : Regions)
collectRegionsConstants(*Region, GVNToConstant, NotSame);
}
void OutlinableGroup::collectGVNStoreSets(Module &M) {
for (OutlinableRegion *OS : Regions)
OutputGVNCombinations.insert(OS->GVNStores);
// We are adding an extracted argument to decide between which output path
// to use in the basic block. It is used in a switch statement and only
// needs to be an integer.
if (OutputGVNCombinations.size() > 1)
ArgumentTypes.push_back(Type::getInt32Ty(M.getContext()));
}
/// Get the subprogram if it exists for one of the outlined regions.
///
/// \param [in] Group - The set of regions to find a subprogram for.
/// \returns the subprogram if it exists, or nullptr.
static DISubprogram *getSubprogramOrNull(OutlinableGroup &Group) {
for (OutlinableRegion *OS : Group.Regions)
if (Function *F = OS->Call->getFunction())
if (DISubprogram *SP = F->getSubprogram())
return SP;
return nullptr;
}
Function *IROutliner::createFunction(Module &M, OutlinableGroup &Group,
unsigned FunctionNameSuffix) {
assert(!Group.OutlinedFunction && "Function is already defined!");
Type *RetTy = Type::getVoidTy(M.getContext());
// All extracted functions _should_ have the same return type at this point
// since the similarity identifier ensures that all branches outside of the
// region occur in the same place.
// NOTE: Should we ever move to the model that uses a switch at every point
// needed, meaning that we could branch within the region or out, it is
// possible that we will need to switch to using the most general case all of
// the time.
for (OutlinableRegion *R : Group.Regions) {
Type *ExtractedFuncType = R->ExtractedFunction->getReturnType();
if ((RetTy->isVoidTy() && !ExtractedFuncType->isVoidTy()) ||
(RetTy->isIntegerTy(1) && ExtractedFuncType->isIntegerTy(16)))
RetTy = ExtractedFuncType;
}
Group.OutlinedFunctionType = FunctionType::get(
RetTy, Group.ArgumentTypes, false);
// These functions will only be called from within the same module, so
// we can set an internal linkage.
Group.OutlinedFunction = Function::Create(
Group.OutlinedFunctionType, GlobalValue::InternalLinkage,
"outlined_ir_func_" + std::to_string(FunctionNameSuffix), M);
// Transfer the swifterr attribute to the correct function parameter.
if (Group.SwiftErrorArgument)
Group.OutlinedFunction->addParamAttr(*Group.SwiftErrorArgument,
Attribute::SwiftError);
Group.OutlinedFunction->addFnAttr(Attribute::OptimizeForSize);
Group.OutlinedFunction->addFnAttr(Attribute::MinSize);
// If there's a DISubprogram associated with this outlined function, then
// emit debug info for the outlined function.
if (DISubprogram *SP = getSubprogramOrNull(Group)) {
Function *F = Group.OutlinedFunction;
// We have a DISubprogram. Get its DICompileUnit.
DICompileUnit *CU = SP->getUnit();
DIBuilder DB(M, true, CU);
DIFile *Unit = SP->getFile();
Mangler Mg;
// Get the mangled name of the function for the linkage name.
std::string Dummy;
llvm::raw_string_ostream MangledNameStream(Dummy);
Mg.getNameWithPrefix(MangledNameStream, F, false);
DISubprogram *OutlinedSP = DB.createFunction(
Unit /* Context */, F->getName(), MangledNameStream.str(),
Unit /* File */,
0 /* Line 0 is reserved for compiler-generated code. */,
DB.createSubroutineType(
DB.getOrCreateTypeArray(std::nullopt)), /* void type */
0, /* Line 0 is reserved for compiler-generated code. */
DINode::DIFlags::FlagArtificial /* Compiler-generated code. */,
/* Outlined code is optimized code by definition. */
DISubprogram::SPFlagDefinition | DISubprogram::SPFlagOptimized);
// Don't add any new variables to the subprogram.
DB.finalizeSubprogram(OutlinedSP);
// Attach subprogram to the function.
F->setSubprogram(OutlinedSP);
// We're done with the DIBuilder.
DB.finalize();
}
return Group.OutlinedFunction;
}
/// Move each BasicBlock in \p Old to \p New.
///
/// \param [in] Old - The function to move the basic blocks from.
/// \param [in] New - The function to move the basic blocks to.
/// \param [out] NewEnds - The return blocks of the new overall function.
static void moveFunctionData(Function &Old, Function &New,
DenseMap<Value *, BasicBlock *> &NewEnds) {
for (BasicBlock &CurrBB : llvm::make_early_inc_range(Old)) {
CurrBB.removeFromParent();
CurrBB.insertInto(&New);
Instruction *I = CurrBB.getTerminator();
// For each block we find a return instruction is, it is a potential exit
// path for the function. We keep track of each block based on the return
// value here.
if (ReturnInst *RI = dyn_cast<ReturnInst>(I))
NewEnds.insert(std::make_pair(RI->getReturnValue(), &CurrBB));
std::vector<Instruction *> DebugInsts;
for (Instruction &Val : CurrBB) {
// We must handle the scoping of called functions differently than
// other outlined instructions.
if (!isa<CallInst>(&Val)) {
// Remove the debug information for outlined functions.
Val.setDebugLoc(DebugLoc());
// Loop info metadata may contain line locations. Update them to have no
// value in the new subprogram since the outlined code could be from
// several locations.
auto updateLoopInfoLoc = [&New](Metadata *MD) -> Metadata * {
if (DISubprogram *SP = New.getSubprogram())
if (auto *Loc = dyn_cast_or_null<DILocation>(MD))
return DILocation::get(New.getContext(), Loc->getLine(),
Loc->getColumn(), SP, nullptr);
return MD;
};
updateLoopMetadataDebugLocations(Val, updateLoopInfoLoc);
continue;
}
// From this point we are only handling call instructions.
CallInst *CI = cast<CallInst>(&Val);
// We add any debug statements here, to be removed after. Since the
// instructions originate from many different locations in the program,
// it will cause incorrect reporting from a debugger if we keep the
// same debug instructions.
if (isa<DbgInfoIntrinsic>(CI)) {
DebugInsts.push_back(&Val);
continue;
}
// Edit the scope of called functions inside of outlined functions.
if (DISubprogram *SP = New.getSubprogram()) {
DILocation *DI = DILocation::get(New.getContext(), 0, 0, SP);
Val.setDebugLoc(DI);
}
}
for (Instruction *I : DebugInsts)
I->eraseFromParent();
}
}
/// Find the the constants that will need to be lifted into arguments
/// as they are not the same in each instance of the region.
///
/// \param [in] C - The IRSimilarityCandidate containing the region we are
/// analyzing.
/// \param [in] NotSame - The set of global value numbers that do not have a
/// single Constant across all OutlinableRegions similar to \p C.
/// \param [out] Inputs - The list containing the global value numbers of the
/// arguments needed for the region of code.
static void findConstants(IRSimilarityCandidate &C, DenseSet<unsigned> &NotSame,
std::vector<unsigned> &Inputs) {
DenseSet<unsigned> Seen;
// Iterate over the instructions, and find what constants will need to be
// extracted into arguments.
for (IRInstructionDataList::iterator IDIt = C.begin(), EndIDIt = C.end();
IDIt != EndIDIt; IDIt++) {
for (Value *V : (*IDIt).OperVals) {
// Since these are stored before any outlining, they will be in the
// global value numbering.
unsigned GVN = *C.getGVN(V);
if (isa<Constant>(V))
if (NotSame.contains(GVN) && !Seen.contains(GVN)) {
Inputs.push_back(GVN);
Seen.insert(GVN);
}
}
}
}
/// Find the GVN for the inputs that have been found by the CodeExtractor.
///
/// \param [in] C - The IRSimilarityCandidate containing the region we are
/// analyzing.
/// \param [in] CurrentInputs - The set of inputs found by the
/// CodeExtractor.
/// \param [in] OutputMappings - The mapping of values that have been replaced
/// by a new output value.
/// \param [out] EndInputNumbers - The global value numbers for the extracted
/// arguments.
static void mapInputsToGVNs(IRSimilarityCandidate &C,
SetVector<Value *> &CurrentInputs,
const DenseMap<Value *, Value *> &OutputMappings,
std::vector<unsigned> &EndInputNumbers) {
// Get the Global Value Number for each input. We check if the Value has been
// replaced by a different value at output, and use the original value before
// replacement.
for (Value *Input : CurrentInputs) {
assert(Input && "Have a nullptr as an input");
if (OutputMappings.contains(Input))
Input = OutputMappings.find(Input)->second;
assert(C.getGVN(Input) && "Could not find a numbering for the given input");
EndInputNumbers.push_back(*C.getGVN(Input));
}
}
/// Find the original value for the \p ArgInput values if any one of them was
/// replaced during a previous extraction.
///
/// \param [in] ArgInputs - The inputs to be extracted by the code extractor.
/// \param [in] OutputMappings - The mapping of values that have been replaced
/// by a new output value.
/// \param [out] RemappedArgInputs - The remapped values according to
/// \p OutputMappings that will be extracted.
static void
remapExtractedInputs(const ArrayRef<Value *> ArgInputs,
const DenseMap<Value *, Value *> &OutputMappings,
SetVector<Value *> &RemappedArgInputs) {
// Get the global value number for each input that will be extracted as an
// argument by the code extractor, remapping if needed for reloaded values.
for (Value *Input : ArgInputs) {
if (OutputMappings.contains(Input))
Input = OutputMappings.find(Input)->second;
RemappedArgInputs.insert(Input);
}
}
/// Find the input GVNs and the output values for a region of Instructions.
/// Using the code extractor, we collect the inputs to the extracted function.
///
/// The \p Region can be identified as needing to be ignored in this function.
/// It should be checked whether it should be ignored after a call to this
/// function.
///
/// \param [in,out] Region - The region of code to be analyzed.
/// \param [out] InputGVNs - The global value numbers for the extracted
/// arguments.
/// \param [in] NotSame - The global value numbers in the region that do not
/// have the same constant value in the regions structurally similar to
/// \p Region.
/// \param [in] OutputMappings - The mapping of values that have been replaced
/// by a new output value after extraction.
/// \param [out] ArgInputs - The values of the inputs to the extracted function.
/// \param [out] Outputs - The set of values extracted by the CodeExtractor
/// as outputs.
static void getCodeExtractorArguments(
OutlinableRegion &Region, std::vector<unsigned> &InputGVNs,
DenseSet<unsigned> &NotSame, DenseMap<Value *, Value *> &OutputMappings,
SetVector<Value *> &ArgInputs, SetVector<Value *> &Outputs) {
IRSimilarityCandidate &C = *Region.Candidate;
// OverallInputs are the inputs to the region found by the CodeExtractor,
// SinkCands and HoistCands are used by the CodeExtractor to find sunken
// allocas of values whose lifetimes are contained completely within the
// outlined region. PremappedInputs are the arguments found by the
// CodeExtractor, removing conditions such as sunken allocas, but that
// may need to be remapped due to the extracted output values replacing
// the original values. We use DummyOutputs for this first run of finding
// inputs and outputs since the outputs could change during findAllocas,
// the correct set of extracted outputs will be in the final Outputs ValueSet.
SetVector<Value *> OverallInputs, PremappedInputs, SinkCands, HoistCands,
DummyOutputs;
// Use the code extractor to get the inputs and outputs, without sunken
// allocas or removing llvm.assumes.
CodeExtractor *CE = Region.CE;
CE->findInputsOutputs(OverallInputs, DummyOutputs, SinkCands);
assert(Region.StartBB && "Region must have a start BasicBlock!");
Function *OrigF = Region.StartBB->getParent();
CodeExtractorAnalysisCache CEAC(*OrigF);
BasicBlock *Dummy = nullptr;
// The region may be ineligible due to VarArgs in the parent function. In this
// case we ignore the region.
if (!CE->isEligible()) {
Region.IgnoreRegion = true;
return;
}
// Find if any values are going to be sunk into the function when extracted
CE->findAllocas(CEAC, SinkCands, HoistCands, Dummy);
CE->findInputsOutputs(PremappedInputs, Outputs, SinkCands);
// TODO: Support regions with sunken allocas: values whose lifetimes are
// contained completely within the outlined region. These are not guaranteed
// to be the same in every region, so we must elevate them all to arguments
// when they appear. If these values are not equal, it means there is some
// Input in OverallInputs that was removed for ArgInputs.
if (OverallInputs.size() != PremappedInputs.size()) {
Region.IgnoreRegion = true;
return;
}
findConstants(C, NotSame, InputGVNs);
mapInputsToGVNs(C, OverallInputs, OutputMappings, InputGVNs);
remapExtractedInputs(PremappedInputs.getArrayRef(), OutputMappings,
ArgInputs);
// Sort the GVNs, since we now have constants included in the \ref InputGVNs
// we need to make sure they are in a deterministic order.
stable_sort(InputGVNs);
}
/// Look over the inputs and map each input argument to an argument in the
/// overall function for the OutlinableRegions. This creates a way to replace
/// the arguments of the extracted function with the arguments of the new
/// overall function.
///
/// \param [in,out] Region - The region of code to be analyzed.
/// \param [in] InputGVNs - The global value numbering of the input values
/// collected.
/// \param [in] ArgInputs - The values of the arguments to the extracted
/// function.
static void
findExtractedInputToOverallInputMapping(OutlinableRegion &Region,
std::vector<unsigned> &InputGVNs,
SetVector<Value *> &ArgInputs) {
IRSimilarityCandidate &C = *Region.Candidate;
OutlinableGroup &Group = *Region.Parent;
// This counts the argument number in the overall function.
unsigned TypeIndex = 0;
// This counts the argument number in the extracted function.
unsigned OriginalIndex = 0;
// Find the mapping of the extracted arguments to the arguments for the
// overall function. Since there may be extra arguments in the overall
// function to account for the extracted constants, we have two different
// counters as we find extracted arguments, and as we come across overall
// arguments.
// Additionally, in our first pass, for the first extracted function,
// we find argument locations for the canonical value numbering. This
// numbering overrides any discovered location for the extracted code.
for (unsigned InputVal : InputGVNs) {
std::optional<unsigned> CanonicalNumberOpt = C.getCanonicalNum(InputVal);
assert(CanonicalNumberOpt && "Canonical number not found?");
unsigned CanonicalNumber = *CanonicalNumberOpt;
std::optional<Value *> InputOpt = C.fromGVN(InputVal);
assert(InputOpt && "Global value number not found?");
Value *Input = *InputOpt;
DenseMap<unsigned, unsigned>::iterator AggArgIt =
Group.CanonicalNumberToAggArg.find(CanonicalNumber);
if (!Group.InputTypesSet) {
Group.ArgumentTypes.push_back(Input->getType());
// If the input value has a swifterr attribute, make sure to mark the
// argument in the overall function.
if (Input->isSwiftError()) {
assert(
!Group.SwiftErrorArgument &&
"Argument already marked with swifterr for this OutlinableGroup!");
Group.SwiftErrorArgument = TypeIndex;
}
}
// Check if we have a constant. If we do add it to the overall argument
// number to Constant map for the region, and continue to the next input.
if (Constant *CST = dyn_cast<Constant>(Input)) {
if (AggArgIt != Group.CanonicalNumberToAggArg.end())
Region.AggArgToConstant.insert(std::make_pair(AggArgIt->second, CST));
else {
Group.CanonicalNumberToAggArg.insert(
std::make_pair(CanonicalNumber, TypeIndex));
Region.AggArgToConstant.insert(std::make_pair(TypeIndex, CST));
}
TypeIndex++;
continue;
}
// It is not a constant, we create the mapping from extracted argument list
// to the overall argument list, using the canonical location, if it exists.
assert(ArgInputs.count(Input) && "Input cannot be found!");
if (AggArgIt != Group.CanonicalNumberToAggArg.end()) {
if (OriginalIndex != AggArgIt->second)
Region.ChangedArgOrder = true;
Region.ExtractedArgToAgg.insert(
std::make_pair(OriginalIndex, AggArgIt->second));
Region.AggArgToExtracted.insert(
std::make_pair(AggArgIt->second, OriginalIndex));
} else {
Group.CanonicalNumberToAggArg.insert(
std::make_pair(CanonicalNumber, TypeIndex));
Region.ExtractedArgToAgg.insert(std::make_pair(OriginalIndex, TypeIndex));
Region.AggArgToExtracted.insert(std::make_pair(TypeIndex, OriginalIndex));
}
OriginalIndex++;
TypeIndex++;
}
// If the function type definitions for the OutlinableGroup holding the region
// have not been set, set the length of the inputs here. We should have the
// same inputs for all of the different regions contained in the
// OutlinableGroup since they are all structurally similar to one another.
if (!Group.InputTypesSet) {
Group.NumAggregateInputs = TypeIndex;
Group.InputTypesSet = true;
}
Region.NumExtractedInputs = OriginalIndex;
}
/// Check if the \p V has any uses outside of the region other than \p PN.
///
/// \param V [in] - The value to check.
/// \param PHILoc [in] - The location in the PHINode of \p V.
/// \param PN [in] - The PHINode using \p V.
/// \param Exits [in] - The potential blocks we exit to from the outlined
/// region.
/// \param BlocksInRegion [in] - The basic blocks contained in the region.
/// \returns true if \p V has any use soutside its region other than \p PN.
static bool outputHasNonPHI(Value *V, unsigned PHILoc, PHINode &PN,
SmallPtrSet<BasicBlock *, 1> &Exits,
DenseSet<BasicBlock *> &BlocksInRegion) {
// We check to see if the value is used by the PHINode from some other
// predecessor not included in the region. If it is, we make sure
// to keep it as an output.
if (any_of(llvm::seq<unsigned>(0, PN.getNumIncomingValues()),
[PHILoc, &PN, V, &BlocksInRegion](unsigned Idx) {
return (Idx != PHILoc && V == PN.getIncomingValue(Idx) &&
!BlocksInRegion.contains(PN.getIncomingBlock(Idx)));
}))
return true;
// Check if the value is used by any other instructions outside the region.
return any_of(V->users(), [&Exits, &BlocksInRegion](User *U) {
Instruction *I = dyn_cast<Instruction>(U);
if (!I)
return false;
// If the use of the item is inside the region, we skip it. Uses
// inside the region give us useful information about how the item could be
// used as an output.
BasicBlock *Parent = I->getParent();
if (BlocksInRegion.contains(Parent))
return false;
// If it's not a PHINode then we definitely know the use matters. This
// output value will not completely combined with another item in a PHINode
// as it is directly reference by another non-phi instruction
if (!isa<PHINode>(I))
return true;
// If we have a PHINode outside one of the exit locations, then it
// can be considered an outside use as well. If there is a PHINode
// contained in the Exit where this values use matters, it will be
// caught when we analyze that PHINode.
if (!Exits.contains(Parent))
return true;
return false;
});
}
/// Test whether \p CurrentExitFromRegion contains any PhiNodes that should be
/// considered outputs. A PHINodes is an output when more than one incoming
/// value has been marked by the CodeExtractor as an output.
///
/// \param CurrentExitFromRegion [in] - The block to analyze.
/// \param PotentialExitsFromRegion [in] - The potential exit blocks from the
/// region.
/// \param RegionBlocks [in] - The basic blocks in the region.
/// \param Outputs [in, out] - The existing outputs for the region, we may add
/// PHINodes to this as we find that they replace output values.
/// \param OutputsReplacedByPHINode [out] - A set containing outputs that are
/// totally replaced by a PHINode.
/// \param OutputsWithNonPhiUses [out] - A set containing outputs that are used
/// in PHINodes, but have other uses, and should still be considered outputs.
static void analyzeExitPHIsForOutputUses(
BasicBlock *CurrentExitFromRegion,
SmallPtrSet<BasicBlock *, 1> &PotentialExitsFromRegion,
DenseSet<BasicBlock *> &RegionBlocks, SetVector<Value *> &Outputs,
DenseSet<Value *> &OutputsReplacedByPHINode,
DenseSet<Value *> &OutputsWithNonPhiUses) {
for (PHINode &PN : CurrentExitFromRegion->phis()) {
// Find all incoming values from the outlining region.
SmallVector<unsigned, 2> IncomingVals;
for (unsigned I = 0, E = PN.getNumIncomingValues(); I < E; ++I)
if (RegionBlocks.contains(PN.getIncomingBlock(I)))
IncomingVals.push_back(I);
// Do not process PHI if there are no predecessors from region.
unsigned NumIncomingVals = IncomingVals.size();
if (NumIncomingVals == 0)
continue;
// If there is one predecessor, we mark it as a value that needs to be kept
// as an output.
if (NumIncomingVals == 1) {
Value *V = PN.getIncomingValue(*IncomingVals.begin());
OutputsWithNonPhiUses.insert(V);
OutputsReplacedByPHINode.erase(V);
continue;
}
// This PHINode will be used as an output value, so we add it to our list.
Outputs.insert(&PN);
// Not all of the incoming values should be ignored as other inputs and
// outputs may have uses in outlined region. If they have other uses
// outside of the single PHINode we should not skip over it.
for (unsigned Idx : IncomingVals) {
Value *V = PN.getIncomingValue(Idx);
if (outputHasNonPHI(V, Idx, PN, PotentialExitsFromRegion, RegionBlocks)) {
OutputsWithNonPhiUses.insert(V);
OutputsReplacedByPHINode.erase(V);
continue;
}
if (!OutputsWithNonPhiUses.contains(V))
OutputsReplacedByPHINode.insert(V);
}
}
}
// Represents the type for the unsigned number denoting the output number for
// phi node, along with the canonical number for the exit block.
using ArgLocWithBBCanon = std::pair<unsigned, unsigned>;
// The list of canonical numbers for the incoming values to a PHINode.
using CanonList = SmallVector<unsigned, 2>;
// The pair type representing the set of canonical values being combined in the
// PHINode, along with the location data for the PHINode.
using PHINodeData = std::pair<ArgLocWithBBCanon, CanonList>;
/// Encode \p PND as an integer for easy lookup based on the argument location,
/// the parent BasicBlock canonical numbering, and the canonical numbering of
/// the values stored in the PHINode.
///
/// \param PND - The data to hash.
/// \returns The hash code of \p PND.
static hash_code encodePHINodeData(PHINodeData &PND) {
return llvm::hash_combine(
llvm::hash_value(PND.first.first), llvm::hash_value(PND.first.second),
llvm::hash_combine_range(PND.second.begin(), PND.second.end()));
}
/// Create a special GVN for PHINodes that will be used outside of
/// the region. We create a hash code based on the Canonical number of the
/// parent BasicBlock, the canonical numbering of the values stored in the
/// PHINode and the aggregate argument location. This is used to find whether
/// this PHINode type has been given a canonical numbering already. If not, we
/// assign it a value and store it for later use. The value is returned to
/// identify different output schemes for the set of regions.
///
/// \param Region - The region that \p PN is an output for.
/// \param PN - The PHINode we are analyzing.
/// \param Blocks - The blocks for the region we are analyzing.
/// \param AggArgIdx - The argument \p PN will be stored into.
/// \returns An optional holding the assigned canonical number, or std::nullopt
/// if there is some attribute of the PHINode blocking it from being used.
static std::optional<unsigned> getGVNForPHINode(OutlinableRegion &Region,
PHINode *PN,
DenseSet<BasicBlock *> &Blocks,
unsigned AggArgIdx) {
OutlinableGroup &Group = *Region.Parent;
IRSimilarityCandidate &Cand = *Region.Candidate;
BasicBlock *PHIBB = PN->getParent();
CanonList PHIGVNs;
Value *Incoming;
BasicBlock *IncomingBlock;
for (unsigned Idx = 0, EIdx = PN->getNumIncomingValues(); Idx < EIdx; Idx++) {
Incoming = PN->getIncomingValue(Idx);
IncomingBlock = PN->getIncomingBlock(Idx);
// If we cannot find a GVN, and the incoming block is included in the region
// this means that the input to the PHINode is not included in the region we
// are trying to analyze, meaning, that if it was outlined, we would be
// adding an extra input. We ignore this case for now, and so ignore the
// region.
std::optional<unsigned> OGVN = Cand.getGVN(Incoming);
if (!OGVN && Blocks.contains(IncomingBlock)) {
Region.IgnoreRegion = true;
return std::nullopt;
}
// If the incoming block isn't in the region, we don't have to worry about
// this incoming value.
if (!Blocks.contains(IncomingBlock))
continue;
// Collect the canonical numbers of the values in the PHINode.
unsigned GVN = *OGVN;
OGVN = Cand.getCanonicalNum(GVN);
assert(OGVN && "No GVN found for incoming value?");
PHIGVNs.push_back(*OGVN);
// Find the incoming block and use the canonical numbering as well to define
// the hash for the PHINode.
OGVN = Cand.getGVN(IncomingBlock);
// If there is no number for the incoming block, it is because we have
// split the candidate basic blocks. So we use the previous block that it
// was split from to find the valid global value numbering for the PHINode.
if (!OGVN) {
assert(Cand.getStartBB() == IncomingBlock &&
"Unknown basic block used in exit path PHINode.");
BasicBlock *PrevBlock = nullptr;
// Iterate over the predecessors to the incoming block of the
// PHINode, when we find a block that is not contained in the region
// we know that this is the first block that we split from, and should
// have a valid global value numbering.
for (BasicBlock *Pred : predecessors(IncomingBlock))
if (!Blocks.contains(Pred)) {
PrevBlock = Pred;
break;
}
assert(PrevBlock && "Expected a predecessor not in the reigon!");
OGVN = Cand.getGVN(PrevBlock);
}
GVN = *OGVN;
OGVN = Cand.getCanonicalNum(GVN);
assert(OGVN && "No GVN found for incoming block?");
PHIGVNs.push_back(*OGVN);
}
// Now that we have the GVNs for the incoming values, we are going to combine
// them with the GVN of the incoming bock, and the output location of the
// PHINode to generate a hash value representing this instance of the PHINode.
DenseMap<hash_code, unsigned>::iterator GVNToPHIIt;
DenseMap<unsigned, PHINodeData>::iterator PHIToGVNIt;
std::optional<unsigned> BBGVN = Cand.getGVN(PHIBB);
assert(BBGVN && "Could not find GVN for the incoming block!");
BBGVN = Cand.getCanonicalNum(*BBGVN);
assert(BBGVN && "Could not find canonical number for the incoming block!");
// Create a pair of the exit block canonical value, and the aggregate
// argument location, connected to the canonical numbers stored in the
// PHINode.
PHINodeData TemporaryPair =
std::make_pair(std::make_pair(*BBGVN, AggArgIdx), PHIGVNs);
hash_code PHINodeDataHash = encodePHINodeData(TemporaryPair);
// Look for and create a new entry in our connection between canonical
// numbers for PHINodes, and the set of objects we just created.
GVNToPHIIt = Group.GVNsToPHINodeGVN.find(PHINodeDataHash);
if (GVNToPHIIt == Group.GVNsToPHINodeGVN.end()) {
bool Inserted = false;
std::tie(PHIToGVNIt, Inserted) = Group.PHINodeGVNToGVNs.insert(
std::make_pair(Group.PHINodeGVNTracker, TemporaryPair));
std::tie(GVNToPHIIt, Inserted) = Group.GVNsToPHINodeGVN.insert(
std::make_pair(PHINodeDataHash, Group.PHINodeGVNTracker--));
}
return GVNToPHIIt->second;
}
/// Create a mapping of the output arguments for the \p Region to the output
/// arguments of the overall outlined function.
///
/// \param [in,out] Region - The region of code to be analyzed.
/// \param [in] Outputs - The values found by the code extractor.
static void
findExtractedOutputToOverallOutputMapping(Module &M, OutlinableRegion &Region,
SetVector<Value *> &Outputs) {
OutlinableGroup &Group = *Region.Parent;
IRSimilarityCandidate &C = *Region.Candidate;
SmallVector<BasicBlock *> BE;
DenseSet<BasicBlock *> BlocksInRegion;
C.getBasicBlocks(BlocksInRegion, BE);
// Find the exits to the region.
SmallPtrSet<BasicBlock *, 1> Exits;
for (BasicBlock *Block : BE)
for (BasicBlock *Succ : successors(Block))
if (!BlocksInRegion.contains(Succ))
Exits.insert(Succ);
// After determining which blocks exit to PHINodes, we add these PHINodes to
// the set of outputs to be processed. We also check the incoming values of
// the PHINodes for whether they should no longer be considered outputs.
DenseSet<Value *> OutputsReplacedByPHINode;
DenseSet<Value *> OutputsWithNonPhiUses;
for (BasicBlock *ExitBB : Exits)
analyzeExitPHIsForOutputUses(ExitBB, Exits, BlocksInRegion, Outputs,
OutputsReplacedByPHINode,
OutputsWithNonPhiUses);
// This counts the argument number in the extracted function.
unsigned OriginalIndex = Region.NumExtractedInputs;
// This counts the argument number in the overall function.
unsigned TypeIndex = Group.NumAggregateInputs;
bool TypeFound;
DenseSet<unsigned> AggArgsUsed;
// Iterate over the output types and identify if there is an aggregate pointer
// type whose base type matches the current output type. If there is, we mark
// that we will use this output register for this value. If not we add another
// type to the overall argument type list. We also store the GVNs used for
// stores to identify which values will need to be moved into an special
// block that holds the stores to the output registers.
for (Value *Output : Outputs) {
TypeFound = false;
// We can do this since it is a result value, and will have a number
// that is necessarily the same. BUT if in the future, the instructions
// do not have to be in same order, but are functionally the same, we will
// have to use a different scheme, as one-to-one correspondence is not
// guaranteed.
unsigned ArgumentSize = Group.ArgumentTypes.size();
// If the output is combined in a PHINode, we make sure to skip over it.
if (OutputsReplacedByPHINode.contains(Output))
continue;
unsigned AggArgIdx = 0;
for (unsigned Jdx = TypeIndex; Jdx < ArgumentSize; Jdx++) {
if (!isa<PointerType>(Group.ArgumentTypes[Jdx]))
continue;
if (AggArgsUsed.contains(Jdx))
continue;
TypeFound = true;
AggArgsUsed.insert(Jdx);
Region.ExtractedArgToAgg.insert(std::make_pair(OriginalIndex, Jdx));
Region.AggArgToExtracted.insert(std::make_pair(Jdx, OriginalIndex));
AggArgIdx = Jdx;
break;
}
// We were unable to find an unused type in the output type set that matches
// the output, so we add a pointer type to the argument types of the overall
// function to handle this output and create a mapping to it.
if (!TypeFound) {
Group.ArgumentTypes.push_back(Output->getType()->getPointerTo(
M.getDataLayout().getAllocaAddrSpace()));
// Mark the new pointer type as the last value in the aggregate argument
// list.
unsigned ArgTypeIdx = Group.ArgumentTypes.size() - 1;
AggArgsUsed.insert(ArgTypeIdx);
Region.ExtractedArgToAgg.insert(
std::make_pair(OriginalIndex, ArgTypeIdx));
Region.AggArgToExtracted.insert(
std::make_pair(ArgTypeIdx, OriginalIndex));
AggArgIdx = ArgTypeIdx;
}
// TODO: Adapt to the extra input from the PHINode.
PHINode *PN = dyn_cast<PHINode>(Output);
std::optional<unsigned> GVN;
if (PN && !BlocksInRegion.contains(PN->getParent())) {
// Values outside the region can be combined into PHINode when we
// have multiple exits. We collect both of these into a list to identify
// which values are being used in the PHINode. Each list identifies a
// different PHINode, and a different output. We store the PHINode as it's
// own canonical value. These canonical values are also dependent on the
// output argument it is saved to.
// If two PHINodes have the same canonical values, but different aggregate
// argument locations, then they will have distinct Canonical Values.
GVN = getGVNForPHINode(Region, PN, BlocksInRegion, AggArgIdx);
if (!GVN)
return;
} else {
// If we do not have a PHINode we use the global value numbering for the
// output value, to find the canonical number to add to the set of stored
// values.
GVN = C.getGVN(Output);
GVN = C.getCanonicalNum(*GVN);
}
// Each region has a potentially unique set of outputs. We save which
// values are output in a list of canonical values so we can differentiate
// among the different store schemes.
Region.GVNStores.push_back(*GVN);
OriginalIndex++;
TypeIndex++;
}
// We sort the stored values to make sure that we are not affected by analysis
// order when determining what combination of items were stored.
stable_sort(Region.GVNStores);
}
void IROutliner::findAddInputsOutputs(Module &M, OutlinableRegion &Region,
DenseSet<unsigned> &NotSame) {
std::vector<unsigned> Inputs;
SetVector<Value *> ArgInputs, Outputs;
getCodeExtractorArguments(Region, Inputs, NotSame, OutputMappings, ArgInputs,
Outputs);
if (Region.IgnoreRegion)
return;
// Map the inputs found by the CodeExtractor to the arguments found for
// the overall function.
findExtractedInputToOverallInputMapping(Region, Inputs, ArgInputs);
// Map the outputs found by the CodeExtractor to the arguments found for
// the overall function.
findExtractedOutputToOverallOutputMapping(M, Region, Outputs);
}
/// Replace the extracted function in the Region with a call to the overall
/// function constructed from the deduplicated similar regions, replacing and
/// remapping the values passed to the extracted function as arguments to the
/// new arguments of the overall function.
///
/// \param [in] M - The module to outline from.
/// \param [in] Region - The regions of extracted code to be replaced with a new
/// function.
/// \returns a call instruction with the replaced function.
CallInst *replaceCalledFunction(Module &M, OutlinableRegion &Region) {
std::vector<Value *> NewCallArgs;
DenseMap<unsigned, unsigned>::iterator ArgPair;
OutlinableGroup &Group = *Region.Parent;
CallInst *Call = Region.Call;
assert(Call && "Call to replace is nullptr?");
Function *AggFunc = Group.OutlinedFunction;
assert(AggFunc && "Function to replace with is nullptr?");
// If the arguments are the same size, there are not values that need to be
// made into an argument, the argument ordering has not been change, or
// different output registers to handle. We can simply replace the called
// function in this case.
if (!Region.ChangedArgOrder && AggFunc->arg_size() == Call->arg_size()) {
LLVM_DEBUG(dbgs() << "Replace call to " << *Call << " with call to "
<< *AggFunc << " with same number of arguments\n");
Call->setCalledFunction(AggFunc);
return Call;
}
// We have a different number of arguments than the new function, so
// we need to use our previously mappings off extracted argument to overall
// function argument, and constants to overall function argument to create the
// new argument list.
for (unsigned AggArgIdx = 0; AggArgIdx < AggFunc->arg_size(); AggArgIdx++) {
if (AggArgIdx == AggFunc->arg_size() - 1 &&
Group.OutputGVNCombinations.size() > 1) {
// If we are on the last argument, and we need to differentiate between
// output blocks, add an integer to the argument list to determine
// what block to take
LLVM_DEBUG(dbgs() << "Set switch block argument to "
<< Region.OutputBlockNum << "\n");
NewCallArgs.push_back(ConstantInt::get(Type::getInt32Ty(M.getContext()),
Region.OutputBlockNum));
continue;
}
ArgPair = Region.AggArgToExtracted.find(AggArgIdx);
if (ArgPair != Region.AggArgToExtracted.end()) {
Value *ArgumentValue = Call->getArgOperand(ArgPair->second);
// If we found the mapping from the extracted function to the overall
// function, we simply add it to the argument list. We use the same
// value, it just needs to honor the new order of arguments.
LLVM_DEBUG(dbgs() << "Setting argument " << AggArgIdx << " to value "
<< *ArgumentValue << "\n");
NewCallArgs.push_back(ArgumentValue);
continue;
}
// If it is a constant, we simply add it to the argument list as a value.
if (Region.AggArgToConstant.contains(AggArgIdx)) {
Constant *CST = Region.AggArgToConstant.find(AggArgIdx)->second;
LLVM_DEBUG(dbgs() << "Setting argument " << AggArgIdx << " to value "
<< *CST << "\n");
NewCallArgs.push_back(CST);
continue;
}
// Add a nullptr value if the argument is not found in the extracted
// function. If we cannot find a value, it means it is not in use
// for the region, so we should not pass anything to it.
LLVM_DEBUG(dbgs() << "Setting argument " << AggArgIdx << " to nullptr\n");
NewCallArgs.push_back(ConstantPointerNull::get(
static_cast<PointerType *>(AggFunc->getArg(AggArgIdx)->getType())));
}
LLVM_DEBUG(dbgs() << "Replace call to " << *Call << " with call to "
<< *AggFunc << " with new set of arguments\n");
// Create the new call instruction and erase the old one.
Call = CallInst::Create(AggFunc->getFunctionType(), AggFunc, NewCallArgs, "",
Call);
// It is possible that the call to the outlined function is either the first
// instruction is in the new block, the last instruction, or both. If either
// of these is the case, we need to make sure that we replace the instruction
// in the IRInstructionData struct with the new call.
CallInst *OldCall = Region.Call;
if (Region.NewFront->Inst == OldCall)
Region.NewFront->Inst = Call;
if (Region.NewBack->Inst == OldCall)
Region.NewBack->Inst = Call;
// Transfer any debug information.
Call->setDebugLoc(Region.Call->getDebugLoc());
// Since our output may determine which branch we go to, we make sure to
// propogate this new call value through the module.
OldCall->replaceAllUsesWith(Call);
// Remove the old instruction.
OldCall->eraseFromParent();
Region.Call = Call;
// Make sure that the argument in the new function has the SwiftError
// argument.
if (Group.SwiftErrorArgument)
Call->addParamAttr(*Group.SwiftErrorArgument, Attribute::SwiftError);
return Call;
}
/// Find or create a BasicBlock in the outlined function containing PhiBlocks
/// for \p RetVal.
///
/// \param Group - The OutlinableGroup containing the information about the
/// overall outlined function.
/// \param RetVal - The return value or exit option that we are currently
/// evaluating.
/// \returns The found or newly created BasicBlock to contain the needed
/// PHINodes to be used as outputs.
static BasicBlock *findOrCreatePHIBlock(OutlinableGroup &Group, Value *RetVal) {
DenseMap<Value *, BasicBlock *>::iterator PhiBlockForRetVal,
ReturnBlockForRetVal;
PhiBlockForRetVal = Group.PHIBlocks.find(RetVal);
ReturnBlockForRetVal = Group.EndBBs.find(RetVal);
assert(ReturnBlockForRetVal != Group.EndBBs.end() &&
"Could not find output value!");
BasicBlock *ReturnBB = ReturnBlockForRetVal->second;
// Find if a PHIBlock exists for this return value already. If it is
// the first time we are analyzing this, we will not, so we record it.
PhiBlockForRetVal = Group.PHIBlocks.find(RetVal);
if (PhiBlockForRetVal != Group.PHIBlocks.end())
return PhiBlockForRetVal->second;
// If we did not find a block, we create one, and insert it into the
// overall function and record it.
bool Inserted = false;
BasicBlock *PHIBlock = BasicBlock::Create(ReturnBB->getContext(), "phi_block",
ReturnBB->getParent());
std::tie(PhiBlockForRetVal, Inserted) =
Group.PHIBlocks.insert(std::make_pair(RetVal, PHIBlock));
// We find the predecessors of the return block in the newly created outlined
// function in order to point them to the new PHIBlock rather than the already
// existing return block.
SmallVector<BranchInst *, 2> BranchesToChange;
for (BasicBlock *Pred : predecessors(ReturnBB))
BranchesToChange.push_back(cast<BranchInst>(Pred->getTerminator()));
// Now we mark the branch instructions found, and change the references of the
// return block to the newly created PHIBlock.
for (BranchInst *BI : BranchesToChange)
for (unsigned Succ = 0, End = BI->getNumSuccessors(); Succ < End; Succ++) {
if (BI->getSuccessor(Succ) != ReturnBB)
continue;
BI->setSuccessor(Succ, PHIBlock);
}
BranchInst::Create(ReturnBB, PHIBlock);
return PhiBlockForRetVal->second;
}
/// For the function call now representing the \p Region, find the passed value
/// to that call that represents Argument \p A at the call location if the
/// call has already been replaced with a call to the overall, aggregate
/// function.
///
/// \param A - The Argument to get the passed value for.
/// \param Region - The extracted Region corresponding to the outlined function.
/// \returns The Value representing \p A at the call site.
static Value *
getPassedArgumentInAlreadyOutlinedFunction(const Argument *A,
const OutlinableRegion &Region) {
// If we don't need to adjust the argument number at all (since the call
// has already been replaced by a call to the overall outlined function)
// we can just get the specified argument.
return Region.Call->getArgOperand(A->getArgNo());
}
/// For the function call now representing the \p Region, find the passed value
/// to that call that represents Argument \p A at the call location if the
/// call has only been replaced by the call to the aggregate function.
///
/// \param A - The Argument to get the passed value for.
/// \param Region - The extracted Region corresponding to the outlined function.
/// \returns The Value representing \p A at the call site.
static Value *
getPassedArgumentAndAdjustArgumentLocation(const Argument *A,
const OutlinableRegion &Region) {
unsigned ArgNum = A->getArgNo();
// If it is a constant, we can look at our mapping from when we created
// the outputs to figure out what the constant value is.
if (Region.AggArgToConstant.count(ArgNum))
return Region.AggArgToConstant.find(ArgNum)->second;
// If it is not a constant, and we are not looking at the overall function, we
// need to adjust which argument we are looking at.
ArgNum = Region.AggArgToExtracted.find(ArgNum)->second;
return Region.Call->getArgOperand(ArgNum);
}
/// Find the canonical numbering for the incoming Values into the PHINode \p PN.
///
/// \param PN [in] - The PHINode that we are finding the canonical numbers for.
/// \param Region [in] - The OutlinableRegion containing \p PN.
/// \param OutputMappings [in] - The mapping of output values from outlined
/// region to their original values.
/// \param CanonNums [out] - The canonical numbering for the incoming values to
/// \p PN paired with their incoming block.
/// \param ReplacedWithOutlinedCall - A flag to use the extracted function call
/// of \p Region rather than the overall function's call.
static void findCanonNumsForPHI(
PHINode *PN, OutlinableRegion &Region,
const DenseMap<Value *, Value *> &OutputMappings,
SmallVector<std::pair<unsigned, BasicBlock *>> &CanonNums,
bool ReplacedWithOutlinedCall = true) {
// Iterate over the incoming values.
for (unsigned Idx = 0, EIdx = PN->getNumIncomingValues(); Idx < EIdx; Idx++) {
Value *IVal = PN->getIncomingValue(Idx);
BasicBlock *IBlock = PN->getIncomingBlock(Idx);
// If we have an argument as incoming value, we need to grab the passed
// value from the call itself.
if (Argument *A = dyn_cast<Argument>(IVal)) {
if (ReplacedWithOutlinedCall)
IVal = getPassedArgumentInAlreadyOutlinedFunction(A, Region);
else
IVal = getPassedArgumentAndAdjustArgumentLocation(A, Region);
}
// Get the original value if it has been replaced by an output value.
IVal = findOutputMapping(OutputMappings, IVal);
// Find and add the canonical number for the incoming value.
std::optional<unsigned> GVN = Region.Candidate->getGVN(IVal);
assert(GVN && "No GVN for incoming value");
std::optional<unsigned> CanonNum = Region.Candidate->getCanonicalNum(*GVN);
assert(CanonNum && "No Canonical Number for GVN");
CanonNums.push_back(std::make_pair(*CanonNum, IBlock));
}
}
/// Find, or add PHINode \p PN to the combined PHINode Block \p OverallPHIBlock
/// in order to condense the number of instructions added to the outlined
/// function.
///
/// \param PN [in] - The PHINode that we are finding the canonical numbers for.
/// \param Region [in] - The OutlinableRegion containing \p PN.
/// \param OverallPhiBlock [in] - The overall PHIBlock we are trying to find
/// \p PN in.
/// \param OutputMappings [in] - The mapping of output values from outlined
/// region to their original values.
/// \param UsedPHIs [in, out] - The PHINodes in the block that have already been
/// matched.
/// \return the newly found or created PHINode in \p OverallPhiBlock.
static PHINode*
findOrCreatePHIInBlock(PHINode &PN, OutlinableRegion &Region,
BasicBlock *OverallPhiBlock,
const DenseMap<Value *, Value *> &OutputMappings,
DenseSet<PHINode *> &UsedPHIs) {
OutlinableGroup &Group = *Region.Parent;
// A list of the canonical numbering assigned to each incoming value, paired
// with the incoming block for the PHINode passed into this function.
SmallVector<std::pair<unsigned, BasicBlock *>> PNCanonNums;
// We have to use the extracted function since we have merged this region into
// the overall function yet. We make sure to reassign the argument numbering
// since it is possible that the argument ordering is different between the
// functions.
findCanonNumsForPHI(&PN, Region, OutputMappings, PNCanonNums,
/* ReplacedWithOutlinedCall = */ false);
OutlinableRegion *FirstRegion = Group.Regions[0];
// A list of the canonical numbering assigned to each incoming value, paired
// with the incoming block for the PHINode that we are currently comparing
// the passed PHINode to.
SmallVector<std::pair<unsigned, BasicBlock *>> CurrentCanonNums;
// Find the Canonical Numbering for each PHINode, if it matches, we replace
// the uses of the PHINode we are searching for, with the found PHINode.
for (PHINode &CurrPN : OverallPhiBlock->phis()) {
// If this PHINode has already been matched to another PHINode to be merged,
// we skip it.
if (UsedPHIs.contains(&CurrPN))
continue;
CurrentCanonNums.clear();
findCanonNumsForPHI(&CurrPN, *FirstRegion, OutputMappings, CurrentCanonNums,
/* ReplacedWithOutlinedCall = */ true);
// If the list of incoming values is not the same length, then they cannot
// match since there is not an analogue for each incoming value.
if (PNCanonNums.size() != CurrentCanonNums.size())
continue;
bool FoundMatch = true;
// We compare the canonical value for each incoming value in the passed
// in PHINode to one already present in the outlined region. If the
// incoming values do not match, then the PHINodes do not match.
// We also check to make sure that the incoming block matches as well by
// finding the corresponding incoming block in the combined outlined region
// for the current outlined region.
for (unsigned Idx = 0, Edx = PNCanonNums.size(); Idx < Edx; ++Idx) {
std::pair<unsigned, BasicBlock *> ToCompareTo = CurrentCanonNums[Idx];
std::pair<unsigned, BasicBlock *> ToAdd = PNCanonNums[Idx];
if (ToCompareTo.first != ToAdd.first) {
FoundMatch = false;
break;
}
BasicBlock *CorrespondingBlock =
Region.findCorrespondingBlockIn(*FirstRegion, ToAdd.second);
assert(CorrespondingBlock && "Found block is nullptr");
if (CorrespondingBlock != ToCompareTo.second) {
FoundMatch = false;
break;
}
}
// If all incoming values and branches matched, then we can merge
// into the found PHINode.
if (FoundMatch) {
UsedPHIs.insert(&CurrPN);
return &CurrPN;
}
}
// If we've made it here, it means we weren't able to replace the PHINode, so
// we must insert it ourselves.
PHINode *NewPN = cast<PHINode>(PN.clone());
NewPN->insertBefore(&*OverallPhiBlock->begin());
for (unsigned Idx = 0, Edx = NewPN->getNumIncomingValues(); Idx < Edx;
Idx++) {
Value *IncomingVal = NewPN->getIncomingValue(Idx);
BasicBlock *IncomingBlock = NewPN->getIncomingBlock(Idx);
// Find corresponding basic block in the overall function for the incoming
// block.
BasicBlock *BlockToUse =
Region.findCorrespondingBlockIn(*FirstRegion, IncomingBlock);
NewPN->setIncomingBlock(Idx, BlockToUse);
// If we have an argument we make sure we replace using the argument from
// the correct function.
if (Argument *A = dyn_cast<Argument>(IncomingVal)) {
Value *Val = Group.OutlinedFunction->getArg(A->getArgNo());
NewPN->setIncomingValue(Idx, Val);
continue;
}
// Find the corresponding value in the overall function.
IncomingVal = findOutputMapping(OutputMappings, IncomingVal);
Value *Val = Region.findCorrespondingValueIn(*FirstRegion, IncomingVal);
assert(Val && "Value is nullptr?");
DenseMap<Value *, Value *>::iterator RemappedIt =
FirstRegion->RemappedArguments.find(Val);
if (RemappedIt != FirstRegion->RemappedArguments.end())
Val = RemappedIt->second;
NewPN->setIncomingValue(Idx, Val);
}
return NewPN;
}
// Within an extracted function, replace the argument uses of the extracted
// region with the arguments of the function for an OutlinableGroup.
//
/// \param [in] Region - The region of extracted code to be changed.
/// \param [in,out] OutputBBs - The BasicBlock for the output stores for this
/// region.
/// \param [in] FirstFunction - A flag to indicate whether we are using this
/// function to define the overall outlined function for all the regions, or
/// if we are operating on one of the following regions.
static void
replaceArgumentUses(OutlinableRegion &Region,
DenseMap<Value *, BasicBlock *> &OutputBBs,
const DenseMap<Value *, Value *> &OutputMappings,
bool FirstFunction = false) {
OutlinableGroup &Group = *Region.Parent;
assert(Region.ExtractedFunction && "Region has no extracted function?");
Function *DominatingFunction = Region.ExtractedFunction;
if (FirstFunction)
DominatingFunction = Group.OutlinedFunction;
DominatorTree DT(*DominatingFunction);
DenseSet<PHINode *> UsedPHIs;
for (unsigned ArgIdx = 0; ArgIdx < Region.ExtractedFunction->arg_size();
ArgIdx++) {
assert(Region.ExtractedArgToAgg.contains(ArgIdx) &&
"No mapping from extracted to outlined?");
unsigned AggArgIdx = Region.ExtractedArgToAgg.find(ArgIdx)->second;
Argument *AggArg = Group.OutlinedFunction->getArg(AggArgIdx);
Argument *Arg = Region.ExtractedFunction->getArg(ArgIdx);
// The argument is an input, so we can simply replace it with the overall
// argument value
if (ArgIdx < Region.NumExtractedInputs) {
LLVM_DEBUG(dbgs() << "Replacing uses of input " << *Arg << " in function "
<< *Region.ExtractedFunction << " with " << *AggArg
<< " in function " << *Group.OutlinedFunction << "\n");
Arg->replaceAllUsesWith(AggArg);
Value *V = Region.Call->getArgOperand(ArgIdx);
Region.RemappedArguments.insert(std::make_pair(V, AggArg));
continue;
}
// If we are replacing an output, we place the store value in its own
// block inside the overall function before replacing the use of the output
// in the function.
assert(Arg->hasOneUse() && "Output argument can only have one use");
User *InstAsUser = Arg->user_back();
assert(InstAsUser && "User is nullptr!");
Instruction *I = cast<Instruction>(InstAsUser);
BasicBlock *BB = I->getParent();
SmallVector<BasicBlock *, 4> Descendants;
DT.getDescendants(BB, Descendants);
bool EdgeAdded = false;
if (Descendants.size() == 0) {
EdgeAdded = true;
DT.insertEdge(&DominatingFunction->getEntryBlock(), BB);
DT.getDescendants(BB, Descendants);
}
// Iterate over the following blocks, looking for return instructions,
// if we find one, find the corresponding output block for the return value
// and move our store instruction there.
for (BasicBlock *DescendBB : Descendants) {
ReturnInst *RI = dyn_cast<ReturnInst>(DescendBB->getTerminator());
if (!RI)
continue;
Value *RetVal = RI->getReturnValue();
auto VBBIt = OutputBBs.find(RetVal);
assert(VBBIt != OutputBBs.end() && "Could not find output value!");
// If this is storing a PHINode, we must make sure it is included in the
// overall function.
StoreInst *SI = cast<StoreInst>(I);
Value *ValueOperand = SI->getValueOperand();
StoreInst *NewI = cast<StoreInst>(I->clone());
NewI->setDebugLoc(DebugLoc());
BasicBlock *OutputBB = VBBIt->second;
NewI->insertInto(OutputBB, OutputBB->end());
LLVM_DEBUG(dbgs() << "Move store for instruction " << *I << " to "
<< *OutputBB << "\n");
// If this is storing a PHINode, we must make sure it is included in the
// overall function.
if (!isa<PHINode>(ValueOperand) ||
Region.Candidate->getGVN(ValueOperand).has_value()) {
if (FirstFunction)
continue;
Value *CorrVal =
Region.findCorrespondingValueIn(*Group.Regions[0], ValueOperand);
assert(CorrVal && "Value is nullptr?");
NewI->setOperand(0, CorrVal);
continue;
}
PHINode *PN = cast<PHINode>(SI->getValueOperand());
// If it has a value, it was not split by the code extractor, which
// is what we are looking for.
if (Region.Candidate->getGVN(PN))
continue;
// We record the parent block for the PHINode in the Region so that
// we can exclude it from checks later on.
Region.PHIBlocks.insert(std::make_pair(RetVal, PN->getParent()));
// If this is the first function, we do not need to worry about mergiing
// this with any other block in the overall outlined function, so we can
// just continue.
if (FirstFunction) {
BasicBlock *PHIBlock = PN->getParent();
Group.PHIBlocks.insert(std::make_pair(RetVal, PHIBlock));
continue;
}
// We look for the aggregate block that contains the PHINodes leading into
// this exit path. If we can't find one, we create one.
BasicBlock *OverallPhiBlock = findOrCreatePHIBlock(Group, RetVal);
// For our PHINode, we find the combined canonical numbering, and
// attempt to find a matching PHINode in the overall PHIBlock. If we
// cannot, we copy the PHINode and move it into this new block.
PHINode *NewPN = findOrCreatePHIInBlock(*PN, Region, OverallPhiBlock,
OutputMappings, UsedPHIs);
NewI->setOperand(0, NewPN);
}
// If we added an edge for basic blocks without a predecessor, we remove it
// here.
if (EdgeAdded)
DT.deleteEdge(&DominatingFunction->getEntryBlock(), BB);
I->eraseFromParent();
LLVM_DEBUG(dbgs() << "Replacing uses of output " << *Arg << " in function "
<< *Region.ExtractedFunction << " with " << *AggArg
<< " in function " << *Group.OutlinedFunction << "\n");
Arg->replaceAllUsesWith(AggArg);
}
}
/// Within an extracted function, replace the constants that need to be lifted
/// into arguments with the actual argument.
///
/// \param Region [in] - The region of extracted code to be changed.
void replaceConstants(OutlinableRegion &Region) {
OutlinableGroup &Group = *Region.Parent;
// Iterate over the constants that need to be elevated into arguments
for (std::pair<unsigned, Constant *> &Const : Region.AggArgToConstant) {
unsigned AggArgIdx = Const.first;
Function *OutlinedFunction = Group.OutlinedFunction;
assert(OutlinedFunction && "Overall Function is not defined?");
Constant *CST = Const.second;
Argument *Arg = Group.OutlinedFunction->getArg(AggArgIdx);
// Identify the argument it will be elevated to, and replace instances of
// that constant in the function.
// TODO: If in the future constants do not have one global value number,
// i.e. a constant 1 could be mapped to several values, this check will
// have to be more strict. It cannot be using only replaceUsesWithIf.
LLVM_DEBUG(dbgs() << "Replacing uses of constant " << *CST
<< " in function " << *OutlinedFunction << " with "
<< *Arg << "\n");
CST->replaceUsesWithIf(Arg, [OutlinedFunction](Use &U) {
if (Instruction *I = dyn_cast<Instruction>(U.getUser()))
return I->getFunction() == OutlinedFunction;
return false;
});
}
}
/// It is possible that there is a basic block that already performs the same
/// stores. This returns a duplicate block, if it exists
///
/// \param OutputBBs [in] the blocks we are looking for a duplicate of.
/// \param OutputStoreBBs [in] The existing output blocks.
/// \returns an optional value with the number output block if there is a match.
std::optional<unsigned> findDuplicateOutputBlock(
DenseMap<Value *, BasicBlock *> &OutputBBs,
std::vector<DenseMap<Value *, BasicBlock *>> &OutputStoreBBs) {
bool Mismatch = false;
unsigned MatchingNum = 0;
// We compare the new set output blocks to the other sets of output blocks.
// If they are the same number, and have identical instructions, they are
// considered to be the same.
for (DenseMap<Value *, BasicBlock *> &CompBBs : OutputStoreBBs) {
Mismatch = false;
for (std::pair<Value *, BasicBlock *> &VToB : CompBBs) {
DenseMap<Value *, BasicBlock *>::iterator OutputBBIt =
OutputBBs.find(VToB.first);
if (OutputBBIt == OutputBBs.end()) {
Mismatch = true;
break;
}
BasicBlock *CompBB = VToB.second;
BasicBlock *OutputBB = OutputBBIt->second;
if (CompBB->size() - 1 != OutputBB->size()) {
Mismatch = true;
break;
}
BasicBlock::iterator NIt = OutputBB->begin();
for (Instruction &I : *CompBB) {
if (isa<BranchInst>(&I))
continue;
if (!I.isIdenticalTo(&(*NIt))) {
Mismatch = true;
break;
}
NIt++;
}
}
if (!Mismatch)
return MatchingNum;
MatchingNum++;
}
return std::nullopt;
}
/// Remove empty output blocks from the outlined region.
///
/// \param BlocksToPrune - Mapping of return values output blocks for the \p
/// Region.
/// \param Region - The OutlinableRegion we are analyzing.
static bool
analyzeAndPruneOutputBlocks(DenseMap<Value *, BasicBlock *> &BlocksToPrune,
OutlinableRegion &Region) {
bool AllRemoved = true;
Value *RetValueForBB;
BasicBlock *NewBB;
SmallVector<Value *, 4> ToRemove;
// Iterate over the output blocks created in the outlined section.
for (std::pair<Value *, BasicBlock *> &VtoBB : BlocksToPrune) {
RetValueForBB = VtoBB.first;
NewBB = VtoBB.second;
// If there are no instructions, we remove it from the module, and also
// mark the value for removal from the return value to output block mapping.
if (NewBB->size() == 0) {
NewBB->eraseFromParent();
ToRemove.push_back(RetValueForBB);
continue;
}
// Mark that we could not remove all the blocks since they were not all
// empty.
AllRemoved = false;
}
// Remove the return value from the mapping.
for (Value *V : ToRemove)
BlocksToPrune.erase(V);
// Mark the region as having the no output scheme.
if (AllRemoved)
Region.OutputBlockNum = -1;
return AllRemoved;
}
/// For the outlined section, move needed the StoreInsts for the output
/// registers into their own block. Then, determine if there is a duplicate
/// output block already created.
///
/// \param [in] OG - The OutlinableGroup of regions to be outlined.
/// \param [in] Region - The OutlinableRegion that is being analyzed.
/// \param [in,out] OutputBBs - the blocks that stores for this region will be
/// placed in.
/// \param [in] EndBBs - the final blocks of the extracted function.
/// \param [in] OutputMappings - OutputMappings the mapping of values that have
/// been replaced by a new output value.
/// \param [in,out] OutputStoreBBs - The existing output blocks.
static void alignOutputBlockWithAggFunc(
OutlinableGroup &OG, OutlinableRegion &Region,
DenseMap<Value *, BasicBlock *> &OutputBBs,
DenseMap<Value *, BasicBlock *> &EndBBs,
const DenseMap<Value *, Value *> &OutputMappings,
std::vector<DenseMap<Value *, BasicBlock *>> &OutputStoreBBs) {
// If none of the output blocks have any instructions, this means that we do
// not have to determine if it matches any of the other output schemes, and we
// don't have to do anything else.
if (analyzeAndPruneOutputBlocks(OutputBBs, Region))
return;
// Determine is there is a duplicate set of blocks.
std::optional<unsigned> MatchingBB =
findDuplicateOutputBlock(OutputBBs, OutputStoreBBs);
// If there is, we remove the new output blocks. If it does not,
// we add it to our list of sets of output blocks.
if (MatchingBB) {
LLVM_DEBUG(dbgs() << "Set output block for region in function"
<< Region.ExtractedFunction << " to " << *MatchingBB);
Region.OutputBlockNum = *MatchingBB;
for (std::pair<Value *, BasicBlock *> &VtoBB : OutputBBs)
VtoBB.second->eraseFromParent();
return;
}
Region.OutputBlockNum = OutputStoreBBs.size();
Value *RetValueForBB;
BasicBlock *NewBB;
OutputStoreBBs.push_back(DenseMap<Value *, BasicBlock *>());
for (std::pair<Value *, BasicBlock *> &VtoBB : OutputBBs) {
RetValueForBB = VtoBB.first;
NewBB = VtoBB.second;
DenseMap<Value *, BasicBlock *>::iterator VBBIt =
EndBBs.find(RetValueForBB);
LLVM_DEBUG(dbgs() << "Create output block for region in"
<< Region.ExtractedFunction << " to "
<< *NewBB);
BranchInst::Create(VBBIt->second, NewBB);
OutputStoreBBs.back().insert(std::make_pair(RetValueForBB, NewBB));
}
}
/// Takes in a mapping, \p OldMap of ConstantValues to BasicBlocks, sorts keys,
/// before creating a basic block for each \p NewMap, and inserting into the new
/// block. Each BasicBlock is named with the scheme "<basename>_<key_idx>".
///
/// \param OldMap [in] - The mapping to base the new mapping off of.
/// \param NewMap [out] - The output mapping using the keys of \p OldMap.
/// \param ParentFunc [in] - The function to put the new basic block in.
/// \param BaseName [in] - The start of the BasicBlock names to be appended to
/// by an index value.
static void createAndInsertBasicBlocks(DenseMap<Value *, BasicBlock *> &OldMap,
DenseMap<Value *, BasicBlock *> &NewMap,
Function *ParentFunc, Twine BaseName) {
unsigned Idx = 0;
std::vector<Value *> SortedKeys;
getSortedConstantKeys(SortedKeys, OldMap);
for (Value *RetVal : SortedKeys) {
BasicBlock *NewBB = BasicBlock::Create(
ParentFunc->getContext(),
Twine(BaseName) + Twine("_") + Twine(static_cast<unsigned>(Idx++)),
ParentFunc);
NewMap.insert(std::make_pair(RetVal, NewBB));
}
}
/// Create the switch statement for outlined function to differentiate between
/// all the output blocks.
///
/// For the outlined section, determine if an outlined block already exists that
/// matches the needed stores for the extracted section.
/// \param [in] M - The module we are outlining from.
/// \param [in] OG - The group of regions to be outlined.
/// \param [in] EndBBs - The final blocks of the extracted function.
/// \param [in,out] OutputStoreBBs - The existing output blocks.
void createSwitchStatement(
Module &M, OutlinableGroup &OG, DenseMap<Value *, BasicBlock *> &EndBBs,
std::vector<DenseMap<Value *, BasicBlock *>> &OutputStoreBBs) {
// We only need the switch statement if there is more than one store
// combination, or there is more than one set of output blocks. The first
// will occur when we store different sets of values for two different
// regions. The second will occur when we have two outputs that are combined
// in a PHINode outside of the region in one outlined instance, and are used
// seaparately in another. This will create the same set of OutputGVNs, but
// will generate two different output schemes.
if (OG.OutputGVNCombinations.size() > 1) {
Function *AggFunc = OG.OutlinedFunction;
// Create a final block for each different return block.
DenseMap<Value *, BasicBlock *> ReturnBBs;
createAndInsertBasicBlocks(OG.EndBBs, ReturnBBs, AggFunc, "final_block");
for (std::pair<Value *, BasicBlock *> &RetBlockPair : ReturnBBs) {
std::pair<Value *, BasicBlock *> &OutputBlock =
*OG.EndBBs.find(RetBlockPair.first);
BasicBlock *ReturnBlock = RetBlockPair.second;
BasicBlock *EndBB = OutputBlock.second;
Instruction *Term = EndBB->getTerminator();
// Move the return value to the final block instead of the original exit
// stub.
Term->moveBefore(*ReturnBlock, ReturnBlock->end());
// Put the switch statement in the old end basic block for the function
// with a fall through to the new return block.
LLVM_DEBUG(dbgs() << "Create switch statement in " << *AggFunc << " for "
<< OutputStoreBBs.size() << "\n");
SwitchInst *SwitchI =
SwitchInst::Create(AggFunc->getArg(AggFunc->arg_size() - 1),
ReturnBlock, OutputStoreBBs.size(), EndBB);
unsigned Idx = 0;
for (DenseMap<Value *, BasicBlock *> &OutputStoreBB : OutputStoreBBs) {
DenseMap<Value *, BasicBlock *>::iterator OSBBIt =
OutputStoreBB.find(OutputBlock.first);
if (OSBBIt == OutputStoreBB.end())
continue;
BasicBlock *BB = OSBBIt->second;
SwitchI->addCase(
ConstantInt::get(Type::getInt32Ty(M.getContext()), Idx), BB);
Term = BB->getTerminator();
Term->setSuccessor(0, ReturnBlock);
Idx++;
}
}
return;
}
assert(OutputStoreBBs.size() < 2 && "Different store sets not handled!");
// If there needs to be stores, move them from the output blocks to their
// corresponding ending block. We do not check that the OutputGVNCombinations
// is equal to 1 here since that could just been the case where there are 0
// outputs. Instead, we check whether there is more than one set of output
// blocks since this is the only case where we would have to move the
// stores, and erase the extraneous blocks.
if (OutputStoreBBs.size() == 1) {
LLVM_DEBUG(dbgs() << "Move store instructions to the end block in "
<< *OG.OutlinedFunction << "\n");
DenseMap<Value *, BasicBlock *> OutputBlocks = OutputStoreBBs[0];
for (std::pair<Value *, BasicBlock *> &VBPair : OutputBlocks) {
DenseMap<Value *, BasicBlock *>::iterator EndBBIt =
EndBBs.find(VBPair.first);
assert(EndBBIt != EndBBs.end() && "Could not find end block");
BasicBlock *EndBB = EndBBIt->second;
BasicBlock *OutputBB = VBPair.second;
Instruction *Term = OutputBB->getTerminator();
Term->eraseFromParent();
Term = EndBB->getTerminator();
moveBBContents(*OutputBB, *EndBB);
Term->moveBefore(*EndBB, EndBB->end());
OutputBB->eraseFromParent();
}
}
}
/// Fill the new function that will serve as the replacement function for all of
/// the extracted regions of a certain structure from the first region in the
/// list of regions. Replace this first region's extracted function with the
/// new overall function.
///
/// \param [in] M - The module we are outlining from.
/// \param [in] CurrentGroup - The group of regions to be outlined.
/// \param [in,out] OutputStoreBBs - The output blocks for each different
/// set of stores needed for the different functions.
/// \param [in,out] FuncsToRemove - Extracted functions to erase from module
/// once outlining is complete.
/// \param [in] OutputMappings - Extracted functions to erase from module
/// once outlining is complete.
static void fillOverallFunction(
Module &M, OutlinableGroup &CurrentGroup,
std::vector<DenseMap<Value *, BasicBlock *>> &OutputStoreBBs,
std::vector<Function *> &FuncsToRemove,
const DenseMap<Value *, Value *> &OutputMappings) {
OutlinableRegion *CurrentOS = CurrentGroup.Regions[0];
// Move first extracted function's instructions into new function.
LLVM_DEBUG(dbgs() << "Move instructions from "
<< *CurrentOS->ExtractedFunction << " to instruction "
<< *CurrentGroup.OutlinedFunction << "\n");
moveFunctionData(*CurrentOS->ExtractedFunction,
*CurrentGroup.OutlinedFunction, CurrentGroup.EndBBs);
// Transfer the attributes from the function to the new function.
for (Attribute A : CurrentOS->ExtractedFunction->getAttributes().getFnAttrs())
CurrentGroup.OutlinedFunction->addFnAttr(A);
// Create a new set of output blocks for the first extracted function.
DenseMap<Value *, BasicBlock *> NewBBs;
createAndInsertBasicBlocks(CurrentGroup.EndBBs, NewBBs,
CurrentGroup.OutlinedFunction, "output_block_0");
CurrentOS->OutputBlockNum = 0;
replaceArgumentUses(*CurrentOS, NewBBs, OutputMappings, true);
replaceConstants(*CurrentOS);
// We first identify if any output blocks are empty, if they are we remove
// them. We then create a branch instruction to the basic block to the return
// block for the function for each non empty output block.
if (!analyzeAndPruneOutputBlocks(NewBBs, *CurrentOS)) {
OutputStoreBBs.push_back(DenseMap<Value *, BasicBlock *>());
for (std::pair<Value *, BasicBlock *> &VToBB : NewBBs) {
DenseMap<Value *, BasicBlock *>::iterator VBBIt =
CurrentGroup.EndBBs.find(VToBB.first);
BasicBlock *EndBB = VBBIt->second;
BranchInst::Create(EndBB, VToBB.second);
OutputStoreBBs.back().insert(VToBB);
}
}
// Replace the call to the extracted function with the outlined function.
CurrentOS->Call = replaceCalledFunction(M, *CurrentOS);
// We only delete the extracted functions at the end since we may need to
// reference instructions contained in them for mapping purposes.
FuncsToRemove.push_back(CurrentOS->ExtractedFunction);
}
void IROutliner::deduplicateExtractedSections(
Module &M, OutlinableGroup &CurrentGroup,
std::vector<Function *> &FuncsToRemove, unsigned &OutlinedFunctionNum) {
createFunction(M, CurrentGroup, OutlinedFunctionNum);
std::vector<DenseMap<Value *, BasicBlock *>> OutputStoreBBs;
OutlinableRegion *CurrentOS;
fillOverallFunction(M, CurrentGroup, OutputStoreBBs, FuncsToRemove,
OutputMappings);
std::vector<Value *> SortedKeys;
for (unsigned Idx = 1; Idx < CurrentGroup.Regions.size(); Idx++) {
CurrentOS = CurrentGroup.Regions[Idx];
AttributeFuncs::mergeAttributesForOutlining(*CurrentGroup.OutlinedFunction,
*CurrentOS->ExtractedFunction);
// Create a set of BasicBlocks, one for each return block, to hold the
// needed store instructions.
DenseMap<Value *, BasicBlock *> NewBBs;
createAndInsertBasicBlocks(
CurrentGroup.EndBBs, NewBBs, CurrentGroup.OutlinedFunction,
"output_block_" + Twine(static_cast<unsigned>(Idx)));
replaceArgumentUses(*CurrentOS, NewBBs, OutputMappings);
alignOutputBlockWithAggFunc(CurrentGroup, *CurrentOS, NewBBs,
CurrentGroup.EndBBs, OutputMappings,
OutputStoreBBs);
CurrentOS->Call = replaceCalledFunction(M, *CurrentOS);
FuncsToRemove.push_back(CurrentOS->ExtractedFunction);
}
// Create a switch statement to handle the different output schemes.
createSwitchStatement(M, CurrentGroup, CurrentGroup.EndBBs, OutputStoreBBs);
OutlinedFunctionNum++;
}
/// Checks that the next instruction in the InstructionDataList matches the
/// next instruction in the module. If they do not, there could be the
/// possibility that extra code has been inserted, and we must ignore it.
///
/// \param ID - The IRInstructionData to check the next instruction of.
/// \returns true if the InstructionDataList and actual instruction match.
static bool nextIRInstructionDataMatchesNextInst(IRInstructionData &ID) {
// We check if there is a discrepancy between the InstructionDataList
// and the actual next instruction in the module. If there is, it means
// that an extra instruction was added, likely by the CodeExtractor.
// Since we do not have any similarity data about this particular
// instruction, we cannot confidently outline it, and must discard this
// candidate.
IRInstructionDataList::iterator NextIDIt = std::next(ID.getIterator());
Instruction *NextIDLInst = NextIDIt->Inst;
Instruction *NextModuleInst = nullptr;
if (!ID.Inst->isTerminator())
NextModuleInst = ID.Inst->getNextNonDebugInstruction();
else if (NextIDLInst != nullptr)
NextModuleInst =
&*NextIDIt->Inst->getParent()->instructionsWithoutDebug().begin();
if (NextIDLInst && NextIDLInst != NextModuleInst)
return false;
return true;
}
bool IROutliner::isCompatibleWithAlreadyOutlinedCode(
const OutlinableRegion &Region) {
IRSimilarityCandidate *IRSC = Region.Candidate;
unsigned StartIdx = IRSC->getStartIdx();
unsigned EndIdx = IRSC->getEndIdx();
// A check to make sure that we are not about to attempt to outline something
// that has already been outlined.
for (unsigned Idx = StartIdx; Idx <= EndIdx; Idx++)
if (Outlined.contains(Idx))
return false;
// We check if the recorded instruction matches the actual next instruction,
// if it does not, we fix it in the InstructionDataList.
if (!Region.Candidate->backInstruction()->isTerminator()) {
Instruction *NewEndInst =
Region.Candidate->backInstruction()->getNextNonDebugInstruction();
assert(NewEndInst && "Next instruction is a nullptr?");
if (Region.Candidate->end()->Inst != NewEndInst) {
IRInstructionDataList *IDL = Region.Candidate->front()->IDL;
IRInstructionData *NewEndIRID = new (InstDataAllocator.Allocate())
IRInstructionData(*NewEndInst,
InstructionClassifier.visit(*NewEndInst), *IDL);
// Insert the first IRInstructionData of the new region after the
// last IRInstructionData of the IRSimilarityCandidate.
IDL->insert(Region.Candidate->end(), *NewEndIRID);
}
}
return none_of(*IRSC, [this](IRInstructionData &ID) {
if (!nextIRInstructionDataMatchesNextInst(ID))
return true;
return !this->InstructionClassifier.visit(ID.Inst);
});
}
void IROutliner::pruneIncompatibleRegions(
std::vector<IRSimilarityCandidate> &CandidateVec,
OutlinableGroup &CurrentGroup) {
bool PreviouslyOutlined;
// Sort from beginning to end, so the IRSimilarityCandidates are in order.
stable_sort(CandidateVec, [](const IRSimilarityCandidate &LHS,
const IRSimilarityCandidate &RHS) {
return LHS.getStartIdx() < RHS.getStartIdx();
});
IRSimilarityCandidate &FirstCandidate = CandidateVec[0];
// Since outlining a call and a branch instruction will be the same as only
// outlinining a call instruction, we ignore it as a space saving.
if (FirstCandidate.getLength() == 2) {
if (isa<CallInst>(FirstCandidate.front()->Inst) &&
isa<BranchInst>(FirstCandidate.back()->Inst))
return;
}
unsigned CurrentEndIdx = 0;
for (IRSimilarityCandidate &IRSC : CandidateVec) {
PreviouslyOutlined = false;
unsigned StartIdx = IRSC.getStartIdx();
unsigned EndIdx = IRSC.getEndIdx();
const Function &FnForCurrCand = *IRSC.getFunction();
for (unsigned Idx = StartIdx; Idx <= EndIdx; Idx++)
if (Outlined.contains(Idx)) {
PreviouslyOutlined = true;
break;
}
if (PreviouslyOutlined)
continue;
// Check over the instructions, and if the basic block has its address
// taken for use somewhere else, we do not outline that block.
bool BBHasAddressTaken = any_of(IRSC, [](IRInstructionData &ID){
return ID.Inst->getParent()->hasAddressTaken();
});
if (BBHasAddressTaken)
continue;
if (FnForCurrCand.hasOptNone())
continue;
if (FnForCurrCand.hasFnAttribute("nooutline")) {
LLVM_DEBUG({
dbgs() << "... Skipping function with nooutline attribute: "
<< FnForCurrCand.getName() << "\n";
});
continue;
}
if (IRSC.front()->Inst->getFunction()->hasLinkOnceODRLinkage() &&
!OutlineFromLinkODRs)
continue;
// Greedily prune out any regions that will overlap with already chosen
// regions.
if (CurrentEndIdx != 0 && StartIdx <= CurrentEndIdx)
continue;
bool BadInst = any_of(IRSC, [this](IRInstructionData &ID) {
if (!nextIRInstructionDataMatchesNextInst(ID))
return true;
return !this->InstructionClassifier.visit(ID.Inst);
});
if (BadInst)
continue;
OutlinableRegion *OS = new (RegionAllocator.Allocate())
OutlinableRegion(IRSC, CurrentGroup);
CurrentGroup.Regions.push_back(OS);
CurrentEndIdx = EndIdx;
}
}
InstructionCost
IROutliner::findBenefitFromAllRegions(OutlinableGroup &CurrentGroup) {
InstructionCost RegionBenefit = 0;
for (OutlinableRegion *Region : CurrentGroup.Regions) {
TargetTransformInfo &TTI = getTTI(*Region->StartBB->getParent());
// We add the number of instructions in the region to the benefit as an
// estimate as to how much will be removed.
RegionBenefit += Region->getBenefit(TTI);
LLVM_DEBUG(dbgs() << "Adding: " << RegionBenefit
<< " saved instructions to overfall benefit.\n");
}
return RegionBenefit;
}
/// For the \p OutputCanon number passed in find the value represented by this
/// canonical number. If it is from a PHINode, we pick the first incoming
/// value and return that Value instead.
///
/// \param Region - The OutlinableRegion to get the Value from.
/// \param OutputCanon - The canonical number to find the Value from.
/// \returns The Value represented by a canonical number \p OutputCanon in \p
/// Region.
static Value *findOutputValueInRegion(OutlinableRegion &Region,
unsigned OutputCanon) {
OutlinableGroup &CurrentGroup = *Region.Parent;
// If the value is greater than the value in the tracker, we have a
// PHINode and will instead use one of the incoming values to find the
// type.
if (OutputCanon > CurrentGroup.PHINodeGVNTracker) {
auto It = CurrentGroup.PHINodeGVNToGVNs.find(OutputCanon);
assert(It != CurrentGroup.PHINodeGVNToGVNs.end() &&
"Could not find GVN set for PHINode number!");
assert(It->second.second.size() > 0 && "PHINode does not have any values!");
OutputCanon = *It->second.second.begin();
}
std::optional<unsigned> OGVN =
Region.Candidate->fromCanonicalNum(OutputCanon);
assert(OGVN && "Could not find GVN for Canonical Number?");
std::optional<Value *> OV = Region.Candidate->fromGVN(*OGVN);
assert(OV && "Could not find value for GVN?");
return *OV;
}
InstructionCost
IROutliner::findCostOutputReloads(OutlinableGroup &CurrentGroup) {
InstructionCost OverallCost = 0;
for (OutlinableRegion *Region : CurrentGroup.Regions) {
TargetTransformInfo &TTI = getTTI(*Region->StartBB->getParent());
// Each output incurs a load after the call, so we add that to the cost.
for (unsigned OutputCanon : Region->GVNStores) {
Value *V = findOutputValueInRegion(*Region, OutputCanon);
InstructionCost LoadCost =
TTI.getMemoryOpCost(Instruction::Load, V->getType(), Align(1), 0,
TargetTransformInfo::TCK_CodeSize);
LLVM_DEBUG(dbgs() << "Adding: " << LoadCost
<< " instructions to cost for output of type "
<< *V->getType() << "\n");
OverallCost += LoadCost;
}
}
return OverallCost;
}
/// Find the extra instructions needed to handle any output values for the
/// region.
///
/// \param [in] M - The Module to outline from.
/// \param [in] CurrentGroup - The collection of OutlinableRegions to analyze.
/// \param [in] TTI - The TargetTransformInfo used to collect information for
/// new instruction costs.
/// \returns the additional cost to handle the outputs.
static InstructionCost findCostForOutputBlocks(Module &M,
OutlinableGroup &CurrentGroup,
TargetTransformInfo &TTI) {
InstructionCost OutputCost = 0;
unsigned NumOutputBranches = 0;
OutlinableRegion &FirstRegion = *CurrentGroup.Regions[0];
IRSimilarityCandidate &Candidate = *CurrentGroup.Regions[0]->Candidate;
DenseSet<BasicBlock *> CandidateBlocks;
Candidate.getBasicBlocks(CandidateBlocks);
// Count the number of different output branches that point to blocks outside
// of the region.
DenseSet<BasicBlock *> FoundBlocks;
for (IRInstructionData &ID : Candidate) {
if (!isa<BranchInst>(ID.Inst))
continue;
for (Value *V : ID.OperVals) {
BasicBlock *BB = static_cast<BasicBlock *>(V);
if (!CandidateBlocks.contains(BB) && FoundBlocks.insert(BB).second)
NumOutputBranches++;
}
}
CurrentGroup.BranchesToOutside = NumOutputBranches;
for (const ArrayRef<unsigned> &OutputUse :
CurrentGroup.OutputGVNCombinations) {
for (unsigned OutputCanon : OutputUse) {
Value *V = findOutputValueInRegion(FirstRegion, OutputCanon);
InstructionCost StoreCost =
TTI.getMemoryOpCost(Instruction::Load, V->getType(), Align(1), 0,
TargetTransformInfo::TCK_CodeSize);
// An instruction cost is added for each store set that needs to occur for
// various output combinations inside the function, plus a branch to
// return to the exit block.
LLVM_DEBUG(dbgs() << "Adding: " << StoreCost
<< " instructions to cost for output of type "
<< *V->getType() << "\n");
OutputCost += StoreCost * NumOutputBranches;
}
InstructionCost BranchCost =
TTI.getCFInstrCost(Instruction::Br, TargetTransformInfo::TCK_CodeSize);
LLVM_DEBUG(dbgs() << "Adding " << BranchCost << " to the current cost for"
<< " a branch instruction\n");
OutputCost += BranchCost * NumOutputBranches;
}
// If there is more than one output scheme, we must have a comparison and
// branch for each different item in the switch statement.
if (CurrentGroup.OutputGVNCombinations.size() > 1) {
InstructionCost ComparisonCost = TTI.getCmpSelInstrCost(
Instruction::ICmp, Type::getInt32Ty(M.getContext()),
Type::getInt32Ty(M.getContext()), CmpInst::BAD_ICMP_PREDICATE,
TargetTransformInfo::TCK_CodeSize);
InstructionCost BranchCost =
TTI.getCFInstrCost(Instruction::Br, TargetTransformInfo::TCK_CodeSize);
unsigned DifferentBlocks = CurrentGroup.OutputGVNCombinations.size();
InstructionCost TotalCost = ComparisonCost * BranchCost * DifferentBlocks;
LLVM_DEBUG(dbgs() << "Adding: " << TotalCost
<< " instructions for each switch case for each different"
<< " output path in a function\n");
OutputCost += TotalCost * NumOutputBranches;
}
return OutputCost;
}
void IROutliner::findCostBenefit(Module &M, OutlinableGroup &CurrentGroup) {
InstructionCost RegionBenefit = findBenefitFromAllRegions(CurrentGroup);
CurrentGroup.Benefit += RegionBenefit;
LLVM_DEBUG(dbgs() << "Current Benefit: " << CurrentGroup.Benefit << "\n");
InstructionCost OutputReloadCost = findCostOutputReloads(CurrentGroup);
CurrentGroup.Cost += OutputReloadCost;
LLVM_DEBUG(dbgs() << "Current Cost: " << CurrentGroup.Cost << "\n");
InstructionCost AverageRegionBenefit =
RegionBenefit / CurrentGroup.Regions.size();
unsigned OverallArgumentNum = CurrentGroup.ArgumentTypes.size();
unsigned NumRegions = CurrentGroup.Regions.size();
TargetTransformInfo &TTI =
getTTI(*CurrentGroup.Regions[0]->Candidate->getFunction());
// We add one region to the cost once, to account for the instructions added
// inside of the newly created function.
LLVM_DEBUG(dbgs() << "Adding: " << AverageRegionBenefit
<< " instructions to cost for body of new function.\n");
CurrentGroup.Cost += AverageRegionBenefit;
LLVM_DEBUG(dbgs() << "Current Cost: " << CurrentGroup.Cost << "\n");
// For each argument, we must add an instruction for loading the argument
// out of the register and into a value inside of the newly outlined function.
LLVM_DEBUG(dbgs() << "Adding: " << OverallArgumentNum
<< " instructions to cost for each argument in the new"
<< " function.\n");
CurrentGroup.Cost +=
OverallArgumentNum * TargetTransformInfo::TCC_Basic;
LLVM_DEBUG(dbgs() << "Current Cost: " << CurrentGroup.Cost << "\n");
// Each argument needs to either be loaded into a register or onto the stack.
// Some arguments will only be loaded into the stack once the argument
// registers are filled.
LLVM_DEBUG(dbgs() << "Adding: " << OverallArgumentNum
<< " instructions to cost for each argument in the new"
<< " function " << NumRegions << " times for the "
<< "needed argument handling at the call site.\n");
CurrentGroup.Cost +=
2 * OverallArgumentNum * TargetTransformInfo::TCC_Basic * NumRegions;
LLVM_DEBUG(dbgs() << "Current Cost: " << CurrentGroup.Cost << "\n");
CurrentGroup.Cost += findCostForOutputBlocks(M, CurrentGroup, TTI);
LLVM_DEBUG(dbgs() << "Current Cost: " << CurrentGroup.Cost << "\n");
}
void IROutliner::updateOutputMapping(OutlinableRegion &Region,
ArrayRef<Value *> Outputs,
LoadInst *LI) {
// For and load instructions following the call
Value *Operand = LI->getPointerOperand();
std::optional<unsigned> OutputIdx;
// Find if the operand it is an output register.
for (unsigned ArgIdx = Region.NumExtractedInputs;
ArgIdx < Region.Call->arg_size(); ArgIdx++) {
if (Operand == Region.Call->getArgOperand(ArgIdx)) {
OutputIdx = ArgIdx - Region.NumExtractedInputs;
break;
}
}
// If we found an output register, place a mapping of the new value
// to the original in the mapping.
if (!OutputIdx)
return;
if (!OutputMappings.contains(Outputs[*OutputIdx])) {
LLVM_DEBUG(dbgs() << "Mapping extracted output " << *LI << " to "
<< *Outputs[*OutputIdx] << "\n");
OutputMappings.insert(std::make_pair(LI, Outputs[*OutputIdx]));
} else {
Value *Orig = OutputMappings.find(Outputs[*OutputIdx])->second;
LLVM_DEBUG(dbgs() << "Mapping extracted output " << *Orig << " to "
<< *Outputs[*OutputIdx] << "\n");
OutputMappings.insert(std::make_pair(LI, Orig));
}
}
bool IROutliner::extractSection(OutlinableRegion &Region) {
SetVector<Value *> ArgInputs, Outputs, SinkCands;
assert(Region.StartBB && "StartBB for the OutlinableRegion is nullptr!");
BasicBlock *InitialStart = Region.StartBB;
Function *OrigF = Region.StartBB->getParent();
CodeExtractorAnalysisCache CEAC(*OrigF);
Region.ExtractedFunction =
Region.CE->extractCodeRegion(CEAC, ArgInputs, Outputs);
// If the extraction was successful, find the BasicBlock, and reassign the
// OutlinableRegion blocks
if (!Region.ExtractedFunction) {
LLVM_DEBUG(dbgs() << "CodeExtractor failed to outline " << Region.StartBB
<< "\n");
Region.reattachCandidate();
return false;
}
// Get the block containing the called branch, and reassign the blocks as
// necessary. If the original block still exists, it is because we ended on
// a branch instruction, and so we move the contents into the block before
// and assign the previous block correctly.
User *InstAsUser = Region.ExtractedFunction->user_back();
BasicBlock *RewrittenBB = cast<Instruction>(InstAsUser)->getParent();
Region.PrevBB = RewrittenBB->getSinglePredecessor();
assert(Region.PrevBB && "PrevBB is nullptr?");
if (Region.PrevBB == InitialStart) {
BasicBlock *NewPrev = InitialStart->getSinglePredecessor();
Instruction *BI = NewPrev->getTerminator();
BI->eraseFromParent();
moveBBContents(*InitialStart, *NewPrev);
Region.PrevBB = NewPrev;
InitialStart->eraseFromParent();
}
Region.StartBB = RewrittenBB;
Region.EndBB = RewrittenBB;
// The sequences of outlinable regions has now changed. We must fix the
// IRInstructionDataList for consistency. Although they may not be illegal
// instructions, they should not be compared with anything else as they
// should not be outlined in this round. So marking these as illegal is
// allowed.
IRInstructionDataList *IDL = Region.Candidate->front()->IDL;
Instruction *BeginRewritten = &*RewrittenBB->begin();
Instruction *EndRewritten = &*RewrittenBB->begin();
Region.NewFront = new (InstDataAllocator.Allocate()) IRInstructionData(
*BeginRewritten, InstructionClassifier.visit(*BeginRewritten), *IDL);
Region.NewBack = new (InstDataAllocator.Allocate()) IRInstructionData(
*EndRewritten, InstructionClassifier.visit(*EndRewritten), *IDL);
// Insert the first IRInstructionData of the new region in front of the
// first IRInstructionData of the IRSimilarityCandidate.
IDL->insert(Region.Candidate->begin(), *Region.NewFront);
// Insert the first IRInstructionData of the new region after the
// last IRInstructionData of the IRSimilarityCandidate.
IDL->insert(Region.Candidate->end(), *Region.NewBack);
// Remove the IRInstructionData from the IRSimilarityCandidate.
IDL->erase(Region.Candidate->begin(), std::prev(Region.Candidate->end()));
assert(RewrittenBB != nullptr &&
"Could not find a predecessor after extraction!");
// Iterate over the new set of instructions to find the new call
// instruction.
for (Instruction &I : *RewrittenBB)
if (CallInst *CI = dyn_cast<CallInst>(&I)) {
if (Region.ExtractedFunction == CI->getCalledFunction())
Region.Call = CI;
} else if (LoadInst *LI = dyn_cast<LoadInst>(&I))
updateOutputMapping(Region, Outputs.getArrayRef(), LI);
Region.reattachCandidate();
return true;
}
unsigned IROutliner::doOutline(Module &M) {
// Find the possible similarity sections.
InstructionClassifier.EnableBranches = !DisableBranches;
InstructionClassifier.EnableIndirectCalls = !DisableIndirectCalls;
InstructionClassifier.EnableIntrinsics = !DisableIntrinsics;
IRSimilarityIdentifier &Identifier = getIRSI(M);
SimilarityGroupList &SimilarityCandidates = *Identifier.getSimilarity();
// Sort them by size of extracted sections
unsigned OutlinedFunctionNum = 0;
// If we only have one SimilarityGroup in SimilarityCandidates, we do not have
// to sort them by the potential number of instructions to be outlined
if (SimilarityCandidates.size() > 1)
llvm::stable_sort(SimilarityCandidates,
[](const std::vector<IRSimilarityCandidate> &LHS,
const std::vector<IRSimilarityCandidate> &RHS) {
return LHS[0].getLength() * LHS.size() >
RHS[0].getLength() * RHS.size();
});
// Creating OutlinableGroups for each SimilarityCandidate to be used in
// each of the following for loops to avoid making an allocator.
std::vector<OutlinableGroup> PotentialGroups(SimilarityCandidates.size());
DenseSet<unsigned> NotSame;
std::vector<OutlinableGroup *> NegativeCostGroups;
std::vector<OutlinableRegion *> OutlinedRegions;
// Iterate over the possible sets of similarity.
unsigned PotentialGroupIdx = 0;
for (SimilarityGroup &CandidateVec : SimilarityCandidates) {
OutlinableGroup &CurrentGroup = PotentialGroups[PotentialGroupIdx++];
// Remove entries that were previously outlined
pruneIncompatibleRegions(CandidateVec, CurrentGroup);
// We pruned the number of regions to 0 to 1, meaning that it's not worth
// trying to outlined since there is no compatible similar instance of this
// code.
if (CurrentGroup.Regions.size() < 2)
continue;
// Determine if there are any values that are the same constant throughout
// each section in the set.
NotSame.clear();
CurrentGroup.findSameConstants(NotSame);
if (CurrentGroup.IgnoreGroup)
continue;
// Create a CodeExtractor for each outlinable region. Identify inputs and
// outputs for each section using the code extractor and create the argument
// types for the Aggregate Outlining Function.
OutlinedRegions.clear();
for (OutlinableRegion *OS : CurrentGroup.Regions) {
// Break the outlinable region out of its parent BasicBlock into its own
// BasicBlocks (see function implementation).
OS->splitCandidate();
// There's a chance that when the region is split, extra instructions are
// added to the region. This makes the region no longer viable
// to be split, so we ignore it for outlining.
if (!OS->CandidateSplit)
continue;
SmallVector<BasicBlock *> BE;
DenseSet<BasicBlock *> BlocksInRegion;
OS->Candidate->getBasicBlocks(BlocksInRegion, BE);
OS->CE = new (ExtractorAllocator.Allocate())
CodeExtractor(BE, nullptr, false, nullptr, nullptr, nullptr, false,
false, nullptr, "outlined");
findAddInputsOutputs(M, *OS, NotSame);
if (!OS->IgnoreRegion)
OutlinedRegions.push_back(OS);
// We recombine the blocks together now that we have gathered all the
// needed information.
OS->reattachCandidate();
}
CurrentGroup.Regions = std::move(OutlinedRegions);
if (CurrentGroup.Regions.empty())
continue;
CurrentGroup.collectGVNStoreSets(M);
if (CostModel)
findCostBenefit(M, CurrentGroup);
// If we are adhering to the cost model, skip those groups where the cost
// outweighs the benefits.
if (CurrentGroup.Cost >= CurrentGroup.Benefit && CostModel) {
OptimizationRemarkEmitter &ORE =
getORE(*CurrentGroup.Regions[0]->Candidate->getFunction());
ORE.emit([&]() {
IRSimilarityCandidate *C = CurrentGroup.Regions[0]->Candidate;
OptimizationRemarkMissed R(DEBUG_TYPE, "WouldNotDecreaseSize",
C->frontInstruction());
R << "did not outline "
<< ore::NV(std::to_string(CurrentGroup.Regions.size()))
<< " regions due to estimated increase of "
<< ore::NV("InstructionIncrease",
CurrentGroup.Cost - CurrentGroup.Benefit)
<< " instructions at locations ";
interleave(
CurrentGroup.Regions.begin(), CurrentGroup.Regions.end(),
[&R](OutlinableRegion *Region) {
R << ore::NV(
"DebugLoc",
Region->Candidate->frontInstruction()->getDebugLoc());
},
[&R]() { R << " "; });
return R;
});
continue;
}
NegativeCostGroups.push_back(&CurrentGroup);
}
ExtractorAllocator.DestroyAll();
if (NegativeCostGroups.size() > 1)
stable_sort(NegativeCostGroups,
[](const OutlinableGroup *LHS, const OutlinableGroup *RHS) {
return LHS->Benefit - LHS->Cost > RHS->Benefit - RHS->Cost;
});
std::vector<Function *> FuncsToRemove;
for (OutlinableGroup *CG : NegativeCostGroups) {
OutlinableGroup &CurrentGroup = *CG;
OutlinedRegions.clear();
for (OutlinableRegion *Region : CurrentGroup.Regions) {
// We check whether our region is compatible with what has already been
// outlined, and whether we need to ignore this item.
if (!isCompatibleWithAlreadyOutlinedCode(*Region))
continue;
OutlinedRegions.push_back(Region);
}
if (OutlinedRegions.size() < 2)
continue;
// Reestimate the cost and benefit of the OutlinableGroup. Continue only if
// we are still outlining enough regions to make up for the added cost.
CurrentGroup.Regions = std::move(OutlinedRegions);
if (CostModel) {
CurrentGroup.Benefit = 0;
CurrentGroup.Cost = 0;
findCostBenefit(M, CurrentGroup);
if (CurrentGroup.Cost >= CurrentGroup.Benefit)
continue;
}
OutlinedRegions.clear();
for (OutlinableRegion *Region : CurrentGroup.Regions) {
Region->splitCandidate();
if (!Region->CandidateSplit)
continue;
OutlinedRegions.push_back(Region);
}
CurrentGroup.Regions = std::move(OutlinedRegions);
if (CurrentGroup.Regions.size() < 2) {
for (OutlinableRegion *R : CurrentGroup.Regions)
R->reattachCandidate();
continue;
}
LLVM_DEBUG(dbgs() << "Outlining regions with cost " << CurrentGroup.Cost
<< " and benefit " << CurrentGroup.Benefit << "\n");
// Create functions out of all the sections, and mark them as outlined.
OutlinedRegions.clear();
for (OutlinableRegion *OS : CurrentGroup.Regions) {
SmallVector<BasicBlock *> BE;
DenseSet<BasicBlock *> BlocksInRegion;
OS->Candidate->getBasicBlocks(BlocksInRegion, BE);
OS->CE = new (ExtractorAllocator.Allocate())
CodeExtractor(BE, nullptr, false, nullptr, nullptr, nullptr, false,
false, nullptr, "outlined");
bool FunctionOutlined = extractSection(*OS);
if (FunctionOutlined) {
unsigned StartIdx = OS->Candidate->getStartIdx();
unsigned EndIdx = OS->Candidate->getEndIdx();
for (unsigned Idx = StartIdx; Idx <= EndIdx; Idx++)
Outlined.insert(Idx);
OutlinedRegions.push_back(OS);
}
}
LLVM_DEBUG(dbgs() << "Outlined " << OutlinedRegions.size()
<< " with benefit " << CurrentGroup.Benefit
<< " and cost " << CurrentGroup.Cost << "\n");
CurrentGroup.Regions = std::move(OutlinedRegions);
if (CurrentGroup.Regions.empty())
continue;
OptimizationRemarkEmitter &ORE =
getORE(*CurrentGroup.Regions[0]->Call->getFunction());
ORE.emit([&]() {
IRSimilarityCandidate *C = CurrentGroup.Regions[0]->Candidate;
OptimizationRemark R(DEBUG_TYPE, "Outlined", C->front()->Inst);
R << "outlined " << ore::NV(std::to_string(CurrentGroup.Regions.size()))
<< " regions with decrease of "
<< ore::NV("Benefit", CurrentGroup.Benefit - CurrentGroup.Cost)
<< " instructions at locations ";
interleave(
CurrentGroup.Regions.begin(), CurrentGroup.Regions.end(),
[&R](OutlinableRegion *Region) {
R << ore::NV("DebugLoc",
Region->Candidate->frontInstruction()->getDebugLoc());
},
[&R]() { R << " "; });
return R;
});
deduplicateExtractedSections(M, CurrentGroup, FuncsToRemove,
OutlinedFunctionNum);
}
for (Function *F : FuncsToRemove)
F->eraseFromParent();
return OutlinedFunctionNum;
}
bool IROutliner::run(Module &M) {
CostModel = !NoCostModel;
OutlineFromLinkODRs = EnableLinkOnceODRIROutlining;
return doOutline(M) > 0;
}
PreservedAnalyses IROutlinerPass::run(Module &M, ModuleAnalysisManager &AM) {
auto &FAM = AM.getResult<FunctionAnalysisManagerModuleProxy>(M).getManager();
std::function<TargetTransformInfo &(Function &)> GTTI =
[&FAM](Function &F) -> TargetTransformInfo & {
return FAM.getResult<TargetIRAnalysis>(F);
};
std::function<IRSimilarityIdentifier &(Module &)> GIRSI =
[&AM](Module &M) -> IRSimilarityIdentifier & {
return AM.getResult<IRSimilarityAnalysis>(M);
};
std::unique_ptr<OptimizationRemarkEmitter> ORE;
std::function<OptimizationRemarkEmitter &(Function &)> GORE =
[&ORE](Function &F) -> OptimizationRemarkEmitter & {
ORE.reset(new OptimizationRemarkEmitter(&F));
return *ORE;
};
if (IROutliner(GTTI, GIRSI, GORE).run(M))
return PreservedAnalyses::none();
return PreservedAnalyses::all();
}