maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/lib/Analysis/MemorySSA.cpp

2486 lines
91 KiB
C++
Raw Blame History

//===- MemorySSA.cpp - Memory SSA Builder ---------------------------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file implements the MemorySSA class.
//
//===----------------------------------------------------------------------===//
#include "llvm/Analysis/MemorySSA.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/DenseMapInfo.h"
#include "llvm/ADT/DenseSet.h"
#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Hashing.h"
#include "llvm/ADT/None.h"
#include "llvm/ADT/Optional.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/iterator.h"
#include "llvm/ADT/iterator_range.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/IteratedDominanceFrontier.h"
#include "llvm/Analysis/MemoryLocation.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/IR/AssemblyAnnotationWriter.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Instruction.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/Use.h"
#include "llvm/InitializePasses.h"
#include "llvm/Pass.h"
#include "llvm/Support/AtomicOrdering.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/FormattedStream.h"
#include "llvm/Support/raw_ostream.h"
#include <algorithm>
#include <cassert>
#include <cstdlib>
#include <iterator>
#include <memory>
#include <utility>
using namespace llvm;
#define DEBUG_TYPE "memoryssa"
INITIALIZE_PASS_BEGIN(MemorySSAWrapperPass, "memoryssa", "Memory SSA", false,
true)
INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
INITIALIZE_PASS_END(MemorySSAWrapperPass, "memoryssa", "Memory SSA", false,
true)
INITIALIZE_PASS_BEGIN(MemorySSAPrinterLegacyPass, "print-memoryssa",
"Memory SSA Printer", false, false)
INITIALIZE_PASS_DEPENDENCY(MemorySSAWrapperPass)
INITIALIZE_PASS_END(MemorySSAPrinterLegacyPass, "print-memoryssa",
"Memory SSA Printer", false, false)
static cl::opt<unsigned> MaxCheckLimit(
"memssa-check-limit", cl::Hidden, cl::init(100),
cl::desc("The maximum number of stores/phis MemorySSA"
"will consider trying to walk past (default = 100)"));
// Always verify MemorySSA if expensive checking is enabled.
#ifdef EXPENSIVE_CHECKS
bool llvm::VerifyMemorySSA = true;
#else
bool llvm::VerifyMemorySSA = false;
#endif
/// Enables memory ssa as a dependency for loop passes in legacy pass manager.
cl::opt<bool> llvm::EnableMSSALoopDependency(
"enable-mssa-loop-dependency", cl::Hidden, cl::init(true),
cl::desc("Enable MemorySSA dependency for loop pass manager"));
static cl::opt<bool, true>
VerifyMemorySSAX("verify-memoryssa", cl::location(VerifyMemorySSA),
cl::Hidden, cl::desc("Enable verification of MemorySSA."));
namespace llvm {
/// An assembly annotator class to print Memory SSA information in
/// comments.
class MemorySSAAnnotatedWriter : public AssemblyAnnotationWriter {
friend class MemorySSA;
const MemorySSA *MSSA;
public:
MemorySSAAnnotatedWriter(const MemorySSA *M) : MSSA(M) {}
void emitBasicBlockStartAnnot(const BasicBlock *BB,
formatted_raw_ostream &OS) override {
if (MemoryAccess *MA = MSSA->getMemoryAccess(BB))
OS << "; " << *MA << "\n";
}
void emitInstructionAnnot(const Instruction *I,
formatted_raw_ostream &OS) override {
if (MemoryAccess *MA = MSSA->getMemoryAccess(I))
OS << "; " << *MA << "\n";
}
};
} // end namespace llvm
namespace {
/// Our current alias analysis API differentiates heavily between calls and
/// non-calls, and functions called on one usually assert on the other.
/// This class encapsulates the distinction to simplify other code that wants
/// "Memory affecting instructions and related data" to use as a key.
/// For example, this class is used as a densemap key in the use optimizer.
class MemoryLocOrCall {
public:
bool IsCall = false;
MemoryLocOrCall(MemoryUseOrDef *MUD)
: MemoryLocOrCall(MUD->getMemoryInst()) {}
MemoryLocOrCall(const MemoryUseOrDef *MUD)
: MemoryLocOrCall(MUD->getMemoryInst()) {}
MemoryLocOrCall(Instruction *Inst) {
if (auto *C = dyn_cast<CallBase>(Inst)) {
IsCall = true;
Call = C;
} else {
IsCall = false;
// There is no such thing as a memorylocation for a fence inst, and it is
// unique in that regard.
if (!isa<FenceInst>(Inst))
Loc = MemoryLocation::get(Inst);
}
}
explicit MemoryLocOrCall(const MemoryLocation &Loc) : Loc(Loc) {}
const CallBase *getCall() const {
assert(IsCall);
return Call;
}
MemoryLocation getLoc() const {
assert(!IsCall);
return Loc;
}
bool operator==(const MemoryLocOrCall &Other) const {
if (IsCall != Other.IsCall)
return false;
if (!IsCall)
return Loc == Other.Loc;
if (Call->getCalledOperand() != Other.Call->getCalledOperand())
return false;
return Call->arg_size() == Other.Call->arg_size() &&
std::equal(Call->arg_begin(), Call->arg_end(),
Other.Call->arg_begin());
}
private:
union {
const CallBase *Call;
MemoryLocation Loc;
};
};
} // end anonymous namespace
namespace llvm {
template <> struct DenseMapInfo<MemoryLocOrCall> {
static inline MemoryLocOrCall getEmptyKey() {
return MemoryLocOrCall(DenseMapInfo<MemoryLocation>::getEmptyKey());
}
static inline MemoryLocOrCall getTombstoneKey() {
return MemoryLocOrCall(DenseMapInfo<MemoryLocation>::getTombstoneKey());
}
static unsigned getHashValue(const MemoryLocOrCall &MLOC) {
if (!MLOC.IsCall)
return hash_combine(
MLOC.IsCall,
DenseMapInfo<MemoryLocation>::getHashValue(MLOC.getLoc()));
hash_code hash =
hash_combine(MLOC.IsCall, DenseMapInfo<const Value *>::getHashValue(
MLOC.getCall()->getCalledOperand()));
for (const Value *Arg : MLOC.getCall()->args())
hash = hash_combine(hash, DenseMapInfo<const Value *>::getHashValue(Arg));
return hash;
}
static bool isEqual(const MemoryLocOrCall &LHS, const MemoryLocOrCall &RHS) {
return LHS == RHS;
}
};
} // end namespace llvm
/// This does one-way checks to see if Use could theoretically be hoisted above
/// MayClobber. This will not check the other way around.
///
/// This assumes that, for the purposes of MemorySSA, Use comes directly after
/// MayClobber, with no potentially clobbering operations in between them.
/// (Where potentially clobbering ops are memory barriers, aliased stores, etc.)
static bool areLoadsReorderable(const LoadInst *Use,
const LoadInst *MayClobber) {
bool VolatileUse = Use->isVolatile();
bool VolatileClobber = MayClobber->isVolatile();
// Volatile operations may never be reordered with other volatile operations.
if (VolatileUse && VolatileClobber)
return false;
// Otherwise, volatile doesn't matter here. From the language reference:
// 'optimizers may change the order of volatile operations relative to
// non-volatile operations.'"
// If a load is seq_cst, it cannot be moved above other loads. If its ordering
// is weaker, it can be moved above other loads. We just need to be sure that
// MayClobber isn't an acquire load, because loads can't be moved above
// acquire loads.
//
// Note that this explicitly *does* allow the free reordering of monotonic (or
// weaker) loads of the same address.
bool SeqCstUse = Use->getOrdering() == AtomicOrdering::SequentiallyConsistent;
bool MayClobberIsAcquire = isAtLeastOrStrongerThan(MayClobber->getOrdering(),
AtomicOrdering::Acquire);
return !(SeqCstUse || MayClobberIsAcquire);
}
namespace {
struct ClobberAlias {
bool IsClobber;
Optional<AliasResult> AR;
};
} // end anonymous namespace
// Return a pair of {IsClobber (bool), AR (AliasResult)}. It relies on AR being
// ignored if IsClobber = false.
template <typename AliasAnalysisType>
static ClobberAlias
instructionClobbersQuery(const MemoryDef *MD, const MemoryLocation &UseLoc,
const Instruction *UseInst, AliasAnalysisType &AA) {
Instruction *DefInst = MD->getMemoryInst();
assert(DefInst && "Defining instruction not actually an instruction");
const auto *UseCall = dyn_cast<CallBase>(UseInst);
Optional<AliasResult> AR;
if (const IntrinsicInst *II = dyn_cast<IntrinsicInst>(DefInst)) {
// These intrinsics will show up as affecting memory, but they are just
// markers, mostly.
//
// FIXME: We probably don't actually want MemorySSA to model these at all
// (including creating MemoryAccesses for them): we just end up inventing
// clobbers where they don't really exist at all. Please see D43269 for
// context.
switch (II->getIntrinsicID()) {
case Intrinsic::lifetime_start:
if (UseCall)
return {false, NoAlias};
AR = AA.alias(MemoryLocation(II->getArgOperand(1)), UseLoc);
return {AR != NoAlias, AR};
case Intrinsic::lifetime_end:
case Intrinsic::invariant_start:
case Intrinsic::invariant_end:
case Intrinsic::assume:
return {false, NoAlias};
case Intrinsic::dbg_addr:
case Intrinsic::dbg_declare:
case Intrinsic::dbg_label:
case Intrinsic::dbg_value:
llvm_unreachable("debuginfo shouldn't have associated defs!");
default:
break;
}
}
if (UseCall) {
ModRefInfo I = AA.getModRefInfo(DefInst, UseCall);
AR = isMustSet(I) ? MustAlias : MayAlias;
return {isModOrRefSet(I), AR};
}
if (auto *DefLoad = dyn_cast<LoadInst>(DefInst))
if (auto *UseLoad = dyn_cast<LoadInst>(UseInst))
return {!areLoadsReorderable(UseLoad, DefLoad), MayAlias};
ModRefInfo I = AA.getModRefInfo(DefInst, UseLoc);
AR = isMustSet(I) ? MustAlias : MayAlias;
return {isModSet(I), AR};
}
template <typename AliasAnalysisType>
static ClobberAlias instructionClobbersQuery(MemoryDef *MD,
const MemoryUseOrDef *MU,
const MemoryLocOrCall &UseMLOC,
AliasAnalysisType &AA) {
// FIXME: This is a temporary hack to allow a single instructionClobbersQuery
// to exist while MemoryLocOrCall is pushed through places.
if (UseMLOC.IsCall)
return instructionClobbersQuery(MD, MemoryLocation(), MU->getMemoryInst(),
AA);
return instructionClobbersQuery(MD, UseMLOC.getLoc(), MU->getMemoryInst(),
AA);
}
// Return true when MD may alias MU, return false otherwise.
bool MemorySSAUtil::defClobbersUseOrDef(MemoryDef *MD, const MemoryUseOrDef *MU,
AliasAnalysis &AA) {
return instructionClobbersQuery(MD, MU, MemoryLocOrCall(MU), AA).IsClobber;
}
namespace {
struct UpwardsMemoryQuery {
// True if our original query started off as a call
bool IsCall = false;
// The pointer location we started the query with. This will be empty if
// IsCall is true.
MemoryLocation StartingLoc;
// This is the instruction we were querying about.
const Instruction *Inst = nullptr;
// The MemoryAccess we actually got called with, used to test local domination
const MemoryAccess *OriginalAccess = nullptr;
Optional<AliasResult> AR = MayAlias;
bool SkipSelfAccess = false;
UpwardsMemoryQuery() = default;
UpwardsMemoryQuery(const Instruction *Inst, const MemoryAccess *Access)
: IsCall(isa<CallBase>(Inst)), Inst(Inst), OriginalAccess(Access) {
if (!IsCall)
StartingLoc = MemoryLocation::get(Inst);
}
};
} // end anonymous namespace
static bool lifetimeEndsAt(MemoryDef *MD, const MemoryLocation &Loc,
BatchAAResults &AA) {
Instruction *Inst = MD->getMemoryInst();
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
switch (II->getIntrinsicID()) {
case Intrinsic::lifetime_end:
return AA.alias(MemoryLocation(II->getArgOperand(1)), Loc) == MustAlias;
default:
return false;
}
}
return false;
}
template <typename AliasAnalysisType>
static bool isUseTriviallyOptimizableToLiveOnEntry(AliasAnalysisType &AA,
const Instruction *I) {
// If the memory can't be changed, then loads of the memory can't be
// clobbered.
return isa<LoadInst>(I) && (I->hasMetadata(LLVMContext::MD_invariant_load) ||
AA.pointsToConstantMemory(MemoryLocation(
cast<LoadInst>(I)->getPointerOperand())));
}
/// Verifies that `Start` is clobbered by `ClobberAt`, and that nothing
/// inbetween `Start` and `ClobberAt` can clobbers `Start`.
///
/// This is meant to be as simple and self-contained as possible. Because it
/// uses no cache, etc., it can be relatively expensive.
///
/// \param Start The MemoryAccess that we want to walk from.
/// \param ClobberAt A clobber for Start.
/// \param StartLoc The MemoryLocation for Start.
/// \param MSSA The MemorySSA instance that Start and ClobberAt belong to.
/// \param Query The UpwardsMemoryQuery we used for our search.
/// \param AA The AliasAnalysis we used for our search.
/// \param AllowImpreciseClobber Always false, unless we do relaxed verify.
template <typename AliasAnalysisType>
LLVM_ATTRIBUTE_UNUSED static void
checkClobberSanity(const MemoryAccess *Start, MemoryAccess *ClobberAt,
const MemoryLocation &StartLoc, const MemorySSA &MSSA,
const UpwardsMemoryQuery &Query, AliasAnalysisType &AA,
bool AllowImpreciseClobber = false) {
assert(MSSA.dominates(ClobberAt, Start) && "Clobber doesn't dominate start?");
if (MSSA.isLiveOnEntryDef(Start)) {
assert(MSSA.isLiveOnEntryDef(ClobberAt) &&
"liveOnEntry must clobber itself");
return;
}
bool FoundClobber = false;
DenseSet<ConstMemoryAccessPair> VisitedPhis;
SmallVector<ConstMemoryAccessPair, 8> Worklist;
Worklist.emplace_back(Start, StartLoc);
// Walk all paths from Start to ClobberAt, while looking for clobbers. If one
// is found, complain.
while (!Worklist.empty()) {
auto MAP = Worklist.pop_back_val();
// All we care about is that nothing from Start to ClobberAt clobbers Start.
// We learn nothing from revisiting nodes.
if (!VisitedPhis.insert(MAP).second)
continue;
for (const auto *MA : def_chain(MAP.first)) {
if (MA == ClobberAt) {
if (const auto *MD = dyn_cast<MemoryDef>(MA)) {
// instructionClobbersQuery isn't essentially free, so don't use `|=`,
// since it won't let us short-circuit.
//
// Also, note that this can't be hoisted out of the `Worklist` loop,
// since MD may only act as a clobber for 1 of N MemoryLocations.
FoundClobber = FoundClobber || MSSA.isLiveOnEntryDef(MD);
if (!FoundClobber) {
ClobberAlias CA =
instructionClobbersQuery(MD, MAP.second, Query.Inst, AA);
if (CA.IsClobber) {
FoundClobber = true;
// Not used: CA.AR;
}
}
}
break;
}
// We should never hit liveOnEntry, unless it's the clobber.
assert(!MSSA.isLiveOnEntryDef(MA) && "Hit liveOnEntry before clobber?");
if (const auto *MD = dyn_cast<MemoryDef>(MA)) {
// If Start is a Def, skip self.
if (MD == Start)
continue;
assert(!instructionClobbersQuery(MD, MAP.second, Query.Inst, AA)
.IsClobber &&
"Found clobber before reaching ClobberAt!");
continue;
}
if (const auto *MU = dyn_cast<MemoryUse>(MA)) {
(void)MU;
assert (MU == Start &&
"Can only find use in def chain if Start is a use");
continue;
}
assert(isa<MemoryPhi>(MA));
// Add reachable phi predecessors
for (auto ItB = upward_defs_begin(
{const_cast<MemoryAccess *>(MA), MAP.second},
MSSA.getDomTree()),
ItE = upward_defs_end();
ItB != ItE; ++ItB)
if (MSSA.getDomTree().isReachableFromEntry(ItB.getPhiArgBlock()))
Worklist.emplace_back(*ItB);
}
}
// If the verify is done following an optimization, it's possible that
// ClobberAt was a conservative clobbering, that we can now infer is not a
// true clobbering access. Don't fail the verify if that's the case.
// We do have accesses that claim they're optimized, but could be optimized
// further. Updating all these can be expensive, so allow it for now (FIXME).
if (AllowImpreciseClobber)
return;
// If ClobberAt is a MemoryPhi, we can assume something above it acted as a
// clobber. Otherwise, `ClobberAt` should've acted as a clobber at some point.
assert((isa<MemoryPhi>(ClobberAt) || FoundClobber) &&
"ClobberAt never acted as a clobber");
}
namespace {
/// Our algorithm for walking (and trying to optimize) clobbers, all wrapped up
/// in one class.
template <class AliasAnalysisType> class ClobberWalker {
/// Save a few bytes by using unsigned instead of size_t.
using ListIndex = unsigned;
/// Represents a span of contiguous MemoryDefs, potentially ending in a
/// MemoryPhi.
struct DefPath {
MemoryLocation Loc;
// Note that, because we always walk in reverse, Last will always dominate
// First. Also note that First and Last are inclusive.
MemoryAccess *First;
MemoryAccess *Last;
Optional<ListIndex> Previous;
DefPath(const MemoryLocation &Loc, MemoryAccess *First, MemoryAccess *Last,
Optional<ListIndex> Previous)
: Loc(Loc), First(First), Last(Last), Previous(Previous) {}
DefPath(const MemoryLocation &Loc, MemoryAccess *Init,
Optional<ListIndex> Previous)
: DefPath(Loc, Init, Init, Previous) {}
};
const MemorySSA &MSSA;
AliasAnalysisType &AA;
DominatorTree &DT;
UpwardsMemoryQuery *Query;
unsigned *UpwardWalkLimit;
// Phi optimization bookkeeping:
// List of DefPath to process during the current phi optimization walk.
SmallVector<DefPath, 32> Paths;
// List of visited <Access, Location> pairs; we can skip paths already
// visited with the same memory location.
DenseSet<ConstMemoryAccessPair> VisitedPhis;
// Record if phi translation has been performed during the current phi
// optimization walk, as merging alias results after phi translation can
// yield incorrect results. Context in PR46156.
bool PerformedPhiTranslation = false;
/// Find the nearest def or phi that `From` can legally be optimized to.
const MemoryAccess *getWalkTarget(const MemoryPhi *From) const {
assert(From->getNumOperands() && "Phi with no operands?");
BasicBlock *BB = From->getBlock();
MemoryAccess *Result = MSSA.getLiveOnEntryDef();
DomTreeNode *Node = DT.getNode(BB);
while ((Node = Node->getIDom())) {
auto *Defs = MSSA.getBlockDefs(Node->getBlock());
if (Defs)
return &*Defs->rbegin();
}
return Result;
}
/// Result of calling walkToPhiOrClobber.
struct UpwardsWalkResult {
/// The "Result" of the walk. Either a clobber, the last thing we walked, or
/// both. Include alias info when clobber found.
MemoryAccess *Result;
bool IsKnownClobber;
Optional<AliasResult> AR;
};
/// Walk to the next Phi or Clobber in the def chain starting at Desc.Last.
/// This will update Desc.Last as it walks. It will (optionally) also stop at
/// StopAt.
///
/// This does not test for whether StopAt is a clobber
UpwardsWalkResult
walkToPhiOrClobber(DefPath &Desc, const MemoryAccess *StopAt = nullptr,
const MemoryAccess *SkipStopAt = nullptr) const {
assert(!isa<MemoryUse>(Desc.Last) && "Uses don't exist in my world");
assert(UpwardWalkLimit && "Need a valid walk limit");
bool LimitAlreadyReached = false;
// (*UpwardWalkLimit) may be 0 here, due to the loop in tryOptimizePhi. Set
// it to 1. This will not do any alias() calls. It either returns in the
// first iteration in the loop below, or is set back to 0 if all def chains
// are free of MemoryDefs.
if (!*UpwardWalkLimit) {
*UpwardWalkLimit = 1;
LimitAlreadyReached = true;
}
for (MemoryAccess *Current : def_chain(Desc.Last)) {
Desc.Last = Current;
if (Current == StopAt || Current == SkipStopAt)
return {Current, false, MayAlias};
if (auto *MD = dyn_cast<MemoryDef>(Current)) {
if (MSSA.isLiveOnEntryDef(MD))
return {MD, true, MustAlias};
if (!--*UpwardWalkLimit)
return {Current, true, MayAlias};
ClobberAlias CA =
instructionClobbersQuery(MD, Desc.Loc, Query->Inst, AA);
if (CA.IsClobber)
return {MD, true, CA.AR};
}
}
if (LimitAlreadyReached)
*UpwardWalkLimit = 0;
assert(isa<MemoryPhi>(Desc.Last) &&
"Ended at a non-clobber that's not a phi?");
return {Desc.Last, false, MayAlias};
}
void addSearches(MemoryPhi *Phi, SmallVectorImpl<ListIndex> &PausedSearches,
ListIndex PriorNode) {
auto UpwardDefsBegin = upward_defs_begin({Phi, Paths[PriorNode].Loc}, DT,
&PerformedPhiTranslation);
auto UpwardDefs = make_range(UpwardDefsBegin, upward_defs_end());
for (const MemoryAccessPair &P : UpwardDefs) {
PausedSearches.push_back(Paths.size());
Paths.emplace_back(P.second, P.first, PriorNode);
}
}
/// Represents a search that terminated after finding a clobber. This clobber
/// may or may not be present in the path of defs from LastNode..SearchStart,
/// since it may have been retrieved from cache.
struct TerminatedPath {
MemoryAccess *Clobber;
ListIndex LastNode;
};
/// Get an access that keeps us from optimizing to the given phi.
///
/// PausedSearches is an array of indices into the Paths array. Its incoming
/// value is the indices of searches that stopped at the last phi optimization
/// target. It's left in an unspecified state.
///
/// If this returns None, NewPaused is a vector of searches that terminated
/// at StopWhere. Otherwise, NewPaused is left in an unspecified state.
Optional<TerminatedPath>
getBlockingAccess(const MemoryAccess *StopWhere,
SmallVectorImpl<ListIndex> &PausedSearches,
SmallVectorImpl<ListIndex> &NewPaused,
SmallVectorImpl<TerminatedPath> &Terminated) {
assert(!PausedSearches.empty() && "No searches to continue?");
// BFS vs DFS really doesn't make a difference here, so just do a DFS with
// PausedSearches as our stack.
while (!PausedSearches.empty()) {
ListIndex PathIndex = PausedSearches.pop_back_val();
DefPath &Node = Paths[PathIndex];
// If we've already visited this path with this MemoryLocation, we don't
// need to do so again.
//
// NOTE: That we just drop these paths on the ground makes caching
// behavior sporadic. e.g. given a diamond:
// A
// B C
// D
//
// ...If we walk D, B, A, C, we'll only cache the result of phi
// optimization for A, B, and D; C will be skipped because it dies here.
// This arguably isn't the worst thing ever, since:
// - We generally query things in a top-down order, so if we got below D
// without needing cache entries for {C, MemLoc}, then chances are
// that those cache entries would end up ultimately unused.
// - We still cache things for A, so C only needs to walk up a bit.
// If this behavior becomes problematic, we can fix without a ton of extra
// work.
if (!VisitedPhis.insert({Node.Last, Node.Loc}).second) {
if (PerformedPhiTranslation) {
// If visiting this path performed Phi translation, don't continue,
// since it may not be correct to merge results from two paths if one
// relies on the phi translation.
TerminatedPath Term{Node.Last, PathIndex};
return Term;
}
continue;
}
const MemoryAccess *SkipStopWhere = nullptr;
if (Query->SkipSelfAccess && Node.Loc == Query->StartingLoc) {
assert(isa<MemoryDef>(Query->OriginalAccess));
SkipStopWhere = Query->OriginalAccess;
}
UpwardsWalkResult Res = walkToPhiOrClobber(Node,
/*StopAt=*/StopWhere,
/*SkipStopAt=*/SkipStopWhere);
if (Res.IsKnownClobber) {
assert(Res.Result != StopWhere && Res.Result != SkipStopWhere);
// If this wasn't a cache hit, we hit a clobber when walking. That's a
// failure.
TerminatedPath Term{Res.Result, PathIndex};
if (!MSSA.dominates(Res.Result, StopWhere))
return Term;
// Otherwise, it's a valid thing to potentially optimize to.
Terminated.push_back(Term);
continue;
}
if (Res.Result == StopWhere || Res.Result == SkipStopWhere) {
// We've hit our target. Save this path off for if we want to continue
// walking. If we are in the mode of skipping the OriginalAccess, and
// we've reached back to the OriginalAccess, do not save path, we've
// just looped back to self.
if (Res.Result != SkipStopWhere)
NewPaused.push_back(PathIndex);
continue;
}
assert(!MSSA.isLiveOnEntryDef(Res.Result) && "liveOnEntry is a clobber");
addSearches(cast<MemoryPhi>(Res.Result), PausedSearches, PathIndex);
}
return None;
}
template <typename T, typename Walker>
struct generic_def_path_iterator
: public iterator_facade_base<generic_def_path_iterator<T, Walker>,
std::forward_iterator_tag, T *> {
generic_def_path_iterator() {}
generic_def_path_iterator(Walker *W, ListIndex N) : W(W), N(N) {}
T &operator*() const { return curNode(); }
generic_def_path_iterator &operator++() {
N = curNode().Previous;
return *this;
}
bool operator==(const generic_def_path_iterator &O) const {
if (N.hasValue() != O.N.hasValue())
return false;
return !N.hasValue() || *N == *O.N;
}
private:
T &curNode() const { return W->Paths[*N]; }
Walker *W = nullptr;
Optional<ListIndex> N = None;
};
using def_path_iterator = generic_def_path_iterator<DefPath, ClobberWalker>;
using const_def_path_iterator =
generic_def_path_iterator<const DefPath, const ClobberWalker>;
iterator_range<def_path_iterator> def_path(ListIndex From) {
return make_range(def_path_iterator(this, From), def_path_iterator());
}
iterator_range<const_def_path_iterator> const_def_path(ListIndex From) const {
return make_range(const_def_path_iterator(this, From),
const_def_path_iterator());
}
struct OptznResult {
/// The path that contains our result.
TerminatedPath PrimaryClobber;
/// The paths that we can legally cache back from, but that aren't
/// necessarily the result of the Phi optimization.
SmallVector<TerminatedPath, 4> OtherClobbers;
};
ListIndex defPathIndex(const DefPath &N) const {
// The assert looks nicer if we don't need to do &N
const DefPath *NP = &N;
assert(!Paths.empty() && NP >= &Paths.front() && NP <= &Paths.back() &&
"Out of bounds DefPath!");
return NP - &Paths.front();
}
/// Try to optimize a phi as best as we can. Returns a SmallVector of Paths
/// that act as legal clobbers. Note that this won't return *all* clobbers.
///
/// Phi optimization algorithm tl;dr:
/// - Find the earliest def/phi, A, we can optimize to
/// - Find if all paths from the starting memory access ultimately reach A
/// - If not, optimization isn't possible.
/// - Otherwise, walk from A to another clobber or phi, A'.
/// - If A' is a def, we're done.
/// - If A' is a phi, try to optimize it.
///
/// A path is a series of {MemoryAccess, MemoryLocation} pairs. A path
/// terminates when a MemoryAccess that clobbers said MemoryLocation is found.
OptznResult tryOptimizePhi(MemoryPhi *Phi, MemoryAccess *Start,
const MemoryLocation &Loc) {
assert(Paths.empty() && VisitedPhis.empty() && !PerformedPhiTranslation &&
"Reset the optimization state.");
Paths.emplace_back(Loc, Start, Phi, None);
// Stores how many "valid" optimization nodes we had prior to calling
// addSearches/getBlockingAccess. Necessary for caching if we had a blocker.
auto PriorPathsSize = Paths.size();
SmallVector<ListIndex, 16> PausedSearches;
SmallVector<ListIndex, 8> NewPaused;
SmallVector<TerminatedPath, 4> TerminatedPaths;
addSearches(Phi, PausedSearches, 0);
// Moves the TerminatedPath with the "most dominated" Clobber to the end of
// Paths.
auto MoveDominatedPathToEnd = [&](SmallVectorImpl<TerminatedPath> &Paths) {
assert(!Paths.empty() && "Need a path to move");
auto Dom = Paths.begin();
for (auto I = std::next(Dom), E = Paths.end(); I != E; ++I)
if (!MSSA.dominates(I->Clobber, Dom->Clobber))
Dom = I;
auto Last = Paths.end() - 1;
if (Last != Dom)
std::iter_swap(Last, Dom);
};
MemoryPhi *Current = Phi;
while (true) {
assert(!MSSA.isLiveOnEntryDef(Current) &&
"liveOnEntry wasn't treated as a clobber?");
const auto *Target = getWalkTarget(Current);
// If a TerminatedPath doesn't dominate Target, then it wasn't a legal
// optimization for the prior phi.
assert(all_of(TerminatedPaths, [&](const TerminatedPath &P) {
return MSSA.dominates(P.Clobber, Target);
}));
// FIXME: This is broken, because the Blocker may be reported to be
// liveOnEntry, and we'll happily wait for that to disappear (read: never)
// For the moment, this is fine, since we do nothing with blocker info.
if (Optional<TerminatedPath> Blocker = getBlockingAccess(
Target, PausedSearches, NewPaused, TerminatedPaths)) {
// Find the node we started at. We can't search based on N->Last, since
// we may have gone around a loop with a different MemoryLocation.
auto Iter = find_if(def_path(Blocker->LastNode), [&](const DefPath &N) {
return defPathIndex(N) < PriorPathsSize;
});
assert(Iter != def_path_iterator());
DefPath &CurNode = *Iter;
assert(CurNode.Last == Current);
// Two things:
// A. We can't reliably cache all of NewPaused back. Consider a case
// where we have two paths in NewPaused; one of which can't optimize
// above this phi, whereas the other can. If we cache the second path
// back, we'll end up with suboptimal cache entries. We can handle
// cases like this a bit better when we either try to find all
// clobbers that block phi optimization, or when our cache starts
// supporting unfinished searches.
// B. We can't reliably cache TerminatedPaths back here without doing
// extra checks; consider a case like:
// T
// / \
// D C
// \ /
// S
// Where T is our target, C is a node with a clobber on it, D is a
// diamond (with a clobber *only* on the left or right node, N), and
// S is our start. Say we walk to D, through the node opposite N
// (read: ignoring the clobber), and see a cache entry in the top
// node of D. That cache entry gets put into TerminatedPaths. We then
// walk up to C (N is later in our worklist), find the clobber, and
// quit. If we append TerminatedPaths to OtherClobbers, we'll cache
// the bottom part of D to the cached clobber, ignoring the clobber
// in N. Again, this problem goes away if we start tracking all
// blockers for a given phi optimization.
TerminatedPath Result{CurNode.Last, defPathIndex(CurNode)};
return {Result, {}};
}
// If there's nothing left to search, then all paths led to valid clobbers
// that we got from our cache; pick the nearest to the start, and allow
// the rest to be cached back.
if (NewPaused.empty()) {
MoveDominatedPathToEnd(TerminatedPaths);
TerminatedPath Result = TerminatedPaths.pop_back_val();
return {Result, std::move(TerminatedPaths)};
}
MemoryAccess *DefChainEnd = nullptr;
SmallVector<TerminatedPath, 4> Clobbers;
for (ListIndex Paused : NewPaused) {
UpwardsWalkResult WR = walkToPhiOrClobber(Paths[Paused]);
if (WR.IsKnownClobber)
Clobbers.push_back({WR.Result, Paused});
else
// Micro-opt: If we hit the end of the chain, save it.
DefChainEnd = WR.Result;
}
if (!TerminatedPaths.empty()) {
// If we couldn't find the dominating phi/liveOnEntry in the above loop,
// do it now.
if (!DefChainEnd)
for (auto *MA : def_chain(const_cast<MemoryAccess *>(Target)))
DefChainEnd = MA;
assert(DefChainEnd && "Failed to find dominating phi/liveOnEntry");
// If any of the terminated paths don't dominate the phi we'll try to
// optimize, we need to figure out what they are and quit.
const BasicBlock *ChainBB = DefChainEnd->getBlock();
for (const TerminatedPath &TP : TerminatedPaths) {
// Because we know that DefChainEnd is as "high" as we can go, we
// don't need local dominance checks; BB dominance is sufficient.
if (DT.dominates(ChainBB, TP.Clobber->getBlock()))
Clobbers.push_back(TP);
}
}
// If we have clobbers in the def chain, find the one closest to Current
// and quit.
if (!Clobbers.empty()) {
MoveDominatedPathToEnd(Clobbers);
TerminatedPath Result = Clobbers.pop_back_val();
return {Result, std::move(Clobbers)};
}
assert(all_of(NewPaused,
[&](ListIndex I) { return Paths[I].Last == DefChainEnd; }));
// Because liveOnEntry is a clobber, this must be a phi.
auto *DefChainPhi = cast<MemoryPhi>(DefChainEnd);
PriorPathsSize = Paths.size();
PausedSearches.clear();
for (ListIndex I : NewPaused)
addSearches(DefChainPhi, PausedSearches, I);
NewPaused.clear();
Current = DefChainPhi;
}
}
void verifyOptResult(const OptznResult &R) const {
assert(all_of(R.OtherClobbers, [&](const TerminatedPath &P) {
return MSSA.dominates(P.Clobber, R.PrimaryClobber.Clobber);
}));
}
void resetPhiOptznState() {
Paths.clear();
VisitedPhis.clear();
PerformedPhiTranslation = false;
}
public:
ClobberWalker(const MemorySSA &MSSA, AliasAnalysisType &AA, DominatorTree &DT)
: MSSA(MSSA), AA(AA), DT(DT) {}
AliasAnalysisType *getAA() { return &AA; }
/// Finds the nearest clobber for the given query, optimizing phis if
/// possible.
MemoryAccess *findClobber(MemoryAccess *Start, UpwardsMemoryQuery &Q,
unsigned &UpWalkLimit) {
Query = &Q;
UpwardWalkLimit = &UpWalkLimit;
// Starting limit must be > 0.
if (!UpWalkLimit)
UpWalkLimit++;
MemoryAccess *Current = Start;
// This walker pretends uses don't exist. If we're handed one, silently grab
// its def. (This has the nice side-effect of ensuring we never cache uses)
if (auto *MU = dyn_cast<MemoryUse>(Start))
Current = MU->getDefiningAccess();
DefPath FirstDesc(Q.StartingLoc, Current, Current, None);
// Fast path for the overly-common case (no crazy phi optimization
// necessary)
UpwardsWalkResult WalkResult = walkToPhiOrClobber(FirstDesc);
MemoryAccess *Result;
if (WalkResult.IsKnownClobber) {
Result = WalkResult.Result;
Q.AR = WalkResult.AR;
} else {
OptznResult OptRes = tryOptimizePhi(cast<MemoryPhi>(FirstDesc.Last),
Current, Q.StartingLoc);
verifyOptResult(OptRes);
resetPhiOptznState();
Result = OptRes.PrimaryClobber.Clobber;
}
#ifdef EXPENSIVE_CHECKS
if (!Q.SkipSelfAccess && *UpwardWalkLimit > 0)
checkClobberSanity(Current, Result, Q.StartingLoc, MSSA, Q, AA);
#endif
return Result;
}
};
struct RenamePassData {
DomTreeNode *DTN;
DomTreeNode::const_iterator ChildIt;
MemoryAccess *IncomingVal;
RenamePassData(DomTreeNode *D, DomTreeNode::const_iterator It,
MemoryAccess *M)
: DTN(D), ChildIt(It), IncomingVal(M) {}
void swap(RenamePassData &RHS) {
std::swap(DTN, RHS.DTN);
std::swap(ChildIt, RHS.ChildIt);
std::swap(IncomingVal, RHS.IncomingVal);
}
};
} // end anonymous namespace
namespace llvm {
template <class AliasAnalysisType> class MemorySSA::ClobberWalkerBase {
ClobberWalker<AliasAnalysisType> Walker;
MemorySSA *MSSA;
public:
ClobberWalkerBase(MemorySSA *M, AliasAnalysisType *A, DominatorTree *D)
: Walker(*M, *A, *D), MSSA(M) {}
MemoryAccess *getClobberingMemoryAccessBase(MemoryAccess *,
const MemoryLocation &,
unsigned &);
// Third argument (bool), defines whether the clobber search should skip the
// original queried access. If true, there will be a follow-up query searching
// for a clobber access past "self". Note that the Optimized access is not
// updated if a new clobber is found by this SkipSelf search. If this
// additional query becomes heavily used we may decide to cache the result.
// Walker instantiations will decide how to set the SkipSelf bool.
MemoryAccess *getClobberingMemoryAccessBase(MemoryAccess *, unsigned &, bool);
};
/// A MemorySSAWalker that does AA walks to disambiguate accesses. It no
/// longer does caching on its own, but the name has been retained for the
/// moment.
template <class AliasAnalysisType>
class MemorySSA::CachingWalker final : public MemorySSAWalker {
ClobberWalkerBase<AliasAnalysisType> *Walker;
public:
CachingWalker(MemorySSA *M, ClobberWalkerBase<AliasAnalysisType> *W)
: MemorySSAWalker(M), Walker(W) {}
~CachingWalker() override = default;
using MemorySSAWalker::getClobberingMemoryAccess;
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA, unsigned &UWL) {
return Walker->getClobberingMemoryAccessBase(MA, UWL, false);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA,
const MemoryLocation &Loc,
unsigned &UWL) {
return Walker->getClobberingMemoryAccessBase(MA, Loc, UWL);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA) override {
unsigned UpwardWalkLimit = MaxCheckLimit;
return getClobberingMemoryAccess(MA, UpwardWalkLimit);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA,
const MemoryLocation &Loc) override {
unsigned UpwardWalkLimit = MaxCheckLimit;
return getClobberingMemoryAccess(MA, Loc, UpwardWalkLimit);
}
void invalidateInfo(MemoryAccess *MA) override {
if (auto *MUD = dyn_cast<MemoryUseOrDef>(MA))
MUD->resetOptimized();
}
};
template <class AliasAnalysisType>
class MemorySSA::SkipSelfWalker final : public MemorySSAWalker {
ClobberWalkerBase<AliasAnalysisType> *Walker;
public:
SkipSelfWalker(MemorySSA *M, ClobberWalkerBase<AliasAnalysisType> *W)
: MemorySSAWalker(M), Walker(W) {}
~SkipSelfWalker() override = default;
using MemorySSAWalker::getClobberingMemoryAccess;
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA, unsigned &UWL) {
return Walker->getClobberingMemoryAccessBase(MA, UWL, true);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA,
const MemoryLocation &Loc,
unsigned &UWL) {
return Walker->getClobberingMemoryAccessBase(MA, Loc, UWL);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA) override {
unsigned UpwardWalkLimit = MaxCheckLimit;
return getClobberingMemoryAccess(MA, UpwardWalkLimit);
}
MemoryAccess *getClobberingMemoryAccess(MemoryAccess *MA,
const MemoryLocation &Loc) override {
unsigned UpwardWalkLimit = MaxCheckLimit;
return getClobberingMemoryAccess(MA, Loc, UpwardWalkLimit);
}
void invalidateInfo(MemoryAccess *MA) override {
if (auto *MUD = dyn_cast<MemoryUseOrDef>(MA))
MUD->resetOptimized();
}
};
} // end namespace llvm
void MemorySSA::renameSuccessorPhis(BasicBlock *BB, MemoryAccess *IncomingVal,
bool RenameAllUses) {
// Pass through values to our successors
for (const BasicBlock *S : successors(BB)) {
auto It = PerBlockAccesses.find(S);
// Rename the phi nodes in our successor block
if (It == PerBlockAccesses.end() || !isa<MemoryPhi>(It->second->front()))
continue;
AccessList *Accesses = It->second.get();
auto *Phi = cast<MemoryPhi>(&Accesses->front());
if (RenameAllUses) {
bool ReplacementDone = false;
for (unsigned I = 0, E = Phi->getNumIncomingValues(); I != E; ++I)
if (Phi->getIncomingBlock(I) == BB) {
Phi->setIncomingValue(I, IncomingVal);
ReplacementDone = true;
}
(void) ReplacementDone;
assert(ReplacementDone && "Incomplete phi during partial rename");
} else
Phi->addIncoming(IncomingVal, BB);
}
}
/// Rename a single basic block into MemorySSA form.
/// Uses the standard SSA renaming algorithm.
/// \returns The new incoming value.
MemoryAccess *MemorySSA::renameBlock(BasicBlock *BB, MemoryAccess *IncomingVal,
bool RenameAllUses) {
auto It = PerBlockAccesses.find(BB);
// Skip most processing if the list is empty.
if (It != PerBlockAccesses.end()) {
AccessList *Accesses = It->second.get();
for (MemoryAccess &L : *Accesses) {
if (MemoryUseOrDef *MUD = dyn_cast<MemoryUseOrDef>(&L)) {
if (MUD->getDefiningAccess() == nullptr || RenameAllUses)
MUD->setDefiningAccess(IncomingVal);
if (isa<MemoryDef>(&L))
IncomingVal = &L;
} else {
IncomingVal = &L;
}
}
}
return IncomingVal;
}
/// This is the standard SSA renaming algorithm.
///
/// We walk the dominator tree in preorder, renaming accesses, and then filling
/// in phi nodes in our successors.
void MemorySSA::renamePass(DomTreeNode *Root, MemoryAccess *IncomingVal,
SmallPtrSetImpl<BasicBlock *> &Visited,
bool SkipVisited, bool RenameAllUses) {
assert(Root && "Trying to rename accesses in an unreachable block");
SmallVector<RenamePassData, 32> WorkStack;
// Skip everything if we already renamed this block and we are skipping.
// Note: You can't sink this into the if, because we need it to occur
// regardless of whether we skip blocks or not.
bool AlreadyVisited = !Visited.insert(Root->getBlock()).second;
if (SkipVisited && AlreadyVisited)
return;
IncomingVal = renameBlock(Root->getBlock(), IncomingVal, RenameAllUses);
renameSuccessorPhis(Root->getBlock(), IncomingVal, RenameAllUses);
WorkStack.push_back({Root, Root->begin(), IncomingVal});
while (!WorkStack.empty()) {
DomTreeNode *Node = WorkStack.back().DTN;
DomTreeNode::const_iterator ChildIt = WorkStack.back().ChildIt;
IncomingVal = WorkStack.back().IncomingVal;
if (ChildIt == Node->end()) {
WorkStack.pop_back();
} else {
DomTreeNode *Child = *ChildIt;
++WorkStack.back().ChildIt;
BasicBlock *BB = Child->getBlock();
// Note: You can't sink this into the if, because we need it to occur
// regardless of whether we skip blocks or not.
AlreadyVisited = !Visited.insert(BB).second;
if (SkipVisited && AlreadyVisited) {
// We already visited this during our renaming, which can happen when
// being asked to rename multiple blocks. Figure out the incoming val,
// which is the last def.
// Incoming value can only change if there is a block def, and in that
// case, it's the last block def in the list.
if (auto *BlockDefs = getWritableBlockDefs(BB))
IncomingVal = &*BlockDefs->rbegin();
} else
IncomingVal = renameBlock(BB, IncomingVal, RenameAllUses);
renameSuccessorPhis(BB, IncomingVal, RenameAllUses);
WorkStack.push_back({Child, Child->begin(), IncomingVal});
}
}
}
/// This handles unreachable block accesses by deleting phi nodes in
/// unreachable blocks, and marking all other unreachable MemoryAccess's as
/// being uses of the live on entry definition.
void MemorySSA::markUnreachableAsLiveOnEntry(BasicBlock *BB) {
assert(!DT->isReachableFromEntry(BB) &&
"Reachable block found while handling unreachable blocks");
// Make sure phi nodes in our reachable successors end up with a
// LiveOnEntryDef for our incoming edge, even though our block is forward
// unreachable. We could just disconnect these blocks from the CFG fully,
// but we do not right now.
for (const BasicBlock *S : successors(BB)) {
if (!DT->isReachableFromEntry(S))
continue;
auto It = PerBlockAccesses.find(S);
// Rename the phi nodes in our successor block
if (It == PerBlockAccesses.end() || !isa<MemoryPhi>(It->second->front()))
continue;
AccessList *Accesses = It->second.get();
auto *Phi = cast<MemoryPhi>(&Accesses->front());
Phi->addIncoming(LiveOnEntryDef.get(), BB);
}
auto It = PerBlockAccesses.find(BB);
if (It == PerBlockAccesses.end())
return;
auto &Accesses = It->second;
for (auto AI = Accesses->begin(), AE = Accesses->end(); AI != AE;) {
auto Next = std::next(AI);
// If we have a phi, just remove it. We are going to replace all
// users with live on entry.
if (auto *UseOrDef = dyn_cast<MemoryUseOrDef>(AI))
UseOrDef->setDefiningAccess(LiveOnEntryDef.get());
else
Accesses->erase(AI);
AI = Next;
}
}
MemorySSA::MemorySSA(Function &Func, AliasAnalysis *AA, DominatorTree *DT)
: AA(nullptr), DT(DT), F(Func), LiveOnEntryDef(nullptr), Walker(nullptr),
SkipWalker(nullptr), NextID(0) {
// Build MemorySSA using a batch alias analysis. This reuses the internal
// state that AA collects during an alias()/getModRefInfo() call. This is
// safe because there are no CFG changes while building MemorySSA and can
// significantly reduce the time spent by the compiler in AA, because we will
// make queries about all the instructions in the Function.
assert(AA && "No alias analysis?");
BatchAAResults BatchAA(*AA);
buildMemorySSA(BatchAA);
// Intentionally leave AA to nullptr while building so we don't accidently
// use non-batch AliasAnalysis.
this->AA = AA;
// Also create the walker here.
getWalker();
}
MemorySSA::~MemorySSA() {
// Drop all our references
for (const auto &Pair : PerBlockAccesses)
for (MemoryAccess &MA : *Pair.second)
MA.dropAllReferences();
}
MemorySSA::AccessList *MemorySSA::getOrCreateAccessList(const BasicBlock *BB) {
auto Res = PerBlockAccesses.insert(std::make_pair(BB, nullptr));
if (Res.second)
Res.first->second = std::make_unique<AccessList>();
return Res.first->second.get();
}
MemorySSA::DefsList *MemorySSA::getOrCreateDefsList(const BasicBlock *BB) {
auto Res = PerBlockDefs.insert(std::make_pair(BB, nullptr));
if (Res.second)
Res.first->second = std::make_unique<DefsList>();
return Res.first->second.get();
}
namespace llvm {
/// This class is a batch walker of all MemoryUse's in the program, and points
/// their defining access at the thing that actually clobbers them. Because it
/// is a batch walker that touches everything, it does not operate like the
/// other walkers. This walker is basically performing a top-down SSA renaming
/// pass, where the version stack is used as the cache. This enables it to be
/// significantly more time and memory efficient than using the regular walker,
/// which is walking bottom-up.
class MemorySSA::OptimizeUses {
public:
OptimizeUses(MemorySSA *MSSA, CachingWalker<BatchAAResults> *Walker,
BatchAAResults *BAA, DominatorTree *DT)
: MSSA(MSSA), Walker(Walker), AA(BAA), DT(DT) {}
void optimizeUses();
private:
/// This represents where a given memorylocation is in the stack.
struct MemlocStackInfo {
// This essentially is keeping track of versions of the stack. Whenever
// the stack changes due to pushes or pops, these versions increase.
unsigned long StackEpoch;
unsigned long PopEpoch;
// This is the lower bound of places on the stack to check. It is equal to
// the place the last stack walk ended.
// Note: Correctness depends on this being initialized to 0, which densemap
// does
unsigned long LowerBound;
const BasicBlock *LowerBoundBlock;
// This is where the last walk for this memory location ended.
unsigned long LastKill;
bool LastKillValid;
Optional<AliasResult> AR;
};
void optimizeUsesInBlock(const BasicBlock *, unsigned long &, unsigned long &,
SmallVectorImpl<MemoryAccess *> &,
DenseMap<MemoryLocOrCall, MemlocStackInfo> &);
MemorySSA *MSSA;
CachingWalker<BatchAAResults> *Walker;
BatchAAResults *AA;
DominatorTree *DT;
};
} // end namespace llvm
/// Optimize the uses in a given block This is basically the SSA renaming
/// algorithm, with one caveat: We are able to use a single stack for all
/// MemoryUses. This is because the set of *possible* reaching MemoryDefs is
/// the same for every MemoryUse. The *actual* clobbering MemoryDef is just
/// going to be some position in that stack of possible ones.
///
/// We track the stack positions that each MemoryLocation needs
/// to check, and last ended at. This is because we only want to check the
/// things that changed since last time. The same MemoryLocation should
/// get clobbered by the same store (getModRefInfo does not use invariantness or
/// things like this, and if they start, we can modify MemoryLocOrCall to
/// include relevant data)
void MemorySSA::OptimizeUses::optimizeUsesInBlock(
const BasicBlock *BB, unsigned long &StackEpoch, unsigned long &PopEpoch,
SmallVectorImpl<MemoryAccess *> &VersionStack,
DenseMap<MemoryLocOrCall, MemlocStackInfo> &LocStackInfo) {
/// If no accesses, nothing to do.
MemorySSA::AccessList *Accesses = MSSA->getWritableBlockAccesses(BB);
if (Accesses == nullptr)
return;
// Pop everything that doesn't dominate the current block off the stack,
// increment the PopEpoch to account for this.
while (true) {
assert(
!VersionStack.empty() &&
"Version stack should have liveOnEntry sentinel dominating everything");
BasicBlock *BackBlock = VersionStack.back()->getBlock();
if (DT->dominates(BackBlock, BB))
break;
while (VersionStack.back()->getBlock() == BackBlock)
VersionStack.pop_back();
++PopEpoch;
}
for (MemoryAccess &MA : *Accesses) {
auto *MU = dyn_cast<MemoryUse>(&MA);
if (!MU) {
VersionStack.push_back(&MA);
++StackEpoch;
continue;
}
if (isUseTriviallyOptimizableToLiveOnEntry(*AA, MU->getMemoryInst())) {
MU->setDefiningAccess(MSSA->getLiveOnEntryDef(), true, None);
continue;
}
MemoryLocOrCall UseMLOC(MU);
auto &LocInfo = LocStackInfo[UseMLOC];
// If the pop epoch changed, it means we've removed stuff from top of
// stack due to changing blocks. We may have to reset the lower bound or
// last kill info.
if (LocInfo.PopEpoch != PopEpoch) {
LocInfo.PopEpoch = PopEpoch;
LocInfo.StackEpoch = StackEpoch;
// If the lower bound was in something that no longer dominates us, we
// have to reset it.
// We can't simply track stack size, because the stack may have had
// pushes/pops in the meantime.
// XXX: This is non-optimal, but only is slower cases with heavily
// branching dominator trees. To get the optimal number of queries would
// be to make lowerbound and lastkill a per-loc stack, and pop it until
// the top of that stack dominates us. This does not seem worth it ATM.
// A much cheaper optimization would be to always explore the deepest
// branch of the dominator tree first. This will guarantee this resets on
// the smallest set of blocks.
if (LocInfo.LowerBoundBlock && LocInfo.LowerBoundBlock != BB &&
!DT->dominates(LocInfo.LowerBoundBlock, BB)) {
// Reset the lower bound of things to check.
// TODO: Some day we should be able to reset to last kill, rather than
// 0.
LocInfo.LowerBound = 0;
LocInfo.LowerBoundBlock = VersionStack[0]->getBlock();
LocInfo.LastKillValid = false;
}
} else if (LocInfo.StackEpoch != StackEpoch) {
// If all that has changed is the StackEpoch, we only have to check the
// new things on the stack, because we've checked everything before. In
// this case, the lower bound of things to check remains the same.
LocInfo.PopEpoch = PopEpoch;
LocInfo.StackEpoch = StackEpoch;
}
if (!LocInfo.LastKillValid) {
LocInfo.LastKill = VersionStack.size() - 1;
LocInfo.LastKillValid = true;
LocInfo.AR = MayAlias;
}
// At this point, we should have corrected last kill and LowerBound to be
// in bounds.
assert(LocInfo.LowerBound < VersionStack.size() &&
"Lower bound out of range");
assert(LocInfo.LastKill < VersionStack.size() &&
"Last kill info out of range");
// In any case, the new upper bound is the top of the stack.
unsigned long UpperBound = VersionStack.size() - 1;
if (UpperBound - LocInfo.LowerBound > MaxCheckLimit) {
LLVM_DEBUG(dbgs() << "MemorySSA skipping optimization of " << *MU << " ("
<< *(MU->getMemoryInst()) << ")"
<< " because there are "
<< UpperBound - LocInfo.LowerBound
<< " stores to disambiguate\n");
// Because we did not walk, LastKill is no longer valid, as this may
// have been a kill.
LocInfo.LastKillValid = false;
continue;
}
bool FoundClobberResult = false;
unsigned UpwardWalkLimit = MaxCheckLimit;
while (UpperBound > LocInfo.LowerBound) {
if (isa<MemoryPhi>(VersionStack[UpperBound])) {
// For phis, use the walker, see where we ended up, go there
MemoryAccess *Result =
Walker->getClobberingMemoryAccess(MU, UpwardWalkLimit);
// We are guaranteed to find it or something is wrong
while (VersionStack[UpperBound] != Result) {
assert(UpperBound != 0);
--UpperBound;
}
FoundClobberResult = true;
break;
}
MemoryDef *MD = cast<MemoryDef>(VersionStack[UpperBound]);
// If the lifetime of the pointer ends at this instruction, it's live on
// entry.
if (!UseMLOC.IsCall && lifetimeEndsAt(MD, UseMLOC.getLoc(), *AA)) {
// Reset UpperBound to liveOnEntryDef's place in the stack
UpperBound = 0;
FoundClobberResult = true;
LocInfo.AR = MustAlias;
break;
}
ClobberAlias CA = instructionClobbersQuery(MD, MU, UseMLOC, *AA);
if (CA.IsClobber) {
FoundClobberResult = true;
LocInfo.AR = CA.AR;
break;
}
--UpperBound;
}
// Note: Phis always have AliasResult AR set to MayAlias ATM.
// At the end of this loop, UpperBound is either a clobber, or lower bound
// PHI walking may cause it to be < LowerBound, and in fact, < LastKill.
if (FoundClobberResult || UpperBound < LocInfo.LastKill) {
// We were last killed now by where we got to
if (MSSA->isLiveOnEntryDef(VersionStack[UpperBound]))
LocInfo.AR = None;
MU->setDefiningAccess(VersionStack[UpperBound], true, LocInfo.AR);
LocInfo.LastKill = UpperBound;
} else {
// Otherwise, we checked all the new ones, and now we know we can get to
// LastKill.
MU->setDefiningAccess(VersionStack[LocInfo.LastKill], true, LocInfo.AR);
}
LocInfo.LowerBound = VersionStack.size() - 1;
LocInfo.LowerBoundBlock = BB;
}
}
/// Optimize uses to point to their actual clobbering definitions.
void MemorySSA::OptimizeUses::optimizeUses() {
SmallVector<MemoryAccess *, 16> VersionStack;
DenseMap<MemoryLocOrCall, MemlocStackInfo> LocStackInfo;
VersionStack.push_back(MSSA->getLiveOnEntryDef());
unsigned long StackEpoch = 1;
unsigned long PopEpoch = 1;
// We perform a non-recursive top-down dominator tree walk.
for (const auto *DomNode : depth_first(DT->getRootNode()))
optimizeUsesInBlock(DomNode->getBlock(), StackEpoch, PopEpoch, VersionStack,
LocStackInfo);
}
void MemorySSA::placePHINodes(
const SmallPtrSetImpl<BasicBlock *> &DefiningBlocks) {
// Determine where our MemoryPhi's should go
ForwardIDFCalculator IDFs(*DT);
IDFs.setDefiningBlocks(DefiningBlocks);
SmallVector<BasicBlock *, 32> IDFBlocks;
IDFs.calculate(IDFBlocks);
// Now place MemoryPhi nodes.
for (auto &BB : IDFBlocks)
createMemoryPhi(BB);
}
void MemorySSA::buildMemorySSA(BatchAAResults &BAA) {
// We create an access to represent "live on entry", for things like
// arguments or users of globals, where the memory they use is defined before
// the beginning of the function. We do not actually insert it into the IR.
// We do not define a live on exit for the immediate uses, and thus our
// semantics do *not* imply that something with no immediate uses can simply
// be removed.
BasicBlock &StartingPoint = F.getEntryBlock();
LiveOnEntryDef.reset(new MemoryDef(F.getContext(), nullptr, nullptr,
&StartingPoint, NextID++));
// We maintain lists of memory accesses per-block, trading memory for time. We
// could just look up the memory access for every possible instruction in the
// stream.
SmallPtrSet<BasicBlock *, 32> DefiningBlocks;
// Go through each block, figure out where defs occur, and chain together all
// the accesses.
for (BasicBlock &B : F) {
bool InsertIntoDef = false;
AccessList *Accesses = nullptr;
DefsList *Defs = nullptr;
for (Instruction &I : B) {
MemoryUseOrDef *MUD = createNewAccess(&I, &BAA);
if (!MUD)
continue;
if (!Accesses)
Accesses = getOrCreateAccessList(&B);
Accesses->push_back(MUD);
if (isa<MemoryDef>(MUD)) {
InsertIntoDef = true;
if (!Defs)
Defs = getOrCreateDefsList(&B);
Defs->push_back(*MUD);
}
}
if (InsertIntoDef)
DefiningBlocks.insert(&B);
}
placePHINodes(DefiningBlocks);
// Now do regular SSA renaming on the MemoryDef/MemoryUse. Visited will get
// filled in with all blocks.
SmallPtrSet<BasicBlock *, 16> Visited;
renamePass(DT->getRootNode(), LiveOnEntryDef.get(), Visited);
ClobberWalkerBase<BatchAAResults> WalkerBase(this, &BAA, DT);
CachingWalker<BatchAAResults> WalkerLocal(this, &WalkerBase);
OptimizeUses(this, &WalkerLocal, &BAA, DT).optimizeUses();
// Mark the uses in unreachable blocks as live on entry, so that they go
// somewhere.
for (auto &BB : F)
if (!Visited.count(&BB))
markUnreachableAsLiveOnEntry(&BB);
}
MemorySSAWalker *MemorySSA::getWalker() { return getWalkerImpl(); }
MemorySSA::CachingWalker<AliasAnalysis> *MemorySSA::getWalkerImpl() {
if (Walker)
return Walker.get();
if (!WalkerBase)
WalkerBase =
std::make_unique<ClobberWalkerBase<AliasAnalysis>>(this, AA, DT);
Walker =
std::make_unique<CachingWalker<AliasAnalysis>>(this, WalkerBase.get());
return Walker.get();
}
MemorySSAWalker *MemorySSA::getSkipSelfWalker() {
if (SkipWalker)
return SkipWalker.get();
if (!WalkerBase)
WalkerBase =
std::make_unique<ClobberWalkerBase<AliasAnalysis>>(this, AA, DT);
SkipWalker =
std::make_unique<SkipSelfWalker<AliasAnalysis>>(this, WalkerBase.get());
return SkipWalker.get();
}
// This is a helper function used by the creation routines. It places NewAccess
// into the access and defs lists for a given basic block, at the given
// insertion point.
void MemorySSA::insertIntoListsForBlock(MemoryAccess *NewAccess,
const BasicBlock *BB,
InsertionPlace Point) {
auto *Accesses = getOrCreateAccessList(BB);
if (Point == Beginning) {
// If it's a phi node, it goes first, otherwise, it goes after any phi
// nodes.
if (isa<MemoryPhi>(NewAccess)) {
Accesses->push_front(NewAccess);
auto *Defs = getOrCreateDefsList(BB);
Defs->push_front(*NewAccess);
} else {
auto AI = find_if_not(
*Accesses, [](const MemoryAccess &MA) { return isa<MemoryPhi>(MA); });
Accesses->insert(AI, NewAccess);
if (!isa<MemoryUse>(NewAccess)) {
auto *Defs = getOrCreateDefsList(BB);
auto DI = find_if_not(
*Defs, [](const MemoryAccess &MA) { return isa<MemoryPhi>(MA); });
Defs->insert(DI, *NewAccess);
}
}
} else {
Accesses->push_back(NewAccess);
if (!isa<MemoryUse>(NewAccess)) {
auto *Defs = getOrCreateDefsList(BB);
Defs->push_back(*NewAccess);
}
}
BlockNumberingValid.erase(BB);
}
void MemorySSA::insertIntoListsBefore(MemoryAccess *What, const BasicBlock *BB,
AccessList::iterator InsertPt) {
auto *Accesses = getWritableBlockAccesses(BB);
bool WasEnd = InsertPt == Accesses->end();
Accesses->insert(AccessList::iterator(InsertPt), What);
if (!isa<MemoryUse>(What)) {
auto *Defs = getOrCreateDefsList(BB);
// If we got asked to insert at the end, we have an easy job, just shove it
// at the end. If we got asked to insert before an existing def, we also get
// an iterator. If we got asked to insert before a use, we have to hunt for
// the next def.
if (WasEnd) {
Defs->push_back(*What);
} else if (isa<MemoryDef>(InsertPt)) {
Defs->insert(InsertPt->getDefsIterator(), *What);
} else {
while (InsertPt != Accesses->end() && !isa<MemoryDef>(InsertPt))
++InsertPt;
// Either we found a def, or we are inserting at the end
if (InsertPt == Accesses->end())
Defs->push_back(*What);
else
Defs->insert(InsertPt->getDefsIterator(), *What);
}
}
BlockNumberingValid.erase(BB);
}
void MemorySSA::prepareForMoveTo(MemoryAccess *What, BasicBlock *BB) {
// Keep it in the lookup tables, remove from the lists
removeFromLists(What, false);
// Note that moving should implicitly invalidate the optimized state of a
// MemoryUse (and Phis can't be optimized). However, it doesn't do so for a
// MemoryDef.
if (auto *MD = dyn_cast<MemoryDef>(What))
MD->resetOptimized();
What->setBlock(BB);
}
// Move What before Where in the IR. The end result is that What will belong to
// the right lists and have the right Block set, but will not otherwise be
// correct. It will not have the right defining access, and if it is a def,
// things below it will not properly be updated.
void MemorySSA::moveTo(MemoryUseOrDef *What, BasicBlock *BB,
AccessList::iterator Where) {
prepareForMoveTo(What, BB);
insertIntoListsBefore(What, BB, Where);
}
void MemorySSA::moveTo(MemoryAccess *What, BasicBlock *BB,
InsertionPlace Point) {
if (isa<MemoryPhi>(What)) {
assert(Point == Beginning &&
"Can only move a Phi at the beginning of the block");
// Update lookup table entry
ValueToMemoryAccess.erase(What->getBlock());
bool Inserted = ValueToMemoryAccess.insert({BB, What}).second;
(void)Inserted;
assert(Inserted && "Cannot move a Phi to a block that already has one");
}
prepareForMoveTo(What, BB);
insertIntoListsForBlock(What, BB, Point);
}
MemoryPhi *MemorySSA::createMemoryPhi(BasicBlock *BB) {
assert(!getMemoryAccess(BB) && "MemoryPhi already exists for this BB");
MemoryPhi *Phi = new MemoryPhi(BB->getContext(), BB, NextID++);
// Phi's always are placed at the front of the block.
insertIntoListsForBlock(Phi, BB, Beginning);
ValueToMemoryAccess[BB] = Phi;
return Phi;
}
MemoryUseOrDef *MemorySSA::createDefinedAccess(Instruction *I,
MemoryAccess *Definition,
const MemoryUseOrDef *Template,
bool CreationMustSucceed) {
assert(!isa<PHINode>(I) && "Cannot create a defined access for a PHI");
MemoryUseOrDef *NewAccess = createNewAccess(I, AA, Template);
if (CreationMustSucceed)
assert(NewAccess != nullptr && "Tried to create a memory access for a "
"non-memory touching instruction");
if (NewAccess) {
assert((!Definition || !isa<MemoryUse>(Definition)) &&
"A use cannot be a defining access");
NewAccess->setDefiningAccess(Definition);
}
return NewAccess;
}
// Return true if the instruction has ordering constraints.
// Note specifically that this only considers stores and loads
// because others are still considered ModRef by getModRefInfo.
static inline bool isOrdered(const Instruction *I) {
if (auto *SI = dyn_cast<StoreInst>(I)) {
if (!SI->isUnordered())
return true;
} else if (auto *LI = dyn_cast<LoadInst>(I)) {
if (!LI->isUnordered())
return true;
}
return false;
}
/// Helper function to create new memory accesses
template <typename AliasAnalysisType>
MemoryUseOrDef *MemorySSA::createNewAccess(Instruction *I,
AliasAnalysisType *AAP,
const MemoryUseOrDef *Template) {
// The assume intrinsic has a control dependency which we model by claiming
// that it writes arbitrarily. Debuginfo intrinsics may be considered
// clobbers when we have a nonstandard AA pipeline. Ignore these fake memory
// dependencies here.
// FIXME: Replace this special casing with a more accurate modelling of
// assume's control dependency.
if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I))
if (II->getIntrinsicID() == Intrinsic::assume)
return nullptr;
// Using a nonstandard AA pipelines might leave us with unexpected modref
// results for I, so add a check to not model instructions that may not read
// from or write to memory. This is necessary for correctness.
if (!I->mayReadFromMemory() && !I->mayWriteToMemory())
return nullptr;
bool Def, Use;
if (Template) {
Def = dyn_cast_or_null<MemoryDef>(Template) != nullptr;
Use = dyn_cast_or_null<MemoryUse>(Template) != nullptr;
#if !defined(NDEBUG)
ModRefInfo ModRef = AAP->getModRefInfo(I, None);
bool DefCheck, UseCheck;
DefCheck = isModSet(ModRef) || isOrdered(I);
UseCheck = isRefSet(ModRef);
assert(Def == DefCheck && (Def || Use == UseCheck) && "Invalid template");
#endif
} else {
// Find out what affect this instruction has on memory.
ModRefInfo ModRef = AAP->getModRefInfo(I, None);
// The isOrdered check is used to ensure that volatiles end up as defs
// (atomics end up as ModRef right now anyway). Until we separate the
// ordering chain from the memory chain, this enables people to see at least
// some relative ordering to volatiles. Note that getClobberingMemoryAccess
// will still give an answer that bypasses other volatile loads. TODO:
// Separate memory aliasing and ordering into two different chains so that
// we can precisely represent both "what memory will this read/write/is
// clobbered by" and "what instructions can I move this past".
Def = isModSet(ModRef) || isOrdered(I);
Use = isRefSet(ModRef);
}
// It's possible for an instruction to not modify memory at all. During
// construction, we ignore them.
if (!Def && !Use)
return nullptr;
MemoryUseOrDef *MUD;
if (Def)
MUD = new MemoryDef(I->getContext(), nullptr, I, I->getParent(), NextID++);
else
MUD = new MemoryUse(I->getContext(), nullptr, I, I->getParent());
ValueToMemoryAccess[I] = MUD;
return MUD;
}
/// Returns true if \p Replacer dominates \p Replacee .
bool MemorySSA::dominatesUse(const MemoryAccess *Replacer,
const MemoryAccess *Replacee) const {
if (isa<MemoryUseOrDef>(Replacee))
return DT->dominates(Replacer->getBlock(), Replacee->getBlock());
const auto *MP = cast<MemoryPhi>(Replacee);
// For a phi node, the use occurs in the predecessor block of the phi node.
// Since we may occur multiple times in the phi node, we have to check each
// operand to ensure Replacer dominates each operand where Replacee occurs.
for (const Use &Arg : MP->operands()) {
if (Arg.get() != Replacee &&
!DT->dominates(Replacer->getBlock(), MP->getIncomingBlock(Arg)))
return false;
}
return true;
}
/// Properly remove \p MA from all of MemorySSA's lookup tables.
void MemorySSA::removeFromLookups(MemoryAccess *MA) {
assert(MA->use_empty() &&
"Trying to remove memory access that still has uses");
BlockNumbering.erase(MA);
if (auto *MUD = dyn_cast<MemoryUseOrDef>(MA))
MUD->setDefiningAccess(nullptr);
// Invalidate our walker's cache if necessary
if (!isa<MemoryUse>(MA))
getWalker()->invalidateInfo(MA);
Value *MemoryInst;
if (const auto *MUD = dyn_cast<MemoryUseOrDef>(MA))
MemoryInst = MUD->getMemoryInst();
else
MemoryInst = MA->getBlock();
auto VMA = ValueToMemoryAccess.find(MemoryInst);
if (VMA->second == MA)
ValueToMemoryAccess.erase(VMA);
}
/// Properly remove \p MA from all of MemorySSA's lists.
///
/// Because of the way the intrusive list and use lists work, it is important to
/// do removal in the right order.
/// ShouldDelete defaults to true, and will cause the memory access to also be
/// deleted, not just removed.
void MemorySSA::removeFromLists(MemoryAccess *MA, bool ShouldDelete) {
BasicBlock *BB = MA->getBlock();
// The access list owns the reference, so we erase it from the non-owning list
// first.
if (!isa<MemoryUse>(MA)) {
auto DefsIt = PerBlockDefs.find(BB);
std::unique_ptr<DefsList> &Defs = DefsIt->second;
Defs->remove(*MA);
if (Defs->empty())
PerBlockDefs.erase(DefsIt);
}
// The erase call here will delete it. If we don't want it deleted, we call
// remove instead.
auto AccessIt = PerBlockAccesses.find(BB);
std::unique_ptr<AccessList> &Accesses = AccessIt->second;
if (ShouldDelete)
Accesses->erase(MA);
else
Accesses->remove(MA);
if (Accesses->empty()) {
PerBlockAccesses.erase(AccessIt);
BlockNumberingValid.erase(BB);
}
}
void MemorySSA::print(raw_ostream &OS) const {
MemorySSAAnnotatedWriter Writer(this);
F.print(OS, &Writer);
}
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
LLVM_DUMP_METHOD void MemorySSA::dump() const { print(dbgs()); }
#endif
void MemorySSA::verifyMemorySSA() const {
verifyOrderingDominationAndDefUses(F);
verifyDominationNumbers(F);
verifyPrevDefInPhis(F);
// Previously, the verification used to also verify that the clobberingAccess
// cached by MemorySSA is the same as the clobberingAccess found at a later
// query to AA. This does not hold true in general due to the current fragility
// of BasicAA which has arbitrary caps on the things it analyzes before giving
// up. As a result, transformations that are correct, will lead to BasicAA
// returning different Alias answers before and after that transformation.
// Invalidating MemorySSA is not an option, as the results in BasicAA can be so
// random, in the worst case we'd need to rebuild MemorySSA from scratch after
// every transformation, which defeats the purpose of using it. For such an
// example, see test4 added in D51960.
}
void MemorySSA::verifyPrevDefInPhis(Function &F) const {
#if !defined(NDEBUG) && defined(EXPENSIVE_CHECKS)
for (const BasicBlock &BB : F) {
if (MemoryPhi *Phi = getMemoryAccess(&BB)) {
for (unsigned I = 0, E = Phi->getNumIncomingValues(); I != E; ++I) {
auto *Pred = Phi->getIncomingBlock(I);
auto *IncAcc = Phi->getIncomingValue(I);
// If Pred has no unreachable predecessors, get last def looking at
// IDoms. If, while walkings IDoms, any of these has an unreachable
// predecessor, then the incoming def can be any access.
if (auto *DTNode = DT->getNode(Pred)) {
while (DTNode) {
if (auto *DefList = getBlockDefs(DTNode->getBlock())) {
auto *LastAcc = &*(--DefList->end());
assert(LastAcc == IncAcc &&
"Incorrect incoming access into phi.");
break;
}
DTNode = DTNode->getIDom();
}
} else {
// If Pred has unreachable predecessors, but has at least a Def, the
// incoming access can be the last Def in Pred, or it could have been
// optimized to LoE. After an update, though, the LoE may have been
// replaced by another access, so IncAcc may be any access.
// If Pred has unreachable predecessors and no Defs, incoming access
// should be LoE; However, after an update, it may be any access.
}
}
}
}
#endif
}
/// Verify that all of the blocks we believe to have valid domination numbers
/// actually have valid domination numbers.
void MemorySSA::verifyDominationNumbers(const Function &F) const {
#ifndef NDEBUG
if (BlockNumberingValid.empty())
return;
SmallPtrSet<const BasicBlock *, 16> ValidBlocks = BlockNumberingValid;
for (const BasicBlock &BB : F) {
if (!ValidBlocks.count(&BB))
continue;
ValidBlocks.erase(&BB);
const AccessList *Accesses = getBlockAccesses(&BB);
// It's correct to say an empty block has valid numbering.
if (!Accesses)
continue;
// Block numbering starts at 1.
unsigned long LastNumber = 0;
for (const MemoryAccess &MA : *Accesses) {
auto ThisNumberIter = BlockNumbering.find(&MA);
assert(ThisNumberIter != BlockNumbering.end() &&
"MemoryAccess has no domination number in a valid block!");
unsigned long ThisNumber = ThisNumberIter->second;
assert(ThisNumber > LastNumber &&
"Domination numbers should be strictly increasing!");
LastNumber = ThisNumber;
}
}
assert(ValidBlocks.empty() &&
"All valid BasicBlocks should exist in F -- dangling pointers?");
#endif
}
/// Verify ordering: the order and existence of MemoryAccesses matches the
/// order and existence of memory affecting instructions.
/// Verify domination: each definition dominates all of its uses.
/// Verify def-uses: the immediate use information - walk all the memory
/// accesses and verifying that, for each use, it appears in the appropriate
/// def's use list
void MemorySSA::verifyOrderingDominationAndDefUses(Function &F) const {
#if !defined(NDEBUG)
// Walk all the blocks, comparing what the lookups think and what the access
// lists think, as well as the order in the blocks vs the order in the access
// lists.
SmallVector<MemoryAccess *, 32> ActualAccesses;
SmallVector<MemoryAccess *, 32> ActualDefs;
for (BasicBlock &B : F) {
const AccessList *AL = getBlockAccesses(&B);
const auto *DL = getBlockDefs(&B);
MemoryPhi *Phi = getMemoryAccess(&B);
if (Phi) {
// Verify ordering.
ActualAccesses.push_back(Phi);
ActualDefs.push_back(Phi);
// Verify domination
for (const Use &U : Phi->uses())
assert(dominates(Phi, U) && "Memory PHI does not dominate it's uses");
#if defined(EXPENSIVE_CHECKS)
// Verify def-uses.
assert(Phi->getNumOperands() == static_cast<unsigned>(std::distance(
pred_begin(&B), pred_end(&B))) &&
"Incomplete MemoryPhi Node");
for (unsigned I = 0, E = Phi->getNumIncomingValues(); I != E; ++I) {
verifyUseInDefs(Phi->getIncomingValue(I), Phi);
assert(find(predecessors(&B), Phi->getIncomingBlock(I)) !=
pred_end(&B) &&
"Incoming phi block not a block predecessor");
}
#endif
}
for (Instruction &I : B) {
MemoryUseOrDef *MA = getMemoryAccess(&I);
assert((!MA || (AL && (isa<MemoryUse>(MA) || DL))) &&
"We have memory affecting instructions "
"in this block but they are not in the "
"access list or defs list");
if (MA) {
// Verify ordering.
ActualAccesses.push_back(MA);
if (MemoryAccess *MD = dyn_cast<MemoryDef>(MA)) {
// Verify ordering.
ActualDefs.push_back(MA);
// Verify domination.
for (const Use &U : MD->uses())
assert(dominates(MD, U) &&
"Memory Def does not dominate it's uses");
}
#if defined(EXPENSIVE_CHECKS)
// Verify def-uses.
verifyUseInDefs(MA->getDefiningAccess(), MA);
#endif
}
}
// Either we hit the assert, really have no accesses, or we have both
// accesses and an access list. Same with defs.
if (!AL && !DL)
continue;
// Verify ordering.
assert(AL->size() == ActualAccesses.size() &&
"We don't have the same number of accesses in the block as on the "
"access list");
assert((DL || ActualDefs.size() == 0) &&
"Either we should have a defs list, or we should have no defs");
assert((!DL || DL->size() == ActualDefs.size()) &&
"We don't have the same number of defs in the block as on the "
"def list");
auto ALI = AL->begin();
auto AAI = ActualAccesses.begin();
while (ALI != AL->end() && AAI != ActualAccesses.end()) {
assert(&*ALI == *AAI && "Not the same accesses in the same order");
++ALI;
++AAI;
}
ActualAccesses.clear();
if (DL) {
auto DLI = DL->begin();
auto ADI = ActualDefs.begin();
while (DLI != DL->end() && ADI != ActualDefs.end()) {
assert(&*DLI == *ADI && "Not the same defs in the same order");
++DLI;
++ADI;
}
}
ActualDefs.clear();
}
#endif
}
/// Verify the def-use lists in MemorySSA, by verifying that \p Use
/// appears in the use list of \p Def.
void MemorySSA::verifyUseInDefs(MemoryAccess *Def, MemoryAccess *Use) const {
#ifndef NDEBUG
// The live on entry use may cause us to get a NULL def here
if (!Def)
assert(isLiveOnEntryDef(Use) &&
"Null def but use not point to live on entry def");
else
assert(is_contained(Def->users(), Use) &&
"Did not find use in def's use list");
#endif
}
/// Perform a local numbering on blocks so that instruction ordering can be
/// determined in constant time.
/// TODO: We currently just number in order. If we numbered by N, we could
/// allow at least N-1 sequences of insertBefore or insertAfter (and at least
/// log2(N) sequences of mixed before and after) without needing to invalidate
/// the numbering.
void MemorySSA::renumberBlock(const BasicBlock *B) const {
// The pre-increment ensures the numbers really start at 1.
unsigned long CurrentNumber = 0;
const AccessList *AL = getBlockAccesses(B);
assert(AL != nullptr && "Asking to renumber an empty block");
for (const auto &I : *AL)
BlockNumbering[&I] = ++CurrentNumber;
BlockNumberingValid.insert(B);
}
/// Determine, for two memory accesses in the same block,
/// whether \p Dominator dominates \p Dominatee.
/// \returns True if \p Dominator dominates \p Dominatee.
bool MemorySSA::locallyDominates(const MemoryAccess *Dominator,
const MemoryAccess *Dominatee) const {
const BasicBlock *DominatorBlock = Dominator->getBlock();
assert((DominatorBlock == Dominatee->getBlock()) &&
"Asking for local domination when accesses are in different blocks!");
// A node dominates itself.
if (Dominatee == Dominator)
return true;
// When Dominatee is defined on function entry, it is not dominated by another
// memory access.
if (isLiveOnEntryDef(Dominatee))
return false;
// When Dominator is defined on function entry, it dominates the other memory
// access.
if (isLiveOnEntryDef(Dominator))
return true;
if (!BlockNumberingValid.count(DominatorBlock))
renumberBlock(DominatorBlock);
unsigned long DominatorNum = BlockNumbering.lookup(Dominator);
// All numbers start with 1
assert(DominatorNum != 0 && "Block was not numbered properly");
unsigned long DominateeNum = BlockNumbering.lookup(Dominatee);
assert(DominateeNum != 0 && "Block was not numbered properly");
return DominatorNum < DominateeNum;
}
bool MemorySSA::dominates(const MemoryAccess *Dominator,
const MemoryAccess *Dominatee) const {
if (Dominator == Dominatee)
return true;
if (isLiveOnEntryDef(Dominatee))
return false;
if (Dominator->getBlock() != Dominatee->getBlock())
return DT->dominates(Dominator->getBlock(), Dominatee->getBlock());
return locallyDominates(Dominator, Dominatee);
}
bool MemorySSA::dominates(const MemoryAccess *Dominator,
const Use &Dominatee) const {
if (MemoryPhi *MP = dyn_cast<MemoryPhi>(Dominatee.getUser())) {
BasicBlock *UseBB = MP->getIncomingBlock(Dominatee);
// The def must dominate the incoming block of the phi.
if (UseBB != Dominator->getBlock())
return DT->dominates(Dominator->getBlock(), UseBB);
// If the UseBB and the DefBB are the same, compare locally.
return locallyDominates(Dominator, cast<MemoryAccess>(Dominatee));
}
// If it's not a PHI node use, the normal dominates can already handle it.
return dominates(Dominator, cast<MemoryAccess>(Dominatee.getUser()));
}
const static char LiveOnEntryStr[] = "liveOnEntry";
void MemoryAccess::print(raw_ostream &OS) const {
switch (getValueID()) {
case MemoryPhiVal: return static_cast<const MemoryPhi *>(this)->print(OS);
case MemoryDefVal: return static_cast<const MemoryDef *>(this)->print(OS);
case MemoryUseVal: return static_cast<const MemoryUse *>(this)->print(OS);
}
llvm_unreachable("invalid value id");
}
void MemoryDef::print(raw_ostream &OS) const {
MemoryAccess *UO = getDefiningAccess();
auto printID = [&OS](MemoryAccess *A) {
if (A && A->getID())
OS << A->getID();
else
OS << LiveOnEntryStr;
};
OS << getID() << " = MemoryDef(";
printID(UO);
OS << ")";
if (isOptimized()) {
OS << "->";
printID(getOptimized());
if (Optional<AliasResult> AR = getOptimizedAccessType())
OS << " " << *AR;
}
}
void MemoryPhi::print(raw_ostream &OS) const {
bool First = true;
OS << getID() << " = MemoryPhi(";
for (const auto &Op : operands()) {
BasicBlock *BB = getIncomingBlock(Op);
MemoryAccess *MA = cast<MemoryAccess>(Op);
if (!First)
OS << ',';
else
First = false;
OS << '{';
if (BB->hasName())
OS << BB->getName();
else
BB->printAsOperand(OS, false);
OS << ',';
if (unsigned ID = MA->getID())
OS << ID;
else
OS << LiveOnEntryStr;
OS << '}';
}
OS << ')';
}
void MemoryUse::print(raw_ostream &OS) const {
MemoryAccess *UO = getDefiningAccess();
OS << "MemoryUse(";
if (UO && UO->getID())
OS << UO->getID();
else
OS << LiveOnEntryStr;
OS << ')';
if (Optional<AliasResult> AR = getOptimizedAccessType())
OS << " " << *AR;
}
void MemoryAccess::dump() const {
// Cannot completely remove virtual function even in release mode.
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
print(dbgs());
dbgs() << "\n";
#endif
}
char MemorySSAPrinterLegacyPass::ID = 0;
MemorySSAPrinterLegacyPass::MemorySSAPrinterLegacyPass() : FunctionPass(ID) {
initializeMemorySSAPrinterLegacyPassPass(*PassRegistry::getPassRegistry());
}
void MemorySSAPrinterLegacyPass::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
AU.addRequired<MemorySSAWrapperPass>();
}
bool MemorySSAPrinterLegacyPass::runOnFunction(Function &F) {
auto &MSSA = getAnalysis<MemorySSAWrapperPass>().getMSSA();
MSSA.print(dbgs());
if (VerifyMemorySSA)
MSSA.verifyMemorySSA();
return false;
}
AnalysisKey MemorySSAAnalysis::Key;
MemorySSAAnalysis::Result MemorySSAAnalysis::run(Function &F,
FunctionAnalysisManager &AM) {
auto &DT = AM.getResult<DominatorTreeAnalysis>(F);
auto &AA = AM.getResult<AAManager>(F);
return MemorySSAAnalysis::Result(std::make_unique<MemorySSA>(F, &AA, &DT));
}
bool MemorySSAAnalysis::Result::invalidate(
Function &F, const PreservedAnalyses &PA,
FunctionAnalysisManager::Invalidator &Inv) {
auto PAC = PA.getChecker<MemorySSAAnalysis>();
return !(PAC.preserved() || PAC.preservedSet<AllAnalysesOn<Function>>()) ||
Inv.invalidate<AAManager>(F, PA) ||
Inv.invalidate<DominatorTreeAnalysis>(F, PA);
}
PreservedAnalyses MemorySSAPrinterPass::run(Function &F,
FunctionAnalysisManager &AM) {
OS << "MemorySSA for function: " << F.getName() << "\n";
AM.getResult<MemorySSAAnalysis>(F).getMSSA().print(OS);
return PreservedAnalyses::all();
}
PreservedAnalyses MemorySSAVerifierPass::run(Function &F,
FunctionAnalysisManager &AM) {
AM.getResult<MemorySSAAnalysis>(F).getMSSA().verifyMemorySSA();
return PreservedAnalyses::all();
}
char MemorySSAWrapperPass::ID = 0;
MemorySSAWrapperPass::MemorySSAWrapperPass() : FunctionPass(ID) {
initializeMemorySSAWrapperPassPass(*PassRegistry::getPassRegistry());
}
void MemorySSAWrapperPass::releaseMemory() { MSSA.reset(); }
void MemorySSAWrapperPass::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
AU.addRequiredTransitive<DominatorTreeWrapperPass>();
AU.addRequiredTransitive<AAResultsWrapperPass>();
}
bool MemorySSAWrapperPass::runOnFunction(Function &F) {
auto &DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree();
auto &AA = getAnalysis<AAResultsWrapperPass>().getAAResults();
MSSA.reset(new MemorySSA(F, &AA, &DT));
return false;
}
void MemorySSAWrapperPass::verifyAnalysis() const {
if (VerifyMemorySSA)
MSSA->verifyMemorySSA();
}
void MemorySSAWrapperPass::print(raw_ostream &OS, const Module *M) const {
MSSA->print(OS);
}
MemorySSAWalker::MemorySSAWalker(MemorySSA *M) : MSSA(M) {}
/// Walk the use-def chains starting at \p StartingAccess and find
/// the MemoryAccess that actually clobbers Loc.
///
/// \returns our clobbering memory access
template <typename AliasAnalysisType>
MemoryAccess *
MemorySSA::ClobberWalkerBase<AliasAnalysisType>::getClobberingMemoryAccessBase(
MemoryAccess *StartingAccess, const MemoryLocation &Loc,
unsigned &UpwardWalkLimit) {
if (isa<MemoryPhi>(StartingAccess))
return StartingAccess;
auto *StartingUseOrDef = cast<MemoryUseOrDef>(StartingAccess);
if (MSSA->isLiveOnEntryDef(StartingUseOrDef))
return StartingUseOrDef;
Instruction *I = StartingUseOrDef->getMemoryInst();
// Conservatively, fences are always clobbers, so don't perform the walk if we
// hit a fence.
if (!isa<CallBase>(I) && I->isFenceLike())
return StartingUseOrDef;
UpwardsMemoryQuery Q;
Q.OriginalAccess = StartingUseOrDef;
Q.StartingLoc = Loc;
Q.Inst = I;
Q.IsCall = false;
// Unlike the other function, do not walk to the def of a def, because we are
// handed something we already believe is the clobbering access.
// We never set SkipSelf to true in Q in this method.
MemoryAccess *DefiningAccess = isa<MemoryUse>(StartingUseOrDef)
? StartingUseOrDef->getDefiningAccess()
: StartingUseOrDef;
MemoryAccess *Clobber =
Walker.findClobber(DefiningAccess, Q, UpwardWalkLimit);
LLVM_DEBUG(dbgs() << "Starting Memory SSA clobber for " << *I << " is ");
LLVM_DEBUG(dbgs() << *StartingUseOrDef << "\n");
LLVM_DEBUG(dbgs() << "Final Memory SSA clobber for " << *I << " is ");
LLVM_DEBUG(dbgs() << *Clobber << "\n");
return Clobber;
}
template <typename AliasAnalysisType>
MemoryAccess *
MemorySSA::ClobberWalkerBase<AliasAnalysisType>::getClobberingMemoryAccessBase(
MemoryAccess *MA, unsigned &UpwardWalkLimit, bool SkipSelf) {
auto *StartingAccess = dyn_cast<MemoryUseOrDef>(MA);
// If this is a MemoryPhi, we can't do anything.
if (!StartingAccess)
return MA;
bool IsOptimized = false;
// If this is an already optimized use or def, return the optimized result.
// Note: Currently, we store the optimized def result in a separate field,
// since we can't use the defining access.
if (StartingAccess->isOptimized()) {
if (!SkipSelf || !isa<MemoryDef>(StartingAccess))
return StartingAccess->getOptimized();
IsOptimized = true;
}
const Instruction *I = StartingAccess->getMemoryInst();
// We can't sanely do anything with a fence, since they conservatively clobber
// all memory, and have no locations to get pointers from to try to
// disambiguate.
if (!isa<CallBase>(I) && I->isFenceLike())
return StartingAccess;
UpwardsMemoryQuery Q(I, StartingAccess);
if (isUseTriviallyOptimizableToLiveOnEntry(*Walker.getAA(), I)) {
MemoryAccess *LiveOnEntry = MSSA->getLiveOnEntryDef();
StartingAccess->setOptimized(LiveOnEntry);
StartingAccess->setOptimizedAccessType(None);
return LiveOnEntry;
}
MemoryAccess *OptimizedAccess;
if (!IsOptimized) {
// Start with the thing we already think clobbers this location
MemoryAccess *DefiningAccess = StartingAccess->getDefiningAccess();
// At this point, DefiningAccess may be the live on entry def.
// If it is, we will not get a better result.
if (MSSA->isLiveOnEntryDef(DefiningAccess)) {
StartingAccess->setOptimized(DefiningAccess);
StartingAccess->setOptimizedAccessType(None);
return DefiningAccess;
}
OptimizedAccess = Walker.findClobber(DefiningAccess, Q, UpwardWalkLimit);
StartingAccess->setOptimized(OptimizedAccess);
if (MSSA->isLiveOnEntryDef(OptimizedAccess))
StartingAccess->setOptimizedAccessType(None);
else if (Q.AR == MustAlias)
StartingAccess->setOptimizedAccessType(MustAlias);
} else
OptimizedAccess = StartingAccess->getOptimized();
LLVM_DEBUG(dbgs() << "Starting Memory SSA clobber for " << *I << " is ");
LLVM_DEBUG(dbgs() << *StartingAccess << "\n");
LLVM_DEBUG(dbgs() << "Optimized Memory SSA clobber for " << *I << " is ");
LLVM_DEBUG(dbgs() << *OptimizedAccess << "\n");
MemoryAccess *Result;
if (SkipSelf && isa<MemoryPhi>(OptimizedAccess) &&
isa<MemoryDef>(StartingAccess) && UpwardWalkLimit) {
assert(isa<MemoryDef>(Q.OriginalAccess));
Q.SkipSelfAccess = true;
Result = Walker.findClobber(OptimizedAccess, Q, UpwardWalkLimit);
} else
Result = OptimizedAccess;
LLVM_DEBUG(dbgs() << "Result Memory SSA clobber [SkipSelf = " << SkipSelf);
LLVM_DEBUG(dbgs() << "] for " << *I << " is " << *Result << "\n");
return Result;
}
MemoryAccess *
DoNothingMemorySSAWalker::getClobberingMemoryAccess(MemoryAccess *MA) {
if (auto *Use = dyn_cast<MemoryUseOrDef>(MA))
return Use->getDefiningAccess();
return MA;
}
MemoryAccess *DoNothingMemorySSAWalker::getClobberingMemoryAccess(
MemoryAccess *StartingAccess, const MemoryLocation &) {
if (auto *Use = dyn_cast<MemoryUseOrDef>(StartingAccess))
return Use->getDefiningAccess();
return StartingAccess;
}
void MemoryPhi::deleteMe(DerivedUser *Self) {
delete static_cast<MemoryPhi *>(Self);
}
void MemoryDef::deleteMe(DerivedUser *Self) {
delete static_cast<MemoryDef *>(Self);
}
void MemoryUse::deleteMe(DerivedUser *Self) {
delete static_cast<MemoryUse *>(Self);
}
Reference in New Issue View Git Blame Copy Permalink