/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp

Bug Summary

File:	llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
Warning:	line 490, column 25 Called C++ object pointer is null

Annotated Source Code

Press '?' to see keyboard shortcuts

Show analyzer invocation

clang -cc1 -cc1 -triple x86_64-pc-linux-gnu -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name MemCpyOptimizer.cpp -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=cplusplus -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -analyzer-config-compatibility-mode=true -mrelocation-model pic -pic-level 2 -mframe-pointer=none -fmath-errno -fno-rounding-math -mconstructor-aliases -munwind-tables -target-cpu x86-64 -tune-cpu generic -debugger-tuning=gdb -ffunction-sections -fdata-sections -fcoverage-compilation-dir=/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/build-llvm/lib/Transforms/Scalar -resource-dir /usr/lib/llvm-14/lib/clang/14.0.0 -D _GNU_SOURCE -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D __STDC_LIMIT_MACROS -I /build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/build-llvm/lib/Transforms/Scalar -I /build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/lib/Transforms/Scalar -I /build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/build-llvm/include -I /build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/include -D NDEBUG -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/c++/10 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/x86_64-linux-gnu/c++/10 -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../include/c++/10/backward -internal-isystem /usr/lib/llvm-14/lib/clang/14.0.0/include -internal-isystem /usr/local/include -internal-isystem /usr/lib/gcc/x86_64-linux-gnu/10/../../../../x86_64-linux-gnu/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O2 -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers -Wno-long-long -Wno-maybe-uninitialized -Wno-class-memaccess -Wno-redundant-move -Wno-pessimizing-move -Wno-noexcept-type -Wno-comment -std=c++14 -fdeprecated-macro -fdebug-compilation-dir=/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/build-llvm/lib/Transforms/Scalar -fdebug-prefix-map=/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e=. -ferror-limit 19 -fvisibility-inlines-hidden -stack-protector 2 -fgnuc-version=4.2.1 -vectorize-loops -vectorize-slp -analyzer-output=html -analyzer-config stable-report-filename=true -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /tmp/scan-build-2021-09-04-040900-46481-1 -x c++ /build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp

/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp

→

1//===- MemCpyOptimizer.cpp - Optimize use of memcpy and friends -----------===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This pass performs various transformations related to eliminating memcpy
10// calls, or transforming sets of stores into memset's.
11//
12//===----------------------------------------------------------------------===//

14#include "llvm/Transforms/Scalar/MemCpyOptimizer.h"
15#include "llvm/ADT/DenseSet.h"
16#include "llvm/ADT/None.h"
17#include "llvm/ADT/STLExtras.h"
18#include "llvm/ADT/SmallVector.h"
19#include "llvm/ADT/Statistic.h"
20#include "llvm/ADT/iterator_range.h"
21#include "llvm/Analysis/AliasAnalysis.h"
22#include "llvm/Analysis/AssumptionCache.h"
23#include "llvm/Analysis/GlobalsModRef.h"
24#include "llvm/Analysis/Loads.h"
25#include "llvm/Analysis/MemoryLocation.h"
26#include "llvm/Analysis/MemorySSA.h"
27#include "llvm/Analysis/MemorySSAUpdater.h"
28#include "llvm/Analysis/TargetLibraryInfo.h"
29#include "llvm/Analysis/ValueTracking.h"
30#include "llvm/IR/Argument.h"
31#include "llvm/IR/BasicBlock.h"
32#include "llvm/IR/Constants.h"
33#include "llvm/IR/DataLayout.h"
34#include "llvm/IR/DerivedTypes.h"
35#include "llvm/IR/Dominators.h"
36#include "llvm/IR/Function.h"
37#include "llvm/IR/GetElementPtrTypeIterator.h"
38#include "llvm/IR/GlobalVariable.h"
39#include "llvm/IR/IRBuilder.h"
40#include "llvm/IR/InstrTypes.h"
41#include "llvm/IR/Instruction.h"
42#include "llvm/IR/Instructions.h"
43#include "llvm/IR/IntrinsicInst.h"
44#include "llvm/IR/Intrinsics.h"
45#include "llvm/IR/LLVMContext.h"
46#include "llvm/IR/Module.h"
47#include "llvm/IR/Operator.h"
48#include "llvm/IR/PassManager.h"
49#include "llvm/IR/Type.h"
50#include "llvm/IR/User.h"
51#include "llvm/IR/Value.h"
52#include "llvm/InitializePasses.h"
53#include "llvm/Pass.h"
54#include "llvm/Support/Casting.h"
55#include "llvm/Support/Debug.h"
56#include "llvm/Support/MathExtras.h"
57#include "llvm/Support/raw_ostream.h"
58#include "llvm/Transforms/Scalar.h"
59#include "llvm/Transforms/Utils/Local.h"
60#include <algorithm>
61#include <cassert>
62#include <cstdint>
63#include <utility>

65using namespace llvm;

67#define DEBUG_TYPE"memcpyopt" "memcpyopt"

69static cl::opt<bool> EnableMemCpyOptWithoutLibcalls(
  "enable-memcpyopt-without-libcalls", cl::init(false), cl::Hidden,
  cl::ZeroOrMore,
  cl::desc("Enable memcpyopt even when libcalls are disabled"));

74STATISTIC(NumMemCpyInstr, "Number of memcpy instructions deleted")static llvm::Statistic NumMemCpyInstr = {"memcpyopt", "NumMemCpyInstr"
, "Number of memcpy instructions deleted"};
75STATISTIC(NumMemSetInfer, "Number of memsets inferred")static llvm::Statistic NumMemSetInfer = {"memcpyopt", "NumMemSetInfer"
, "Number of memsets inferred"};
76STATISTIC(NumMoveToCpy,   "Number of memmoves converted to memcpy")static llvm::Statistic NumMoveToCpy = {"memcpyopt", "NumMoveToCpy"
, "Number of memmoves converted to memcpy"};
77STATISTIC(NumCpyToSet,    "Number of memcpys converted to memset")static llvm::Statistic NumCpyToSet = {"memcpyopt", "NumCpyToSet"
, "Number of memcpys converted to memset"};
78STATISTIC(NumCallSlot,    "Number of call slot optimizations performed")static llvm::Statistic NumCallSlot = {"memcpyopt", "NumCallSlot"
, "Number of call slot optimizations performed"};

80namespace {

82/// Represents a range of memset'd bytes with the ByteVal value.
83/// This allows us to analyze stores like:
84///   store 0 -> P+1
85///   store 0 -> P+0
86///   store 0 -> P+3
87///   store 0 -> P+2
88/// which sometimes happens with stores to arrays of structs etc.  When we see
89/// the first store, we make a range [1, 2).  The second store extends the range
90/// to [0, 2).  The third makes a new range [2, 3).  The fourth store joins the
91/// two ranges into [0, 3) which is memset'able.
92struct MemsetRange {
// Start/End - A semi range that describes the span that this range covers.
// The range is closed at the start and open at the end: [Start, End).
int64_t Start, End;

/// StartPtr - The getelementptr instruction that points to the start of the
/// range.
Value *StartPtr;

/// Alignment - The known alignment of the first store.
unsigned Alignment;

/// TheStores - The actual stores that make up this range.
SmallVector<Instruction*, 16> TheStores;

bool isProfitableToUseMemset(const DataLayout &DL) const;
108};

110} // end anonymous namespace

112bool MemsetRange::isProfitableToUseMemset(const DataLayout &DL) const {
// If we found more than 4 stores to merge or 16 bytes, use memset.
if (TheStores.size() >= 4 || End-Start >= 16) return true;

// If there is nothing to merge, don't do anything.
if (TheStores.size() < 2) return false;

// If any of the stores are a memset, then it is always good to extend the
// memset.
for (Instruction *SI : TheStores)
  if (!isa<StoreInst>(SI))
    return true;

// Assume that the code generator is capable of merging pairs of stores
// together if it wants to.
if (TheStores.size() == 2) return false;

// If we have fewer than 8 stores, it can still be worthwhile to do this.
// For example, merging 4 i8 stores into an i32 store is useful almost always.
// However, merging 2 32-bit stores isn't useful on a 32-bit architecture (the
// memset will be split into 2 32-bit stores anyway) and doing so can
// pessimize the llvm optimizer.
//
// Since we don't have perfect knowledge here, make some assumptions: assume
// the maximum GPR width is the same size as the largest legal integer
// size. If so, check to see whether we will end up actually reducing the
// number of stores used.
unsigned Bytes = unsigned(End-Start);
unsigned MaxIntSize = DL.getLargestLegalIntTypeSizeInBits() / 8;
if (MaxIntSize == 0)
  MaxIntSize = 1;
unsigned NumPointerStores = Bytes / MaxIntSize;

// Assume the remaining bytes if any are done a byte at a time.
unsigned NumByteStores = Bytes % MaxIntSize;

// If we will reduce the # stores (according to this heuristic), do the
// transformation.  This encourages merging 4 x i8 -> i32 and 2 x i16 -> i32
// etc.
return TheStores.size() > NumPointerStores+NumByteStores;
152}

154namespace {

156class MemsetRanges {
using range_iterator = SmallVectorImpl<MemsetRange>::iterator;

/// A sorted list of the memset ranges.
SmallVector<MemsetRange, 8> Ranges;

const DataLayout &DL;

164public:
MemsetRanges(const DataLayout &DL) : DL(DL) {}
16
←
Returning without writing to 'this->Ranges.Size', which participates in a condition later→

using const_iterator = SmallVectorImpl<MemsetRange>::const_iterator;

const_iterator begin() const { return Ranges.begin(); }
const_iterator end() const { return Ranges.end(); }
bool empty() const { return Ranges.empty(); }
22
←
Calling 'SmallVectorBase::empty'→
25
←
Returning from 'SmallVectorBase::empty'→
26
←
Returning zero, which participates in a condition later→

void addInst(int64_t OffsetFromFirst, Instruction *Inst) {
  if (StoreInst *SI = dyn_cast<StoreInst>(Inst))
    addStore(OffsetFromFirst, SI);
  else
    addMemSet(OffsetFromFirst, cast<MemSetInst>(Inst));
}

void addStore(int64_t OffsetFromFirst, StoreInst *SI) {
  int64_t StoreSize = DL.getTypeStoreSize(SI->getOperand(0)->getType());

  addRange(OffsetFromFirst, StoreSize, SI->getPointerOperand(),
           SI->getAlign().value(), SI);
}

void addMemSet(int64_t OffsetFromFirst, MemSetInst *MSI) {
  int64_t Size = cast<ConstantInt>(MSI->getLength())->getZExtValue();
  addRange(OffsetFromFirst, Size, MSI->getDest(), MSI->getDestAlignment(), MSI);
}

void addRange(int64_t Start, int64_t Size, Value *Ptr,
              unsigned Alignment, Instruction *Inst);
194};

196} // end anonymous namespace

198/// Add a new store to the MemsetRanges data structure.  This adds a
199/// new range for the specified store at the specified offset, merging into
200/// existing ranges as appropriate.
201void MemsetRanges::addRange(int64_t Start, int64_t Size, Value *Ptr,
                          unsigned Alignment, Instruction *Inst) {
int64_t End = Start+Size;

range_iterator I = partition_point(
    Ranges, [=](const MemsetRange &O) { return O.End < Start; });

// We now know that I == E, in which case we didn't find anything to merge
// with, or that Start <= I->End.  If End < I->Start or I == E, then we need
// to insert a new range.  Handle this now.
if (I == Ranges.end() || End < I->Start) {
  MemsetRange &R = *Ranges.insert(I, MemsetRange());
  R.Start        = Start;
  R.End          = End;
  R.StartPtr     = Ptr;
  R.Alignment    = Alignment;
  R.TheStores.push_back(Inst);
  return;
}

// This store overlaps with I, add it.
I->TheStores.push_back(Inst);

// At this point, we may have an interval that completely contains our store.
// If so, just add it to the interval and return.
if (I->Start <= Start && I->End >= End)
  return;

// Now we know that Start <= I->End and End >= I->Start so the range overlaps
// but is not entirely contained within the range.

// See if the range extends the start of the range.  In this case, it couldn't
// possibly cause it to join the prior range, because otherwise we would have
// stopped on *it*.
if (Start < I->Start) {
  I->Start = Start;
  I->StartPtr = Ptr;
  I->Alignment = Alignment;
}

// Now we know that Start <= I->End and Start >= I->Start (so the startpoint
// is in or right at the end of I), and that End >= I->Start.  Extend I out to
// End.
if (End > I->End) {
  I->End = End;
  range_iterator NextI = I;
  while (++NextI != Ranges.end() && End >= NextI->Start) {
    // Merge the range in.
    I->TheStores.append(NextI->TheStores.begin(), NextI->TheStores.end());
    if (NextI->End > I->End)
      I->End = NextI->End;
    Ranges.erase(NextI);
    NextI = I;
  }
}
256}

258//===----------------------------------------------------------------------===//
259//                         MemCpyOptLegacyPass Pass
260//===----------------------------------------------------------------------===//

262namespace {

264class MemCpyOptLegacyPass : public FunctionPass {
MemCpyOptPass Impl;

267public:
static char ID; // Pass identification, replacement for typeid

MemCpyOptLegacyPass() : FunctionPass(ID) {
  initializeMemCpyOptLegacyPassPass(*PassRegistry::getPassRegistry());
}

bool runOnFunction(Function &F) override;

276private:
// This transformation requires dominator postdominator info
void getAnalysisUsage(AnalysisUsage &AU) const override {
  AU.setPreservesCFG();
  AU.addRequired<AssumptionCacheTracker>();
  AU.addRequired<DominatorTreeWrapperPass>();
  AU.addPreserved<DominatorTreeWrapperPass>();
  AU.addPreserved<GlobalsAAWrapperPass>();
  AU.addRequired<TargetLibraryInfoWrapperPass>();
  AU.addRequired<AAResultsWrapperPass>();
  AU.addPreserved<AAResultsWrapperPass>();
  AU.addRequired<MemorySSAWrapperPass>();
  AU.addPreserved<MemorySSAWrapperPass>();
}
290};

292} // end anonymous namespace

294char MemCpyOptLegacyPass::ID = 0;

296/// The public interface to this file...
297FunctionPass *llvm::createMemCpyOptPass() { return new MemCpyOptLegacyPass(); }

299INITIALIZE_PASS_BEGIN(MemCpyOptLegacyPass, "memcpyopt", "MemCpy Optimization",static void *initializeMemCpyOptLegacyPassPassOnce(PassRegistry
 &Registry) {
                    false, false)static void *initializeMemCpyOptLegacyPassPassOnce(PassRegistry
 &Registry) {
301INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)initializeAssumptionCacheTrackerPass(Registry);
302INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)initializeDominatorTreeWrapperPassPass(Registry);
303INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)initializeTargetLibraryInfoWrapperPassPass(Registry);
304INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)initializeAAResultsWrapperPassPass(Registry);
305INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass)initializeGlobalsAAWrapperPassPass(Registry);
306INITIALIZE_PASS_DEPENDENCY(MemorySSAWrapperPass)initializeMemorySSAWrapperPassPass(Registry);
307INITIALIZE_PASS_END(MemCpyOptLegacyPass, "memcpyopt", "MemCpy Optimization",PassInfo *PI = new PassInfo( "MemCpy Optimization", "memcpyopt"
, &MemCpyOptLegacyPass::ID, PassInfo::NormalCtor_t(callDefaultCtor
<MemCpyOptLegacyPass>), false, false); Registry.registerPass
(*PI, true); return PI; } static llvm::once_flag InitializeMemCpyOptLegacyPassPassFlag
; void llvm::initializeMemCpyOptLegacyPassPass(PassRegistry &
Registry) { llvm::call_once(InitializeMemCpyOptLegacyPassPassFlag
, initializeMemCpyOptLegacyPassPassOnce, std::ref(Registry));
 }
                  false, false)PassInfo *PI = new PassInfo( "MemCpy Optimization", "memcpyopt"
, &MemCpyOptLegacyPass::ID, PassInfo::NormalCtor_t(callDefaultCtor
<MemCpyOptLegacyPass>), false, false); Registry.registerPass
(*PI, true); return PI; } static llvm::once_flag InitializeMemCpyOptLegacyPassPassFlag
; void llvm::initializeMemCpyOptLegacyPassPass(PassRegistry &
Registry) { llvm::call_once(InitializeMemCpyOptLegacyPassPassFlag
, initializeMemCpyOptLegacyPassPassOnce, std::ref(Registry));
 }

310// Check that V is either not accessible by the caller, or unwinding cannot
311// occur between Start and End.
312static bool mayBeVisibleThroughUnwinding(Value *V, Instruction *Start,
                                       Instruction *End) {
assert(Start->getParent() == End->getParent() && "Must be in same block")(static_cast<void> (0));
if (!Start->getFunction()->doesNotThrow() &&
    !isa<AllocaInst>(getUnderlyingObject(V))) {
  for (const Instruction &I :
       make_range(Start->getIterator(), End->getIterator())) {
    if (I.mayThrow())
      return true;
  }
}
return false;
324}

326void MemCpyOptPass::eraseInstruction(Instruction *I) {
MSSAU->removeMemoryAccess(I);
I->eraseFromParent();
329}

331// Check for mod or ref of Loc between Start and End, excluding both boundaries.
332// Start and End must be in the same block
333static bool accessedBetween(AliasAnalysis &AA, MemoryLocation Loc,
                          const MemoryUseOrDef *Start,
                          const MemoryUseOrDef *End) {
assert(Start->getBlock() == End->getBlock() && "Only local supported")(static_cast<void> (0));
for (const MemoryAccess &MA :
     make_range(++Start->getIterator(), End->getIterator())) {
  if (isModOrRefSet(AA.getModRefInfo(cast<MemoryUseOrDef>(MA).getMemoryInst(),
                                     Loc)))
    return true;
}
return false;
344}

346// Check for mod of Loc between Start and End, excluding both boundaries.
347// Start and End can be in different blocks.
348static bool writtenBetween(MemorySSA *MSSA, MemoryLocation Loc,
                         const MemoryUseOrDef *Start,
                         const MemoryUseOrDef *End) {
// TODO: Only walk until we hit Start.
MemoryAccess *Clobber = MSSA->getWalker()->getClobberingMemoryAccess(
    End->getDefiningAccess(), Loc);
return !MSSA->dominates(Clobber, Start);
355}

357/// When scanning forward over instructions, we look for some other patterns to
358/// fold away. In particular, this looks for stores to neighboring locations of
359/// memory. If it sees enough consecutive ones, it attempts to merge them
360/// together into a memcpy/memset.
361Instruction *MemCpyOptPass::tryMergingIntoMemset(Instruction *StartInst,
                                               Value *StartPtr,
                                               Value *ByteVal) {
const DataLayout &DL = StartInst->getModule()->getDataLayout();

// Okay, so we now have a single store that can be splatable.  Scan to find
// all subsequent stores of the same value to offset from the same pointer.
// Join these together into ranges, so we can decide whether contiguous blocks
// are stored.
MemsetRanges Ranges(DL);
15
←
Calling constructor for 'MemsetRanges'→
17
←
Returning from constructor for 'MemsetRanges'→

BasicBlock::iterator BI(StartInst);

// Keeps track of the last memory use or def before the insertion point for
// the new memset. The new MemoryDef for the inserted memsets will be inserted
// after MemInsertPoint. It points to either LastMemDef or to the last user
// before the insertion point of the memset, if there are any such users.
MemoryUseOrDef *MemInsertPoint = nullptr;
18
←
'MemInsertPoint' initialized to a null pointer value→
// Keeps track of the last MemoryDef between StartInst and the insertion point
// for the new memset. This will become the defining access of the inserted
// memsets.
MemoryDef *LastMemDef = nullptr;
for (++BI; !BI->isTerminator(); ++BI) {
19
←
Assuming the condition is false→
20
←
Loop condition is false. Execution continues on line 451→
  auto *CurrentAcc = cast_or_null<MemoryUseOrDef>(
      MSSAU->getMemorySSA()->getMemoryAccess(&*BI));
  if (CurrentAcc) {
    MemInsertPoint = CurrentAcc;
    if (auto *CurrentDef = dyn_cast<MemoryDef>(CurrentAcc))
      LastMemDef = CurrentDef;
  }

  // Calls that only access inaccessible memory do not block merging
  // accessible stores.
  if (auto *CB = dyn_cast<CallBase>(BI)) {
    if (CB->onlyAccessesInaccessibleMemory())
      continue;
  }

  if (!isa<StoreInst>(BI) && !isa<MemSetInst>(BI)) {
    // If the instruction is readnone, ignore it, otherwise bail out.  We
    // don't even allow readonly here because we don't want something like:
    // A[1] = 2; strlen(A); A[2] = 2; -> memcpy(A, ...); strlen(A).
    if (BI->mayWriteToMemory() || BI->mayReadFromMemory())
      break;
    continue;
  }

  if (StoreInst *NextStore = dyn_cast<StoreInst>(BI)) {
    // If this is a store, see if we can merge it in.
    if (!NextStore->isSimple()) break;

    Value *StoredVal = NextStore->getValueOperand();

    // Don't convert stores of non-integral pointer types to memsets (which
    // stores integers).
    if (DL.isNonIntegralPointerType(StoredVal->getType()->getScalarType()))
      break;

    // Check to see if this stored value is of the same byte-splattable value.
    Value *StoredByte = isBytewiseValue(StoredVal, DL);
    if (isa<UndefValue>(ByteVal) && StoredByte)
      ByteVal = StoredByte;
    if (ByteVal != StoredByte)
      break;

    // Check to see if this store is to a constant offset from the start ptr.
    Optional<int64_t> Offset =
        isPointerOffset(StartPtr, NextStore->getPointerOperand(), DL);
    if (!Offset)
      break;

    Ranges.addStore(*Offset, NextStore);
  } else {
    MemSetInst *MSI = cast<MemSetInst>(BI);

    if (MSI->isVolatile() || ByteVal != MSI->getValue() ||
        !isa<ConstantInt>(MSI->getLength()))
      break;

    // Check to see if this store is to a constant offset from the start ptr.
    Optional<int64_t> Offset = isPointerOffset(StartPtr, MSI->getDest(), DL);
    if (!Offset)
      break;

    Ranges.addMemSet(*Offset, MSI);
  }
}

// If we have no ranges, then we just had a single store with nothing that
// could be merged in.  This is a very common case of course.
if (Ranges.empty())
21
←
Calling 'MemsetRanges::empty'→
27
←
Returning from 'MemsetRanges::empty'→
28
←
Taking false branch→
  return nullptr;

// If we had at least one store that could be merged in, add the starting
// store as well.  We try to avoid this unless there is at least something
// interesting as a small compile-time optimization.
Ranges.addInst(0, StartInst);

// If we create any memsets, we put it right before the first instruction that
// isn't part of the memset block.  This ensure that the memset is dominated
// by any addressing instruction needed by the start of the block.
IRBuilder<> Builder(&*BI);

// Now that we have full information about ranges, loop over the ranges and
// emit memset's for anything big enough to be worthwhile.
Instruction *AMemSet = nullptr;
for (const MemsetRange &Range : Ranges) {
29
←
Assuming '__begin1' is not equal to '__end1'→
  if (Range.TheStores.size() == 1) continue;
30
←
Assuming the condition is false→
31
←
Taking false branch→

  // If it is profitable to lower this range to memset, do so now.
  if (!Range.isProfitableToUseMemset(DL))
32
←
Assuming the condition is false→
33
←
Taking false branch→
    continue;

  // Otherwise, we do want to transform this!  Create a new memset.
  // Get the starting pointer of the block.
  StartPtr = Range.StartPtr;

  AMemSet = Builder.CreateMemSet(StartPtr, ByteVal, Range.End - Range.Start,
                                 MaybeAlign(Range.Alignment));
  LLVM_DEBUG(dbgs() << "Replace stores:\n"; for (Instruction *SIdo { } while (false)
34
←
Loop condition is false.  Exiting loop→
                                                 : Range.TheStores) dbgs()do { } while (false)
                                            << *SI << '\n';do { } while (false)
             dbgs() << "With: " << *AMemSet << '\n')do { } while (false);
  if (!Range.TheStores.empty())
35
←
Taking false branch→
    AMemSet->setDebugLoc(Range.TheStores[0]->getDebugLoc());

  assert(LastMemDef && MemInsertPoint &&(static_cast<void> (0))
         "Both LastMemDef and MemInsertPoint need to be set")(static_cast<void> (0));
  auto *NewDef =
      cast<MemoryDef>(MemInsertPoint->getMemoryInst() == &*BI
36
←
Called C++ object pointer is null
                          ? MSSAU->createMemoryAccessBefore(
                                AMemSet, LastMemDef, MemInsertPoint)
                          : MSSAU->createMemoryAccessAfter(
                                AMemSet, LastMemDef, MemInsertPoint));
  MSSAU->insertDef(NewDef, /*RenameUses=*/true);
  LastMemDef = NewDef;
  MemInsertPoint = NewDef;

  // Zap all the stores.
  for (Instruction *SI : Range.TheStores)
    eraseInstruction(SI);

  ++NumMemSetInfer;
}

return AMemSet;
507}

509// This method try to lift a store instruction before position P.
510// It will lift the store and its argument + that anything that
511// may alias with these.
512// The method returns true if it was successful.
513bool MemCpyOptPass::moveUp(StoreInst *SI, Instruction *P, const LoadInst *LI) {
// If the store alias this position, early bail out.
MemoryLocation StoreLoc = MemoryLocation::get(SI);
if (isModOrRefSet(AA->getModRefInfo(P, StoreLoc)))
  return false;

// Keep track of the arguments of all instruction we plan to lift
// so we can make sure to lift them as well if appropriate.
DenseSet<Instruction*> Args;
if (auto *Ptr = dyn_cast<Instruction>(SI->getPointerOperand()))
  if (Ptr->getParent() == SI->getParent())
    Args.insert(Ptr);

// Instruction to lift before P.
SmallVector<Instruction *, 8> ToLift{SI};

// Memory locations of lifted instructions.
SmallVector<MemoryLocation, 8> MemLocs{StoreLoc};

// Lifted calls.
SmallVector<const CallBase *, 8> Calls;

const MemoryLocation LoadLoc = MemoryLocation::get(LI);

for (auto I = --SI->getIterator(), E = P->getIterator(); I != E; --I) {
  auto *C = &*I;

  // Make sure hoisting does not perform a store that was not guaranteed to
  // happen.
  if (!isGuaranteedToTransferExecutionToSuccessor(C))
    return false;

  bool MayAlias = isModOrRefSet(AA->getModRefInfo(C, None));

  bool NeedLift = false;
  if (Args.erase(C))
    NeedLift = true;
  else if (MayAlias) {
    NeedLift = llvm::any_of(MemLocs, [C, this](const MemoryLocation &ML) {
      return isModOrRefSet(AA->getModRefInfo(C, ML));
    });

    if (!NeedLift)
      NeedLift = llvm::any_of(Calls, [C, this](const CallBase *Call) {
        return isModOrRefSet(AA->getModRefInfo(C, Call));
      });
  }

  if (!NeedLift)
    continue;

  if (MayAlias) {
    // Since LI is implicitly moved downwards past the lifted instructions,
    // none of them may modify its source.
    if (isModSet(AA->getModRefInfo(C, LoadLoc)))
      return false;
    else if (const auto *Call = dyn_cast<CallBase>(C)) {
      // If we can't lift this before P, it's game over.
      if (isModOrRefSet(AA->getModRefInfo(P, Call)))
        return false;

      Calls.push_back(Call);
    } else if (isa<LoadInst>(C) || isa<StoreInst>(C) || isa<VAArgInst>(C)) {
      // If we can't lift this before P, it's game over.
      auto ML = MemoryLocation::get(C);
      if (isModOrRefSet(AA->getModRefInfo(P, ML)))
        return false;

      MemLocs.push_back(ML);
    } else
      // We don't know how to lift this instruction.
      return false;
  }

  ToLift.push_back(C);
  for (unsigned k = 0, e = C->getNumOperands(); k != e; ++k)
    if (auto *A = dyn_cast<Instruction>(C->getOperand(k))) {
      if (A->getParent() == SI->getParent()) {
        // Cannot hoist user of P above P
        if(A == P) return false;
        Args.insert(A);
      }
    }
}

// Find MSSA insertion point. Normally P will always have a corresponding
// memory access before which we can insert. However, with non-standard AA
// pipelines, there may be a mismatch between AA and MSSA, in which case we
// will scan for a memory access before P. In either case, we know for sure
// that at least the load will have a memory access.
// TODO: Simplify this once P will be determined by MSSA, in which case the
// discrepancy can no longer occur.
MemoryUseOrDef *MemInsertPoint = nullptr;
if (MemoryUseOrDef *MA = MSSAU->getMemorySSA()->getMemoryAccess(P)) {
  MemInsertPoint = cast<MemoryUseOrDef>(--MA->getIterator());
} else {
  const Instruction *ConstP = P;
  for (const Instruction &I : make_range(++ConstP->getReverseIterator(),
                                         ++LI->getReverseIterator())) {
    if (MemoryUseOrDef *MA = MSSAU->getMemorySSA()->getMemoryAccess(&I)) {
      MemInsertPoint = MA;
      break;
    }
  }
}

// We made it, we need to lift.
for (auto *I : llvm::reverse(ToLift)) {
  LLVM_DEBUG(dbgs() << "Lifting " << *I << " before " << *P << "\n")do { } while (false);
  I->moveBefore(P);
  assert(MemInsertPoint && "Must have found insert point")(static_cast<void> (0));
  if (MemoryUseOrDef *MA = MSSAU->getMemorySSA()->getMemoryAccess(I)) {
    MSSAU->moveAfter(MA, MemInsertPoint);
    MemInsertPoint = MA;
  }
}

return true;
631}

633bool MemCpyOptPass::processStore(StoreInst *SI, BasicBlock::iterator &BBI) {
if (!SI->isSimple()) return false;

// Avoid merging nontemporal stores since the resulting
// memcpy/memset would not be able to preserve the nontemporal hint.
// In theory we could teach how to propagate the !nontemporal metadata to
// memset calls. However, that change would force the backend to
// conservatively expand !nontemporal memset calls back to sequences of
// store instructions (effectively undoing the merging).
if (SI->getMetadata(LLVMContext::MD_nontemporal))
  return false;

const DataLayout &DL = SI->getModule()->getDataLayout();

Value *StoredVal = SI->getValueOperand();

// Not all the transforms below are correct for non-integral pointers, bail
// until we've audited the individual pieces.
if (DL.isNonIntegralPointerType(StoredVal->getType()->getScalarType()))
  return false;

// Load to store forwarding can be interpreted as memcpy.
if (LoadInst *LI = dyn_cast<LoadInst>(StoredVal)) {
  if (LI->isSimple() && LI->hasOneUse() &&
      LI->getParent() == SI->getParent()) {

    auto *T = LI->getType();
    // Don't introduce calls to memcpy/memmove intrinsics out of thin air if
    // the corresponding libcalls are not available.
    // TODO: We should really distinguish between libcall availability and
    // our ability to introduce intrinsics.
    if (T->isAggregateType() &&
        (EnableMemCpyOptWithoutLibcalls ||
         (TLI->has(LibFunc_memcpy) && TLI->has(LibFunc_memmove)))) {
      MemoryLocation LoadLoc = MemoryLocation::get(LI);

      // We use alias analysis to check if an instruction may store to
      // the memory we load from in between the load and the store. If
      // such an instruction is found, we try to promote there instead
      // of at the store position.
      // TODO: Can use MSSA for this.
      Instruction *P = SI;
      for (auto &I : make_range(++LI->getIterator(), SI->getIterator())) {
        if (isModSet(AA->getModRefInfo(&I, LoadLoc))) {
          P = &I;
          break;
        }
      }

      // We found an instruction that may write to the loaded memory.
      // We can try to promote at this position instead of the store
      // position if nothing aliases the store memory after this and the store
      // destination is not in the range.
      if (P && P != SI) {
        if (!moveUp(SI, P, LI))
          P = nullptr;
      }

      // If a valid insertion position is found, then we can promote
      // the load/store pair to a memcpy.
      if (P) {
        // If we load from memory that may alias the memory we store to,
        // memmove must be used to preserve semantic. If not, memcpy can
        // be used. Also, if we load from constant memory, memcpy can be used
        // as the constant memory won't be modified.
        bool UseMemMove = false;
        if (isModSet(AA->getModRefInfo(SI, LoadLoc)))
          UseMemMove = true;

        uint64_t Size = DL.getTypeStoreSize(T);

        IRBuilder<> Builder(P);
        Instruction *M;
        if (UseMemMove)
          M = Builder.CreateMemMove(
              SI->getPointerOperand(), SI->getAlign(),
              LI->getPointerOperand(), LI->getAlign(), Size);
        else
          M = Builder.CreateMemCpy(
              SI->getPointerOperand(), SI->getAlign(),
              LI->getPointerOperand(), LI->getAlign(), Size);

        LLVM_DEBUG(dbgs() << "Promoting " << *LI << " to " << *SI << " => "do { } while (false)
                          << *M << "\n")do { } while (false);

        auto *LastDef =
            cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(SI));
        auto *NewAccess = MSSAU->createMemoryAccessAfter(M, LastDef, LastDef);
        MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/true);

        eraseInstruction(SI);
        eraseInstruction(LI);
        ++NumMemCpyInstr;

        // Make sure we do not invalidate the iterator.
        BBI = M->getIterator();
        return true;
      }
    }

    // Detect cases where we're performing call slot forwarding, but
    // happen to be using a load-store pair to implement it, rather than
    // a memcpy.
    CallInst *C = nullptr;
    if (auto *LoadClobber = dyn_cast<MemoryUseOrDef>(
            MSSA->getWalker()->getClobberingMemoryAccess(LI))) {
      // The load most post-dom the call. Limit to the same block for now.
      // TODO: Support non-local call-slot optimization?
      if (LoadClobber->getBlock() == SI->getParent())
        C = dyn_cast_or_null<CallInst>(LoadClobber->getMemoryInst());
    }

    if (C) {
      // Check that nothing touches the dest of the "copy" between
      // the call and the store.
      MemoryLocation StoreLoc = MemoryLocation::get(SI);
      if (accessedBetween(*AA, StoreLoc, MSSA->getMemoryAccess(C),
                          MSSA->getMemoryAccess(SI)))
        C = nullptr;
    }

    if (C) {
      bool changed = performCallSlotOptzn(
          LI, SI, SI->getPointerOperand()->stripPointerCasts(),
          LI->getPointerOperand()->stripPointerCasts(),
          DL.getTypeStoreSize(SI->getOperand(0)->getType()),
          commonAlignment(SI->getAlign(), LI->getAlign()), C);
      if (changed) {
        eraseInstruction(SI);
        eraseInstruction(LI);
        ++NumMemCpyInstr;
        return true;
      }
    }
  }
}

// The following code creates memset intrinsics out of thin air. Don't do
// this if the corresponding libfunc is not available.
// TODO: We should really distinguish between libcall availability and
// our ability to introduce intrinsics.
if (!(TLI->has(LibFunc_memset) || EnableMemCpyOptWithoutLibcalls))
  return false;

// There are two cases that are interesting for this code to handle: memcpy
// and memset.  Right now we only handle memset.

// Ensure that the value being stored is something that can be memset'able a
// byte at a time like "0" or "-1" or any width, as well as things like
// 0xA0A0A0A0 and 0.0.
auto *V = SI->getOperand(0);
if (Value *ByteVal = isBytewiseValue(V, DL)) {
  if (Instruction *I = tryMergingIntoMemset(SI, SI->getPointerOperand(),
                                            ByteVal)) {
    BBI = I->getIterator(); // Don't invalidate iterator.
    return true;
  }

  // If we have an aggregate, we try to promote it to memset regardless
  // of opportunity for merging as it can expose optimization opportunities
  // in subsequent passes.
  auto *T = V->getType();
  if (T->isAggregateType()) {
    uint64_t Size = DL.getTypeStoreSize(T);
    IRBuilder<> Builder(SI);
    auto *M = Builder.CreateMemSet(SI->getPointerOperand(), ByteVal, Size,
                                   SI->getAlign());

    LLVM_DEBUG(dbgs() << "Promoting " << *SI << " to " << *M << "\n")do { } while (false);

    // The newly inserted memset is immediately overwritten by the original
    // store, so we do not need to rename uses.
    auto *StoreDef = cast<MemoryDef>(MSSA->getMemoryAccess(SI));
    auto *NewAccess = MSSAU->createMemoryAccessBefore(
        M, StoreDef->getDefiningAccess(), StoreDef);
    MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/false);

    eraseInstruction(SI);
    NumMemSetInfer++;

    // Make sure we do not invalidate the iterator.
    BBI = M->getIterator();
    return true;
  }
}

return false;
820}

822bool MemCpyOptPass::processMemSet(MemSetInst *MSI, BasicBlock::iterator &BBI) {
// See if there is another memset or store neighboring this memset which
// allows us to widen out the memset to do a single larger store.
if (isa<ConstantInt>(MSI->getLength()) && !MSI->isVolatile())
12
←
Assuming the object is a 'ConstantInt'→
13
←
Taking true branch→
  if (Instruction *I = tryMergingIntoMemset(MSI, MSI->getDest(),
14
←
Calling 'MemCpyOptPass::tryMergingIntoMemset'→
                                            MSI->getValue())) {
    BBI = I->getIterator(); // Don't invalidate iterator.
    return true;
  }
return false;
832}

834/// Takes a memcpy and a call that it depends on,
835/// and checks for the possibility of a call slot optimization by having
836/// the call write its result directly into the destination of the memcpy.
837bool MemCpyOptPass::performCallSlotOptzn(Instruction *cpyLoad,
                                       Instruction *cpyStore, Value *cpyDest,
                                       Value *cpySrc, uint64_t cpyLen,
                                       Align cpyAlign, CallInst *C) {
// The general transformation to keep in mind is
//
//   call @func(..., src, ...)
//   memcpy(dest, src, ...)
//
// ->
//
//   memcpy(dest, src, ...)
//   call @func(..., dest, ...)
//
// Since moving the memcpy is technically awkward, we additionally check that
// src only holds uninitialized values at the moment of the call, meaning that
// the memcpy can be discarded rather than moved.

// Lifetime marks shouldn't be operated on.
if (Function *F = C->getCalledFunction())
  if (F->isIntrinsic() && F->getIntrinsicID() == Intrinsic::lifetime_start)
    return false;

// Require that src be an alloca.  This simplifies the reasoning considerably.
AllocaInst *srcAlloca = dyn_cast<AllocaInst>(cpySrc);
if (!srcAlloca)
  return false;

ConstantInt *srcArraySize = dyn_cast<ConstantInt>(srcAlloca->getArraySize());
if (!srcArraySize)
  return false;

const DataLayout &DL = cpyLoad->getModule()->getDataLayout();
uint64_t srcSize = DL.getTypeAllocSize(srcAlloca->getAllocatedType()) *
                   srcArraySize->getZExtValue();

if (cpyLen < srcSize)
  return false;

// Check that accessing the first srcSize bytes of dest will not cause a
// trap.  Otherwise the transform is invalid since it might cause a trap
// to occur earlier than it otherwise would.
if (!isDereferenceableAndAlignedPointer(cpyDest, Align(1), APInt(64, cpyLen),
                                        DL, C, DT))
  return false;

// Make sure that nothing can observe cpyDest being written early. There are
// a number of cases to consider:
//  1. cpyDest cannot be accessed between C and cpyStore as a precondition of
//     the transform.
//  2. C itself may not access cpyDest (prior to the transform). This is
//     checked further below.
//  3. If cpyDest is accessible to the caller of this function (potentially
//     captured and not based on an alloca), we need to ensure that we cannot
//     unwind between C and cpyStore. This is checked here.
//  4. If cpyDest is potentially captured, there may be accesses to it from
//     another thread. In this case, we need to check that cpyStore is
//     guaranteed to be executed if C is. As it is a non-atomic access, it
//     renders accesses from other threads undefined.
//     TODO: This is currently not checked.
if (mayBeVisibleThroughUnwinding(cpyDest, C, cpyStore))
  return false;

// Check that dest points to memory that is at least as aligned as src.
Align srcAlign = srcAlloca->getAlign();
bool isDestSufficientlyAligned = srcAlign <= cpyAlign;
// If dest is not aligned enough and we can't increase its alignment then
// bail out.
if (!isDestSufficientlyAligned && !isa<AllocaInst>(cpyDest))
  return false;

// Check that src is not accessed except via the call and the memcpy.  This
// guarantees that it holds only undefined values when passed in (so the final
// memcpy can be dropped), that it is not read or written between the call and
// the memcpy, and that writing beyond the end of it is undefined.
SmallVector<User *, 8> srcUseList(srcAlloca->users());
while (!srcUseList.empty()) {
  User *U = srcUseList.pop_back_val();

  if (isa<BitCastInst>(U) || isa<AddrSpaceCastInst>(U)) {
    append_range(srcUseList, U->users());
    continue;
  }
  if (GetElementPtrInst *G = dyn_cast<GetElementPtrInst>(U)) {
    if (!G->hasAllZeroIndices())
      return false;

    append_range(srcUseList, U->users());
    continue;
  }
  if (const IntrinsicInst *IT = dyn_cast<IntrinsicInst>(U))
    if (IT->isLifetimeStartOrEnd())
      continue;

  if (U != C && U != cpyLoad)
    return false;
}

// Check that src isn't captured by the called function since the
// transformation can cause aliasing issues in that case.
for (unsigned ArgI = 0, E = C->arg_size(); ArgI != E; ++ArgI)
  if (C->getArgOperand(ArgI) == cpySrc && !C->doesNotCapture(ArgI))
    return false;

// Since we're changing the parameter to the callsite, we need to make sure
// that what would be the new parameter dominates the callsite.
if (!DT->dominates(cpyDest, C)) {
  // Support moving a constant index GEP before the call.
  auto *GEP = dyn_cast<GetElementPtrInst>(cpyDest);
  if (GEP && GEP->hasAllConstantIndices() &&
      DT->dominates(GEP->getPointerOperand(), C))
    GEP->moveBefore(C);
  else
    return false;
}

// In addition to knowing that the call does not access src in some
// unexpected manner, for example via a global, which we deduce from
// the use analysis, we also need to know that it does not sneakily
// access dest.  We rely on AA to figure this out for us.
ModRefInfo MR = AA->getModRefInfo(C, cpyDest, LocationSize::precise(srcSize));
// If necessary, perform additional analysis.
if (isModOrRefSet(MR))
  MR = AA->callCapturesBefore(C, cpyDest, LocationSize::precise(srcSize), DT);
if (isModOrRefSet(MR))
  return false;

// We can't create address space casts here because we don't know if they're
// safe for the target.
if (cpySrc->getType()->getPointerAddressSpace() !=
    cpyDest->getType()->getPointerAddressSpace())
  return false;
for (unsigned ArgI = 0; ArgI < C->arg_size(); ++ArgI)
  if (C->getArgOperand(ArgI)->stripPointerCasts() == cpySrc &&
      cpySrc->getType()->getPointerAddressSpace() !=
          C->getArgOperand(ArgI)->getType()->getPointerAddressSpace())
    return false;

// All the checks have passed, so do the transformation.
bool changedArgument = false;
for (unsigned ArgI = 0; ArgI < C->arg_size(); ++ArgI)
  if (C->getArgOperand(ArgI)->stripPointerCasts() == cpySrc) {
    Value *Dest = cpySrc->getType() == cpyDest->getType() ?  cpyDest
      : CastInst::CreatePointerCast(cpyDest, cpySrc->getType(),
                                    cpyDest->getName(), C);
    changedArgument = true;
    if (C->getArgOperand(ArgI)->getType() == Dest->getType())
      C->setArgOperand(ArgI, Dest);
    else
      C->setArgOperand(ArgI, CastInst::CreatePointerCast(
                                 Dest, C->getArgOperand(ArgI)->getType(),
                                 Dest->getName(), C));
  }

if (!changedArgument)
  return false;

// If the destination wasn't sufficiently aligned then increase its alignment.
if (!isDestSufficientlyAligned) {
  assert(isa<AllocaInst>(cpyDest) && "Can only increase alloca alignment!")(static_cast<void> (0));
  cast<AllocaInst>(cpyDest)->setAlignment(srcAlign);
}

// Update AA metadata
// FIXME: MD_tbaa_struct and MD_mem_parallel_loop_access should also be
// handled here, but combineMetadata doesn't support them yet
unsigned KnownIDs[] = {LLVMContext::MD_tbaa, LLVMContext::MD_alias_scope,
                       LLVMContext::MD_noalias,
                       LLVMContext::MD_invariant_group,
                       LLVMContext::MD_access_group};
combineMetadata(C, cpyLoad, KnownIDs, true);

++NumCallSlot;
return true;
1011}

1013/// We've found that the (upward scanning) memory dependence of memcpy 'M' is
1014/// the memcpy 'MDep'. Try to simplify M to copy from MDep's input if we can.
1015bool MemCpyOptPass::processMemCpyMemCpyDependence(MemCpyInst *M,
                                                MemCpyInst *MDep) {
// We can only transforms memcpy's where the dest of one is the source of the
// other.
if (M->getSource() != MDep->getDest() || MDep->isVolatile())
  return false;

// If dep instruction is reading from our current input, then it is a noop
// transfer and substituting the input won't change this instruction.  Just
// ignore the input and let someone else zap MDep.  This handles cases like:
//    memcpy(a <- a)
//    memcpy(b <- a)
if (M->getSource() == MDep->getSource())
  return false;

// Second, the length of the memcpy's must be the same, or the preceding one
// must be larger than the following one.
if (MDep->getLength() != M->getLength()) {
  ConstantInt *MDepLen = dyn_cast<ConstantInt>(MDep->getLength());
  ConstantInt *MLen = dyn_cast<ConstantInt>(M->getLength());
  if (!MDepLen || !MLen || MDepLen->getZExtValue() < MLen->getZExtValue())
    return false;
}

// Verify that the copied-from memory doesn't change in between the two
// transfers.  For example, in:
//    memcpy(a <- b)
//    *b = 42;
//    memcpy(c <- a)
// It would be invalid to transform the second memcpy into memcpy(c <- b).
//
// TODO: If the code between M and MDep is transparent to the destination "c",
// then we could still perform the xform by moving M up to the first memcpy.
// TODO: It would be sufficient to check the MDep source up to the memcpy
// size of M, rather than MDep.
if (writtenBetween(MSSA, MemoryLocation::getForSource(MDep),
                   MSSA->getMemoryAccess(MDep), MSSA->getMemoryAccess(M)))
  return false;

// If the dest of the second might alias the source of the first, then the
// source and dest might overlap. In addition, if the source of the first
// points to constant memory, they won't overlap by definition. Otherwise, we
// still want to eliminate the intermediate value, but we have to generate a
// memmove instead of memcpy.
bool UseMemMove = false;
if (isModSet(AA->getModRefInfo(M, MemoryLocation::getForSource(MDep))))
  UseMemMove = true;

// If all checks passed, then we can transform M.
LLVM_DEBUG(dbgs() << "MemCpyOptPass: Forwarding memcpy->memcpy src:\n"do { } while (false)
                  << *MDep << '\n' << *M << '\n')do { } while (false);

// TODO: Is this worth it if we're creating a less aligned memcpy? For
// example we could be moving from movaps -> movq on x86.
IRBuilder<> Builder(M);
Instruction *NewM;
if (UseMemMove)
  NewM = Builder.CreateMemMove(M->getRawDest(), M->getDestAlign(),
                               MDep->getRawSource(), MDep->getSourceAlign(),
                               M->getLength(), M->isVolatile());
else if (isa<MemCpyInlineInst>(M)) {
  // llvm.memcpy may be promoted to llvm.memcpy.inline, but the converse is
  // never allowed since that would allow the latter to be lowered as a call
  // to an external function.
  NewM = Builder.CreateMemCpyInline(
      M->getRawDest(), M->getDestAlign(), MDep->getRawSource(),
      MDep->getSourceAlign(), M->getLength(), M->isVolatile());
} else
  NewM = Builder.CreateMemCpy(M->getRawDest(), M->getDestAlign(),
                              MDep->getRawSource(), MDep->getSourceAlign(),
                              M->getLength(), M->isVolatile());

assert(isa<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(M)))(static_cast<void> (0));
auto *LastDef = cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(M));
auto *NewAccess = MSSAU->createMemoryAccessAfter(NewM, LastDef, LastDef);
MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/true);

// Remove the instruction we're replacing.
eraseInstruction(M);
++NumMemCpyInstr;
return true;
1096}

1098/// We've found that the (upward scanning) memory dependence of \p MemCpy is
1099/// \p MemSet.  Try to simplify \p MemSet to only set the trailing bytes that
1100/// weren't copied over by \p MemCpy.
1101///
1102/// In other words, transform:
1103/// \code
1104///   memset(dst, c, dst_size);
1105///   memcpy(dst, src, src_size);
1106/// \endcode
1107/// into:
1108/// \code
1109///   memcpy(dst, src, src_size);
1110///   memset(dst + src_size, c, dst_size <= src_size ? 0 : dst_size - src_size);
1111/// \endcode
1112bool MemCpyOptPass::processMemSetMemCpyDependence(MemCpyInst *MemCpy,
                                                MemSetInst *MemSet) {
// We can only transform memset/memcpy with the same destination.
if (!AA->isMustAlias(MemSet->getDest(), MemCpy->getDest()))
  return false;

// Check that src and dst of the memcpy aren't the same. While memcpy
// operands cannot partially overlap, exact equality is allowed.
if (isModSet(AA->getModRefInfo(MemCpy, MemoryLocation::getForSource(MemCpy))))
  return false;

// We know that dst up to src_size is not written. We now need to make sure
// that dst up to dst_size is not accessed. (If we did not move the memset,
// checking for reads would be sufficient.)
if (accessedBetween(*AA, MemoryLocation::getForDest(MemSet),
                    MSSA->getMemoryAccess(MemSet),
                    MSSA->getMemoryAccess(MemCpy)))
  return false;

// Use the same i8* dest as the memcpy, killing the memset dest if different.
Value *Dest = MemCpy->getRawDest();
Value *DestSize = MemSet->getLength();
Value *SrcSize = MemCpy->getLength();

if (mayBeVisibleThroughUnwinding(Dest, MemSet, MemCpy))
  return false;

// If the sizes are the same, simply drop the memset instead of generating
// a replacement with zero size.
if (DestSize == SrcSize) {
  eraseInstruction(MemSet);
  return true;
}

// By default, create an unaligned memset.
unsigned Align = 1;
// If Dest is aligned, and SrcSize is constant, use the minimum alignment
// of the sum.
const unsigned DestAlign =
    std::max(MemSet->getDestAlignment(), MemCpy->getDestAlignment());
if (DestAlign > 1)
  if (ConstantInt *SrcSizeC = dyn_cast<ConstantInt>(SrcSize))
    Align = MinAlign(SrcSizeC->getZExtValue(), DestAlign);

IRBuilder<> Builder(MemCpy);

// If the sizes have different types, zext the smaller one.
if (DestSize->getType() != SrcSize->getType()) {
  if (DestSize->getType()->getIntegerBitWidth() >
      SrcSize->getType()->getIntegerBitWidth())
    SrcSize = Builder.CreateZExt(SrcSize, DestSize->getType());
  else
    DestSize = Builder.CreateZExt(DestSize, SrcSize->getType());
}

Value *Ule = Builder.CreateICmpULE(DestSize, SrcSize);
Value *SizeDiff = Builder.CreateSub(DestSize, SrcSize);
Value *MemsetLen = Builder.CreateSelect(
    Ule, ConstantInt::getNullValue(DestSize->getType()), SizeDiff);
unsigned DestAS = Dest->getType()->getPointerAddressSpace();
Instruction *NewMemSet = Builder.CreateMemSet(
    Builder.CreateGEP(Builder.getInt8Ty(),
                      Builder.CreatePointerCast(Dest,
                                                Builder.getInt8PtrTy(DestAS)),
                      SrcSize),
    MemSet->getOperand(1), MemsetLen, MaybeAlign(Align));

assert(isa<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(MemCpy)) &&(static_cast<void> (0))
       "MemCpy must be a MemoryDef")(static_cast<void> (0));
// The new memset is inserted after the memcpy, but it is known that its
// defining access is the memset about to be removed which immediately
// precedes the memcpy.
auto *LastDef =
    cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(MemCpy));
auto *NewAccess = MSSAU->createMemoryAccessBefore(
    NewMemSet, LastDef->getDefiningAccess(), LastDef);
MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/true);

eraseInstruction(MemSet);
return true;
1192}

1194/// Determine whether the instruction has undefined content for the given Size,
1195/// either because it was freshly alloca'd or started its lifetime.
1196static bool hasUndefContents(MemorySSA *MSSA, AliasAnalysis *AA, Value *V,
                           MemoryDef *Def, Value *Size) {
if (MSSA->isLiveOnEntryDef(Def))
  return isa<AllocaInst>(getUnderlyingObject(V));

if (IntrinsicInst *II =
        dyn_cast_or_null<IntrinsicInst>(Def->getMemoryInst())) {
  if (II->getIntrinsicID() == Intrinsic::lifetime_start) {
    ConstantInt *LTSize = cast<ConstantInt>(II->getArgOperand(0));

    if (ConstantInt *CSize = dyn_cast<ConstantInt>(Size)) {
      if (AA->isMustAlias(V, II->getArgOperand(1)) &&
          LTSize->getZExtValue() >= CSize->getZExtValue())
        return true;
    }

    // If the lifetime.start covers a whole alloca (as it almost always
    // does) and we're querying a pointer based on that alloca, then we know
    // the memory is definitely undef, regardless of how exactly we alias.
    // The size also doesn't matter, as an out-of-bounds access would be UB.
    AllocaInst *Alloca = dyn_cast<AllocaInst>(getUnderlyingObject(V));
    if (getUnderlyingObject(II->getArgOperand(1)) == Alloca) {
      const DataLayout &DL = Alloca->getModule()->getDataLayout();
      if (Optional<TypeSize> AllocaSize = Alloca->getAllocationSizeInBits(DL))
        if (*AllocaSize == LTSize->getValue() * 8)
          return true;
    }
  }
}

return false;
1227}

1229/// Transform memcpy to memset when its source was just memset.
1230/// In other words, turn:
1231/// \code
1232///   memset(dst1, c, dst1_size);
1233///   memcpy(dst2, dst1, dst2_size);
1234/// \endcode
1235/// into:
1236/// \code
1237///   memset(dst1, c, dst1_size);
1238///   memset(dst2, c, dst2_size);
1239/// \endcode
1240/// When dst2_size <= dst1_size.
1241bool MemCpyOptPass::performMemCpyToMemSetOptzn(MemCpyInst *MemCpy,
                                             MemSetInst *MemSet) {
// Make sure that memcpy(..., memset(...), ...), that is we are memsetting and
// memcpying from the same address. Otherwise it is hard to reason about.
if (!AA->isMustAlias(MemSet->getRawDest(), MemCpy->getRawSource()))
  return false;

Value *MemSetSize = MemSet->getLength();
Value *CopySize = MemCpy->getLength();

if (MemSetSize != CopySize) {
  // Make sure the memcpy doesn't read any more than what the memset wrote.
  // Don't worry about sizes larger than i64.

  // A known memset size is required.
  ConstantInt *CMemSetSize = dyn_cast<ConstantInt>(MemSetSize);
  if (!CMemSetSize)
    return false;

  // A known memcpy size is also required.
  ConstantInt *CCopySize = dyn_cast<ConstantInt>(CopySize);
  if (!CCopySize)
    return false;
  if (CCopySize->getZExtValue() > CMemSetSize->getZExtValue()) {
    // If the memcpy is larger than the memset, but the memory was undef prior
    // to the memset, we can just ignore the tail. Technically we're only
    // interested in the bytes from MemSetSize..CopySize here, but as we can't
    // easily represent this location, we use the full 0..CopySize range.
    MemoryLocation MemCpyLoc = MemoryLocation::getForSource(MemCpy);
    bool CanReduceSize = false;
    MemoryUseOrDef *MemSetAccess = MSSA->getMemoryAccess(MemSet);
    MemoryAccess *Clobber = MSSA->getWalker()->getClobberingMemoryAccess(
        MemSetAccess->getDefiningAccess(), MemCpyLoc);
    if (auto *MD = dyn_cast<MemoryDef>(Clobber))
      if (hasUndefContents(MSSA, AA, MemCpy->getSource(), MD, CopySize))
        CanReduceSize = true;

    if (!CanReduceSize)
      return false;
    CopySize = MemSetSize;
  }
}

IRBuilder<> Builder(MemCpy);
Instruction *NewM =
    Builder.CreateMemSet(MemCpy->getRawDest(), MemSet->getOperand(1),
                         CopySize, MaybeAlign(MemCpy->getDestAlignment()));
auto *LastDef =
    cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(MemCpy));
auto *NewAccess = MSSAU->createMemoryAccessAfter(NewM, LastDef, LastDef);
MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/true);

return true;
1294}

1296/// Perform simplification of memcpy's.  If we have memcpy A
1297/// which copies X to Y, and memcpy B which copies Y to Z, then we can rewrite
1298/// B to be a memcpy from X to Z (or potentially a memmove, depending on
1299/// circumstances). This allows later passes to remove the first memcpy
1300/// altogether.
1301bool MemCpyOptPass::processMemCpy(MemCpyInst *M, BasicBlock::iterator &BBI) {
// We can only optimize non-volatile memcpy's.
if (M->isVolatile()) return false;

// If the source and destination of the memcpy are the same, then zap it.
if (M->getSource() == M->getDest()) {
  ++BBI;
  eraseInstruction(M);
  return true;
}

// If copying from a constant, try to turn the memcpy into a memset.
if (GlobalVariable *GV = dyn_cast<GlobalVariable>(M->getSource()))
  if (GV->isConstant() && GV->hasDefinitiveInitializer())
    if (Value *ByteVal = isBytewiseValue(GV->getInitializer(),
                                         M->getModule()->getDataLayout())) {
      IRBuilder<> Builder(M);
      Instruction *NewM =
          Builder.CreateMemSet(M->getRawDest(), ByteVal, M->getLength(),
                               MaybeAlign(M->getDestAlignment()), false);
      auto *LastDef =
          cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(M));
      auto *NewAccess =
          MSSAU->createMemoryAccessAfter(NewM, LastDef, LastDef);
      MSSAU->insertDef(cast<MemoryDef>(NewAccess), /*RenameUses=*/true);

      eraseInstruction(M);
      ++NumCpyToSet;
      return true;
    }

MemoryUseOrDef *MA = MSSA->getMemoryAccess(M);
MemoryAccess *AnyClobber = MSSA->getWalker()->getClobberingMemoryAccess(MA);
MemoryLocation DestLoc = MemoryLocation::getForDest(M);
const MemoryAccess *DestClobber =
    MSSA->getWalker()->getClobberingMemoryAccess(AnyClobber, DestLoc);

// Try to turn a partially redundant memset + memcpy into
// memcpy + smaller memset.  We don't need the memcpy size for this.
// The memcpy most post-dom the memset, so limit this to the same basic
// block. A non-local generalization is likely not worthwhile.
if (auto *MD = dyn_cast<MemoryDef>(DestClobber))
  if (auto *MDep = dyn_cast_or_null<MemSetInst>(MD->getMemoryInst()))
    if (DestClobber->getBlock() == M->getParent())
      if (processMemSetMemCpyDependence(M, MDep))
        return true;

MemoryAccess *SrcClobber = MSSA->getWalker()->getClobberingMemoryAccess(
    AnyClobber, MemoryLocation::getForSource(M));

// There are four possible optimizations we can do for memcpy:
//   a) memcpy-memcpy xform which exposes redundance for DSE.
//   b) call-memcpy xform for return slot optimization.
//   c) memcpy from freshly alloca'd space or space that has just started
//      its lifetime copies undefined data, and we can therefore eliminate
//      the memcpy in favor of the data that was already at the destination.
//   d) memcpy from a just-memset'd source can be turned into memset.
if (auto *MD = dyn_cast<MemoryDef>(SrcClobber)) {
  if (Instruction *MI = MD->getMemoryInst()) {
    if (ConstantInt *CopySize = dyn_cast<ConstantInt>(M->getLength())) {
      if (auto *C = dyn_cast<CallInst>(MI)) {
        // The memcpy must post-dom the call. Limit to the same block for
        // now. Additionally, we need to ensure that there are no accesses
        // to dest between the call and the memcpy. Accesses to src will be
        // checked by performCallSlotOptzn().
        // TODO: Support non-local call-slot optimization?
        if (C->getParent() == M->getParent() &&
            !accessedBetween(*AA, DestLoc, MD, MA)) {
          // FIXME: Can we pass in either of dest/src alignment here instead
          // of conservatively taking the minimum?
          Align Alignment = std::min(M->getDestAlign().valueOrOne(),
                                     M->getSourceAlign().valueOrOne());
          if (performCallSlotOptzn(M, M, M->getDest(), M->getSource(),
                                   CopySize->getZExtValue(), Alignment, C)) {
            LLVM_DEBUG(dbgs() << "Performed call slot optimization:\n"do { } while (false)
                              << "    call: " << *C << "\n"do { } while (false)
                              << "    memcpy: " << *M << "\n")do { } while (false);
            eraseInstruction(M);
            ++NumMemCpyInstr;
            return true;
          }
        }
      }
    }
    if (auto *MDep = dyn_cast<MemCpyInst>(MI))
      return processMemCpyMemCpyDependence(M, MDep);
    if (auto *MDep = dyn_cast<MemSetInst>(MI)) {
      if (performMemCpyToMemSetOptzn(M, MDep)) {
        LLVM_DEBUG(dbgs() << "Converted memcpy to memset\n")do { } while (false);
        eraseInstruction(M);
        ++NumCpyToSet;
        return true;
      }
    }
  }

  if (hasUndefContents(MSSA, AA, M->getSource(), MD, M->getLength())) {
    LLVM_DEBUG(dbgs() << "Removed memcpy from undef\n")do { } while (false);
    eraseInstruction(M);
    ++NumMemCpyInstr;
    return true;
  }
}

return false;
1406}

1408/// Transforms memmove calls to memcpy calls when the src/dst are guaranteed
1409/// not to alias.
1410bool MemCpyOptPass::processMemMove(MemMoveInst *M) {
// See if the source could be modified by this memmove potentially.
if (isModSet(AA->getModRefInfo(M, MemoryLocation::getForSource(M))))
  return false;

LLVM_DEBUG(dbgs() << "MemCpyOptPass: Optimizing memmove -> memcpy: " << *Mdo { } while (false)
                  << "\n")do { } while (false);

// If not, then we know we can transform this.
Type *ArgTys[3] = { M->getRawDest()->getType(),
                    M->getRawSource()->getType(),
                    M->getLength()->getType() };
M->setCalledFunction(Intrinsic::getDeclaration(M->getModule(),
                                               Intrinsic::memcpy, ArgTys));

// For MemorySSA nothing really changes (except that memcpy may imply stricter
// aliasing guarantees).

++NumMoveToCpy;
return true;
1430}

1432/// This is called on every byval argument in call sites.
1433bool MemCpyOptPass::processByValArgument(CallBase &CB, unsigned ArgNo) {
const DataLayout &DL = CB.getCaller()->getParent()->getDataLayout();
// Find out what feeds this byval argument.
Value *ByValArg = CB.getArgOperand(ArgNo);
Type *ByValTy = CB.getParamByValType(ArgNo);
uint64_t ByValSize = DL.getTypeAllocSize(ByValTy);
MemoryLocation Loc(ByValArg, LocationSize::precise(ByValSize));
MemoryUseOrDef *CallAccess = MSSA->getMemoryAccess(&CB);
if (!CallAccess)
  return false;
MemCpyInst *MDep = nullptr;
MemoryAccess *Clobber = MSSA->getWalker()->getClobberingMemoryAccess(
    CallAccess->getDefiningAccess(), Loc);
if (auto *MD = dyn_cast<MemoryDef>(Clobber))
  MDep = dyn_cast_or_null<MemCpyInst>(MD->getMemoryInst());

// If the byval argument isn't fed by a memcpy, ignore it.  If it is fed by
// a memcpy, see if we can byval from the source of the memcpy instead of the
// result.
if (!MDep || MDep->isVolatile() ||
    ByValArg->stripPointerCasts() != MDep->getDest())
  return false;

// The length of the memcpy must be larger or equal to the size of the byval.
ConstantInt *C1 = dyn_cast<ConstantInt>(MDep->getLength());
if (!C1 || C1->getValue().getZExtValue() < ByValSize)
  return false;

// Get the alignment of the byval.  If the call doesn't specify the alignment,
// then it is some target specific value that we can't know.
MaybeAlign ByValAlign = CB.getParamAlign(ArgNo);
if (!ByValAlign) return false;

// If it is greater than the memcpy, then we check to see if we can force the
// source of the memcpy to the alignment we need.  If we fail, we bail out.
MaybeAlign MemDepAlign = MDep->getSourceAlign();
if ((!MemDepAlign || *MemDepAlign < *ByValAlign) &&
    getOrEnforceKnownAlignment(MDep->getSource(), ByValAlign, DL, &CB, AC,
                               DT) < *ByValAlign)
  return false;

// The address space of the memcpy source must match the byval argument
if (MDep->getSource()->getType()->getPointerAddressSpace() !=
    ByValArg->getType()->getPointerAddressSpace())
  return false;

// Verify that the copied-from memory doesn't change in between the memcpy and
// the byval call.
//    memcpy(a <- b)
//    *b = 42;
//    foo(*a)
// It would be invalid to transform the second memcpy into foo(*b).
if (writtenBetween(MSSA, MemoryLocation::getForSource(MDep),
                   MSSA->getMemoryAccess(MDep), MSSA->getMemoryAccess(&CB)))
  return false;

Value *TmpCast = MDep->getSource();
if (MDep->getSource()->getType() != ByValArg->getType()) {
  BitCastInst *TmpBitCast = new BitCastInst(MDep->getSource(), ByValArg->getType(),
                                            "tmpcast", &CB);
  // Set the tmpcast's DebugLoc to MDep's
  TmpBitCast->setDebugLoc(MDep->getDebugLoc());
  TmpCast = TmpBitCast;
}

LLVM_DEBUG(dbgs() << "MemCpyOptPass: Forwarding memcpy to byval:\n"do { } while (false)
                  << "  " << *MDep << "\n"do { } while (false)
                  << "  " << CB << "\n")do { } while (false);

// Otherwise we're good!  Update the byval argument.
CB.setArgOperand(ArgNo, TmpCast);
++NumMemCpyInstr;
return true;
1506}

1508/// Executes one iteration of MemCpyOptPass.
1509bool MemCpyOptPass::iterateOnFunction(Function &F) {
bool MadeChange = false;

// Walk all instruction in the function.
for (BasicBlock &BB : F) {
  // Skip unreachable blocks. For example processStore assumes that an
  // instruction in a BB can't be dominated by a later instruction in the
  // same BB (which is a scenario that can happen for an unreachable BB that
  // has itself as a predecessor).
  if (!DT->isReachableFromEntry(&BB))
4
←
Assuming the condition is false→
5
←
Taking false branch→
    continue;

  for (BasicBlock::iterator BI = BB.begin(), BE = BB.end(); BI != BE;) {
6
←
Loop condition is true.  Entering loop body→
      // Avoid invalidating the iterator.
    Instruction *I = &*BI++;

    bool RepeatInstruction = false;

    if (StoreInst *SI7.1
'SI' is null
1
'SI' is null
 = dyn_cast<StoreInst>(I))
7
←
Assuming 'I' is not a 'StoreInst'→
8
←
Taking false branch→
      MadeChange |= processStore(SI, BI);
    else if (MemSetInst *M9.1
'M' is non-null
1
'M' is non-null
 = dyn_cast<MemSetInst>(I))
9
←
Assuming 'I' is a 'MemSetInst'→
10
←
Taking true branch→
      RepeatInstruction = processMemSet(M, BI);
11
←
Calling 'MemCpyOptPass::processMemSet'→
    else if (MemCpyInst *M = dyn_cast<MemCpyInst>(I))
      RepeatInstruction = processMemCpy(M, BI);
    else if (MemMoveInst *M = dyn_cast<MemMoveInst>(I))
      RepeatInstruction = processMemMove(M);
    else if (auto *CB = dyn_cast<CallBase>(I)) {
      for (unsigned i = 0, e = CB->arg_size(); i != e; ++i)
        if (CB->isByValArgument(i))
          MadeChange |= processByValArgument(*CB, i);
    }

    // Reprocess the instruction if desired.
    if (RepeatInstruction) {
      if (BI != BB.begin())
        --BI;
      MadeChange = true;
    }
  }
}

return MadeChange;
1551}

1553PreservedAnalyses MemCpyOptPass::run(Function &F, FunctionAnalysisManager &AM) {
auto &TLI = AM.getResult<TargetLibraryAnalysis>(F);
auto *AA = &AM.getResult<AAManager>(F);
auto *AC = &AM.getResult<AssumptionAnalysis>(F);
auto *DT = &AM.getResult<DominatorTreeAnalysis>(F);
auto *MSSA = &AM.getResult<MemorySSAAnalysis>(F);

bool MadeChange = runImpl(F, &TLI, AA, AC, DT, &MSSA->getMSSA());
1
Calling 'MemCpyOptPass::runImpl'→
if (!MadeChange)
  return PreservedAnalyses::all();

PreservedAnalyses PA;
PA.preserveSet<CFGAnalyses>();
PA.preserve<MemorySSAAnalysis>();
return PA;
1568}

1570bool MemCpyOptPass::runImpl(Function &F, TargetLibraryInfo *TLI_,
                          AliasAnalysis *AA_, AssumptionCache *AC_,
                          DominatorTree *DT_, MemorySSA *MSSA_) {
bool MadeChange = false;
TLI = TLI_;
AA = AA_;
AC = AC_;
DT = DT_;
MSSA = MSSA_;
MemorySSAUpdater MSSAU_(MSSA_);
MSSAU = &MSSAU_;

while (true) {
2
←
Loop condition is true.  Entering loop body→
  if (!iterateOnFunction(F))
3
←
Calling 'MemCpyOptPass::iterateOnFunction'→
    break;
  MadeChange = true;
}

if (VerifyMemorySSA)
  MSSA_->verifyMemorySSA();

return MadeChange;
1592}

1594/// This is the main transformation entry point for a function.
1595bool MemCpyOptLegacyPass::runOnFunction(Function &F) {
if (skipFunction(F))
  return false;

auto *TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI(F);
auto *AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
auto *AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
auto *DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
auto *MSSA = &getAnalysis<MemorySSAWrapperPass>().getMSSA();

return Impl.runImpl(F, TLI, AA, AC, DT, MSSA);
1606}

←

/build/llvm-toolchain-snapshot-14~++20210903100615+fd66b44ec19e/llvm/include/llvm/ADT/SmallVector.h

1//===- llvm/ADT/SmallVector.h - 'Normally small' vectors --------*- C++ -*-===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This file defines the SmallVector class.
10//
11//===----------------------------------------------------------------------===//

13#ifndef LLVM_ADT_SMALLVECTOR_H
14#define LLVM_ADT_SMALLVECTOR_H

16#include "llvm/ADT/iterator_range.h"
17#include "llvm/Support/Compiler.h"
18#include "llvm/Support/ErrorHandling.h"
19#include "llvm/Support/MemAlloc.h"
20#include "llvm/Support/type_traits.h"
21#include <algorithm>
22#include <cassert>
23#include <cstddef>
24#include <cstdlib>
25#include <cstring>
26#include <functional>
27#include <initializer_list>
28#include <iterator>
29#include <limits>
30#include <memory>
31#include <new>
32#include <type_traits>
33#include <utility>

35namespace llvm {

37/// This is all the stuff common to all SmallVectors.
38///
39/// The template parameter specifies the type which should be used to hold the
40/// Size and Capacity of the SmallVector, so it can be adjusted.
41/// Using 32 bit size is desirable to shrink the size of the SmallVector.
42/// Using 64 bit size is desirable for cases like SmallVector<char>, where a
43/// 32 bit size would limit the vector to ~4GB. SmallVectors are used for
44/// buffering bitcode output - which can exceed 4GB.
45template <class Size_T> class SmallVectorBase {
46protected:
void *BeginX;
Size_T Size = 0, Capacity;

/// The maximum value of the Size_T used.
static constexpr size_t SizeTypeMax() {
  return std::numeric_limits<Size_T>::max();
}

SmallVectorBase() = delete;
SmallVectorBase(void *FirstEl, size_t TotalCapacity)
    : BeginX(FirstEl), Capacity(TotalCapacity) {}

/// This is a helper for \a grow() that's out of line to reduce code
/// duplication.  This function will report a fatal error if it can't grow at
/// least to \p MinSize.
void *mallocForGrow(size_t MinSize, size_t TSize, size_t &NewCapacity);

/// This is an implementation of the grow() method which only works
/// on POD-like data types and is out of line to reduce code duplication.
/// This function will report a fatal error if it cannot increase capacity.
void grow_pod(void *FirstEl, size_t MinSize, size_t TSize);

69public:
size_t size() const { return Size; }
size_t capacity() const { return Capacity; }

LLVM_NODISCARD[[clang::warn_unused_result]] bool empty() const { return !Size; }
23
←
Assuming field 'Size' is not equal to 0, which participates in a condition later→
24
←
Returning zero, which participates in a condition later→

/// Set the array size to \p N, which the current array must have enough
/// capacity for.
///
/// This does not construct or destroy any elements in the vector.
///
/// Clients can use this in conjunction with capacity() to write past the end
/// of the buffer when they know that more elements are available, and only
/// update the size later. This avoids the cost of value initializing elements
/// which will only be overwritten.
void set_size(size_t N) {
  assert(N <= capacity())(static_cast<void> (0));
  Size = N;
}
88};

90template <class T>
91using SmallVectorSizeType =
  typename std::conditional<sizeof(T) < 4 && sizeof(void *) >= 8, uint64_t,
                            uint32_t>::type;

95/// Figure out the offset of the first element.
96template <class T, typename = void> struct SmallVectorAlignmentAndSize {
alignas(SmallVectorBase<SmallVectorSizeType<T>>) char Base[sizeof(
    SmallVectorBase<SmallVectorSizeType<T>>)];
alignas(T) char FirstEl[sizeof(T)];
100};

102/// This is the part of SmallVectorTemplateBase which does not depend on whether
103/// the type T is a POD. The extra dummy template argument is used by ArrayRef
104/// to avoid unnecessarily requiring T to be complete.
105template <typename T, typename = void>
106class SmallVectorTemplateCommon
  : public SmallVectorBase<SmallVectorSizeType<T>> {
using Base = SmallVectorBase<SmallVectorSizeType<T>>;

/// Find the address of the first element.  For this pointer math to be valid
/// with small-size of 0 for T with lots of alignment, it's important that
/// SmallVectorStorage is properly-aligned even for small-size of 0.
void *getFirstEl() const {
  return const_cast<void *>(reinterpret_cast<const void *>(
      reinterpret_cast<const char *>(this) +
      offsetof(SmallVectorAlignmentAndSize<T>, FirstEl)__builtin_offsetof(SmallVectorAlignmentAndSize<T>, FirstEl
)));
}
// Space after 'FirstEl' is clobbered, do not add any instance vars after it.

120protected:
SmallVectorTemplateCommon(size_t Size) : Base(getFirstEl(), Size) {}

void grow_pod(size_t MinSize, size_t TSize) {
  Base::grow_pod(getFirstEl(), MinSize, TSize);
}

/// Return true if this is a smallvector which has not had dynamic
/// memory allocated for it.
bool isSmall() const { return this->BeginX == getFirstEl(); }

/// Put this vector in a state of being small.
void resetToSmall() {
  this->BeginX = getFirstEl();
  this->Size = this->Capacity = 0; // FIXME: Setting Capacity to 0 is suspect.
}

/// Return true if V is an internal reference to the given range.
bool isReferenceToRange(const void *V, const void *First, const void *Last) const {
  // Use std::less to avoid UB.
  std::less<> LessThan;
  return !LessThan(V, First) && LessThan(V, Last);
}

/// Return true if V is an internal reference to this vector.
bool isReferenceToStorage(const void *V) const {
  return isReferenceToRange(V, this->begin(), this->end());
}

/// Return true if First and Last form a valid (possibly empty) range in this
/// vector's storage.
bool isRangeInStorage(const void *First, const void *Last) const {
  // Use std::less to avoid UB.
  std::less<> LessThan;
  return !LessThan(First, this->begin()) && !LessThan(Last, First) &&
         !LessThan(this->end(), Last);
}

/// Return true unless Elt will be invalidated by resizing the vector to
/// NewSize.
bool isSafeToReferenceAfterResize(const void *Elt, size_t NewSize) {
  // Past the end.
  if (LLVM_LIKELY(!isReferenceToStorage(Elt))__builtin_expect((bool)(!isReferenceToStorage(Elt)), true))
    return true;

  // Return false if Elt will be destroyed by shrinking.
  if (NewSize <= this->size())
    return Elt < this->begin() + NewSize;

  // Return false if we need to grow.
  return NewSize <= this->capacity();
}

/// Check whether Elt will be invalidated by resizing the vector to NewSize.
void assertSafeToReferenceAfterResize(const void *Elt, size_t NewSize) {
  assert(isSafeToReferenceAfterResize(Elt, NewSize) &&(static_cast<void> (0))
         "Attempting to reference an element of the vector in an operation "(static_cast<void> (0))
         "that invalidates it")(static_cast<void> (0));
}

/// Check whether Elt will be invalidated by increasing the size of the
/// vector by N.
void assertSafeToAdd(const void *Elt, size_t N = 1) {
  this->assertSafeToReferenceAfterResize(Elt, this->size() + N);
}

/// Check whether any part of the range will be invalidated by clearing.
void assertSafeToReferenceAfterClear(const T *From, const T *To) {
  if (From == To)
    return;
  this->assertSafeToReferenceAfterResize(From, 0);
  this->assertSafeToReferenceAfterResize(To - 1, 0);
}
template <
    class ItTy,
    std::enable_if_t<!std::is_same<std::remove_const_t<ItTy>, T *>::value,
                     bool> = false>
void assertSafeToReferenceAfterClear(ItTy, ItTy) {}

/// Check whether any part of the range will be invalidated by growing.
void assertSafeToAddRange(const T *From, const T *To) {
  if (From == To)
    return;
  this->assertSafeToAdd(From, To - From);
  this->assertSafeToAdd(To - 1, To - From);
}
template <
    class ItTy,
    std::enable_if_t<!std::is_same<std::remove_const_t<ItTy>, T *>::value,
                     bool> = false>
void assertSafeToAddRange(ItTy, ItTy) {}

/// Reserve enough space to add one element, and return the updated element
/// pointer in case it was a reference to the storage.
template <class U>
static const T *reserveForParamAndGetAddressImpl(U *This, const T &Elt,
                                                 size_t N) {
  size_t NewSize = This->size() + N;
  if (LLVM_LIKELY(NewSize <= This->capacity())__builtin_expect((bool)(NewSize <= This->capacity()), true
))
    return &Elt;

  bool ReferencesStorage = false;
  int64_t Index = -1;
  if (!U::TakesParamByValue) {
    if (LLVM_UNLIKELY(This->isReferenceToStorage(&Elt))__builtin_expect((bool)(This->isReferenceToStorage(&Elt
)), false)) {
      ReferencesStorage = true;
      Index = &Elt - This->begin();
    }
  }
  This->grow(NewSize);
  return ReferencesStorage ? This->begin() + Index : &Elt;
}

233public:
using size_type = size_t;
using difference_type = ptrdiff_t;
using value_type = T;
using iterator = T *;
using const_iterator = const T *;

using const_reverse_iterator = std::reverse_iterator<const_iterator>;
using reverse_iterator = std::reverse_iterator<iterator>;

using reference = T &;
using const_reference = const T &;
using pointer = T *;
using const_pointer = const T *;

using Base::capacity;
using Base::empty;
using Base::size;

// forward iterator creation methods.
iterator begin() { return (iterator)this->BeginX; }
const_iterator begin() const { return (const_iterator)this->BeginX; }
iterator end() { return begin() + size(); }
const_iterator end() const { return begin() + size(); }

// reverse iterator creation methods.
reverse_iterator rbegin()            { return reverse_iterator(end()); }
const_reverse_iterator rbegin() const{ return const_reverse_iterator(end()); }
reverse_iterator rend()              { return reverse_iterator(begin()); }
const_reverse_iterator rend() const { return const_reverse_iterator(begin());}

size_type size_in_bytes() const { return size() * sizeof(T); }
size_type max_size() const {
  return std::min(this->SizeTypeMax(), size_type(-1) / sizeof(T));
}

size_t capacity_in_bytes() const { return capacity() * sizeof(T); }

/// Return a pointer to the vector's buffer, even if empty().
pointer data() { return pointer(begin()); }
/// Return a pointer to the vector's buffer, even if empty().
const_pointer data() const { return const_pointer(begin()); }

reference operator[](size_type idx) {
  assert(idx < size())(static_cast<void> (0));
  return begin()[idx];
}
const_reference operator[](size_type idx) const {
  assert(idx < size())(static_cast<void> (0));
  return begin()[idx];
}

reference front() {
  assert(!empty())(static_cast<void> (0));
  return begin()[0];
}
const_reference front() const {
  assert(!empty())(static_cast<void> (0));
  return begin()[0];
}

reference back() {
  assert(!empty())(static_cast<void> (0));
  return end()[-1];
}
const_reference back() const {
  assert(!empty())(static_cast<void> (0));
  return end()[-1];
}
302};

304/// SmallVectorTemplateBase<TriviallyCopyable = false> - This is where we put
305/// method implementations that are designed to work with non-trivial T's.
306///
307/// We approximate is_trivially_copyable with trivial move/copy construction and
308/// trivial destruction. While the standard doesn't specify that you're allowed
309/// copy these types with memcpy, there is no way for the type to observe this.
310/// This catches the important case of std::pair<POD, POD>, which is not
311/// trivially assignable.
312template <typename T, bool = (is_trivially_copy_constructible<T>::value) &&
                           (is_trivially_move_constructible<T>::value) &&
                           std::is_trivially_destructible<T>::value>
315class SmallVectorTemplateBase : public SmallVectorTemplateCommon<T> {
friend class SmallVectorTemplateCommon<T>;

318protected:
static constexpr bool TakesParamByValue = false;
using ValueParamT = const T &;

SmallVectorTemplateBase(size_t Size) : SmallVectorTemplateCommon<T>(Size) {}

static void destroy_range(T *S, T *E) {
  while (S != E) {
    --E;
    E->~T();
  }
}

/// Move the range [I, E) into the uninitialized memory starting with "Dest",
/// constructing elements as needed.
template<typename It1, typename It2>
static void uninitialized_move(It1 I, It1 E, It2 Dest) {
  std::uninitialized_copy(std::make_move_iterator(I),
                          std::make_move_iterator(E), Dest);
}

/// Copy the range [I, E) onto the uninitialized memory starting with "Dest",
/// constructing elements as needed.
template<typename It1, typename It2>
static void uninitialized_copy(It1 I, It1 E, It2 Dest) {
  std::uninitialized_copy(I, E, Dest);
}

/// Grow the allocated memory (without initializing new elements), doubling
/// the size of the allocated memory. Guarantees space for at least one more
/// element, or MinSize more elements if specified.
void grow(size_t MinSize = 0);

/// Create a new allocation big enough for \p MinSize and pass back its size
/// in \p NewCapacity. This is the first section of \a grow().
T *mallocForGrow(size_t MinSize, size_t &NewCapacity) {
  return static_cast<T *>(
      SmallVectorBase<SmallVectorSizeType<T>>::mallocForGrow(
          MinSize, sizeof(T), NewCapacity));
}

/// Move existing elements over to the new allocation \p NewElts, the middle
/// section of \a grow().
void moveElementsForGrow(T *NewElts);

/// Transfer ownership of the allocation, finishing up \a grow().
void takeAllocationForGrow(T *NewElts, size_t NewCapacity);

/// Reserve enough space to add one element, and return the updated element
/// pointer in case it was a reference to the storage.
const T *reserveForParamAndGetAddress(const T &Elt, size_t N = 1) {
  return this->reserveForParamAndGetAddressImpl(this, Elt, N);
}

/// Reserve enough space to add one element, and return the updated element
/// pointer in case it was a reference to the storage.
T *reserveForParamAndGetAddress(T &Elt, size_t N = 1) {
  return const_cast<T *>(
      this->reserveForParamAndGetAddressImpl(this, Elt, N));
}

static T &&forward_value_param(T &&V) { return std::move(V); }
static const T &forward_value_param(const T &V) { return V; }

void growAndAssign(size_t NumElts, const T &Elt) {
  // Grow manually in case Elt is an internal reference.
  size_t NewCapacity;
  T *NewElts = mallocForGrow(NumElts, NewCapacity);
  std::uninitialized_fill_n(NewElts, NumElts, Elt);
  this->destroy_range(this->begin(), this->end());
  takeAllocationForGrow(NewElts, NewCapacity);
  this->set_size(NumElts);
}

template <typename... ArgTypes> T &growAndEmplaceBack(ArgTypes &&... Args) {
  // Grow manually in case one of Args is an internal reference.
  size_t NewCapacity;
  T *NewElts = mallocForGrow(0, NewCapacity);
  ::new ((void *)(NewElts + this->size())) T(std::forward<ArgTypes>(Args)...);
  moveElementsForGrow(NewElts);
  takeAllocationForGrow(NewElts, NewCapacity);
  this->set_size(this->size() + 1);
  return this->back();
}

403public:
void push_back(const T &Elt) {
  const T *EltPtr = reserveForParamAndGetAddress(Elt);
  ::new ((void *)this->end()) T(*EltPtr);
  this->set_size(this->size() + 1);
}

void push_back(T &&Elt) {
  T *EltPtr = reserveForParamAndGetAddress(Elt);
  ::new ((void *)this->end()) T(::std::move(*EltPtr));
  this->set_size(this->size() + 1);
}

void pop_back() {
  this->set_size(this->size() - 1);
  this->end()->~T();
}
420};

422// Define this out-of-line to dissuade the C++ compiler from inlining it.
423template <typename T, bool TriviallyCopyable>
424void SmallVectorTemplateBase<T, TriviallyCopyable>::grow(size_t MinSize) {
size_t NewCapacity;
T *NewElts = mallocForGrow(MinSize, NewCapacity);
moveElementsForGrow(NewElts);
takeAllocationForGrow(NewElts, NewCapacity);
429}

431// Define this out-of-line to dissuade the C++ compiler from inlining it.
432template <typename T, bool TriviallyCopyable>
433void SmallVectorTemplateBase<T, TriviallyCopyable>::moveElementsForGrow(
  T *NewElts) {
// Move the elements over.
this->uninitialized_move(this->begin(), this->end(), NewElts);

// Destroy the original elements.
destroy_range(this->begin(), this->end());
440}

442// Define this out-of-line to dissuade the C++ compiler from inlining it.
443template <typename T, bool TriviallyCopyable>
444void SmallVectorTemplateBase<T, TriviallyCopyable>::takeAllocationForGrow(
  T *NewElts, size_t NewCapacity) {
// If this wasn't grown from the inline copy, deallocate the old space.
if (!this->isSmall())
  free(this->begin());

this->BeginX = NewElts;
this->Capacity = NewCapacity;
452}

454/// SmallVectorTemplateBase<TriviallyCopyable = true> - This is where we put
455/// method implementations that are designed to work with trivially copyable
456/// T's. This allows using memcpy in place of copy/move construction and
457/// skipping destruction.
458template <typename T>
459class SmallVectorTemplateBase<T, true> : public SmallVectorTemplateCommon<T> {
friend class SmallVectorTemplateCommon<T>;

462protected:
/// True if it's cheap enough to take parameters by value. Doing so avoids
/// overhead related to mitigations for reference invalidation.
static constexpr bool TakesParamByValue = sizeof(T) <= 2 * sizeof(void *);

/// Either const T& or T, depending on whether it's cheap enough to take
/// parameters by value.
using ValueParamT =
    typename std::conditional<TakesParamByValue, T, const T &>::type;

SmallVectorTemplateBase(size_t Size) : SmallVectorTemplateCommon<T>(Size) {}

// No need to do a destroy loop for POD's.
static void destroy_range(T *, T *) {}

/// Move the range [I, E) onto the uninitialized memory
/// starting with "Dest", constructing elements into it as needed.
template<typename It1, typename It2>
static void uninitialized_move(It1 I, It1 E, It2 Dest) {
  // Just do a copy.
  uninitialized_copy(I, E, Dest);
}

/// Copy the range [I, E) onto the uninitialized memory
/// starting with "Dest", constructing elements into it as needed.
template<typename It1, typename It2>
static void uninitialized_copy(It1 I, It1 E, It2 Dest) {
  // Arbitrary iterator types; just use the basic implementation.
  std::uninitialized_copy(I, E, Dest);
}

/// Copy the range [I, E) onto the uninitialized memory
/// starting with "Dest", constructing elements into it as needed.
template <typename T1, typename T2>
static void uninitialized_copy(
    T1 *I, T1 *E, T2 *Dest,
    std::enable_if_t<std::is_same<typename std::remove_const<T1>::type,
                                  T2>::value> * = nullptr) {
  // Use memcpy for PODs iterated by pointers (which includes SmallVector
  // iterators): std::uninitialized_copy optimizes to memmove, but we can
  // use memcpy here. Note that I and E are iterators and thus might be
  // invalid for memcpy if they are equal.
  if (I != E)
    memcpy(reinterpret_cast<void *>(Dest), I, (E - I) * sizeof(T));
}

/// Double the size of the allocated memory, guaranteeing space for at
/// least one more element or MinSize if specified.
void grow(size_t MinSize = 0) { this->grow_pod(MinSize, sizeof(T)); }

/// Reserve enough space to add one element, and return the updated element
/// pointer in case it was a reference to the storage.
const T *reserveForParamAndGetAddress(const T &Elt, size_t N = 1) {
  return this->reserveForParamAndGetAddressImpl(this, Elt, N);
}

/// Reserve enough space to add one element, and return the updated element
/// pointer in case it was a reference to the storage.
T *reserveForParamAndGetAddress(T &Elt, size_t N = 1) {
  return const_cast<T *>(
      this->reserveForParamAndGetAddressImpl(this, Elt, N));
}

/// Copy \p V or return a reference, depending on \a ValueParamT.
static ValueParamT forward_value_param(ValueParamT V) { return V; }

void growAndAssign(size_t NumElts, T Elt) {
  // Elt has been copied in case it's an internal reference, side-stepping
  // reference invalidation problems without losing the realloc optimization.
  this->set_size(0);
  this->grow(NumElts);
  std::uninitialized_fill_n(this->begin(), NumElts, Elt);
  this->set_size(NumElts);
}

template <typename... ArgTypes> T &growAndEmplaceBack(ArgTypes &&... Args) {
  // Use push_back with a copy in case Args has an internal reference,
  // side-stepping reference invalidation problems without losing the realloc
  // optimization.
  push_back(T(std::forward<ArgTypes>(Args)...));
  return this->back();
}

545public:
void push_back(ValueParamT Elt) {
  const T *EltPtr = reserveForParamAndGetAddress(Elt);
  memcpy(reinterpret_cast<void *>(this->end()), EltPtr, sizeof(T));
  this->set_size(this->size() + 1);
}

void pop_back() { this->set_size(this->size() - 1); }
553};

555/// This class consists of common code factored out of the SmallVector class to
556/// reduce code duplication based on the SmallVector 'N' template parameter.
557template <typename T>
558class SmallVectorImpl : public SmallVectorTemplateBase<T> {
using SuperClass = SmallVectorTemplateBase<T>;

561public:
using iterator = typename SuperClass::iterator;
using const_iterator = typename SuperClass::const_iterator;
using reference = typename SuperClass::reference;
using size_type = typename SuperClass::size_type;

567protected:
using SmallVectorTemplateBase<T>::TakesParamByValue;
using ValueParamT = typename SuperClass::ValueParamT;

// Default ctor - Initialize to empty.
explicit SmallVectorImpl(unsigned N)
    : SmallVectorTemplateBase<T>(N) {}

575public:
SmallVectorImpl(const SmallVectorImpl &) = delete;

~SmallVectorImpl() {
  // Subclass has already destructed this vector's elements.
  // If this wasn't grown from the inline copy, deallocate the old space.
  if (!this->isSmall())
    free(this->begin());
}

void clear() {
  this->destroy_range(this->begin(), this->end());
  this->Size = 0;
}

590private:
template <bool ForOverwrite> void resizeImpl(size_type N) {
  if (N < this->size()) {
    this->pop_back_n(this->size() - N);
  } else if (N > this->size()) {
    this->reserve(N);
    for (auto I = this->end(), E = this->begin() + N; I != E; ++I)
      if (ForOverwrite)
        new (&*I) T;
      else
        new (&*I) T();
    this->set_size(N);
  }
}

605public:
void resize(size_type N) { resizeImpl<false>(N); }

/// Like resize, but \ref T is POD, the new values won't be initialized.
void resize_for_overwrite(size_type N) { resizeImpl<true>(N); }

void resize(size_type N, ValueParamT NV) {
  if (N == this->size())
    return;

  if (N < this->size()) {
    this->pop_back_n(this->size() - N);
    return;
  }

  // N > this->size(). Defer to append.
  this->append(N - this->size(), NV);
}

void reserve(size_type N) {
  if (this->capacity() < N)
    this->grow(N);
}

void pop_back_n(size_type NumItems) {
  assert(this->size() >= NumItems)(static_cast<void> (0));
  this->destroy_range(this->end() - NumItems, this->end());
  this->set_size(this->size() - NumItems);
}

LLVM_NODISCARD[[clang::warn_unused_result]] T pop_back_val() {
  T Result = ::std::move(this->back());
  this->pop_back();
  return Result;
}

void swap(SmallVectorImpl &RHS);

/// Add the specified range to the end of the SmallVector.
template <typename in_iter,
          typename = std::enable_if_t<std::is_convertible<
              typename std::iterator_traits<in_iter>::iterator_category,
              std::input_iterator_tag>::value>>
void append(in_iter in_start, in_iter in_end) {
  this->assertSafeToAddRange(in_start, in_end);
  size_type NumInputs = std::distance(in_start, in_end);
  this->reserve(this->size() + NumInputs);
  this->uninitialized_copy(in_start, in_end, this->end());
  this->set_size(this->size() + NumInputs);
}

/// Append \p NumInputs copies of \p Elt to the end.
void append(size_type NumInputs, ValueParamT Elt) {
  const T *EltPtr = this->reserveForParamAndGetAddress(Elt, NumInputs);
  std::uninitialized_fill_n(this->end(), NumInputs, *EltPtr);
  this->set_size(this->size() + NumInputs);
}

void append(std::initializer_list<T> IL) {
  append(IL.begin(), IL.end());
}

void append(const SmallVectorImpl &RHS) { append(RHS.begin(), RHS.end()); }

void assign(size_type NumElts, ValueParamT Elt) {
  // Note that Elt could be an internal reference.
  if (NumElts > this->capacity()) {
    this->growAndAssign(NumElts, Elt);
    return;
  }

  // Assign over existing elements.
  std::fill_n(this->begin(), std::min(NumElts, this->size()), Elt);
  if (NumElts > this->size())
    std::uninitialized_fill_n(this->end(), NumElts - this->size(), Elt);
  else if (NumElts < this->size())
    this->destroy_range(this->begin() + NumElts, this->end());
  this->set_size(NumElts);
}

// FIXME: Consider assigning over existing elements, rather than clearing &
// re-initializing them - for all assign(...) variants.

template <typename in_iter,
          typename = std::enable_if_t<std::is_convertible<
              typename std::iterator_traits<in_iter>::iterator_category,
              std::input_iterator_tag>::value>>
void assign(in_iter in_start, in_iter in_end) {
  this->assertSafeToReferenceAfterClear(in_start, in_end);
  clear();
  append(in_start, in_end);
}

void assign(std::initializer_list<T> IL) {
  clear();
  append(IL);
}

void assign(const SmallVectorImpl &RHS) { assign(RHS.begin(), RHS.end()); }

iterator erase(const_iterator CI) {
  // Just cast away constness because this is a non-const member function.
  iterator I = const_cast<iterator>(CI);

  assert(this->isReferenceToStorage(CI) && "Iterator to erase is out of bounds.")(static_cast<void> (0));

  iterator N = I;
  // Shift all elts down one.
  std::move(I+1, this->end(), I);
  // Drop the last elt.
  this->pop_back();
  return(N);
}

iterator erase(const_iterator CS, const_iterator CE) {
  // Just cast away constness because this is a non-const member function.
  iterator S = const_cast<iterator>(CS);
  iterator E = const_cast<iterator>(CE);

  assert(this->isRangeInStorage(S, E) && "Range to erase is out of bounds.")(static_cast<void> (0));

  iterator N = S;
  // Shift all elts down.
  iterator I = std::move(E, this->end(), S);
  // Drop the last elts.
  this->destroy_range(I, this->end());
  this->set_size(I - this->begin());
  return(N);
}

735private:
template <class ArgType> iterator insert_one_impl(iterator I, ArgType &&Elt) {
  // Callers ensure that ArgType is derived from T.
  static_assert(
      std::is_same<std::remove_const_t<std::remove_reference_t<ArgType>>,
                   T>::value,
      "ArgType must be derived from T!");

  if (I == this->end()) {  // Important special case for empty vector.
    this->push_back(::std::forward<ArgType>(Elt));
    return this->end()-1;
  }

  assert(this->isReferenceToStorage(I) && "Insertion iterator is out of bounds.")(static_cast<void> (0));

  // Grow if necessary.
  size_t Index = I - this->begin();
  std::remove_reference_t<ArgType> *EltPtr =
      this->reserveForParamAndGetAddress(Elt);
  I = this->begin() + Index;

  ::new ((void*) this->end()) T(::std::move(this->back()));
  // Push everything else over.
  std::move_backward(I, this->end()-1, this->end());
  this->set_size(this->size() + 1);

  // If we just moved the element we're inserting, be sure to update
  // the reference (never happens if TakesParamByValue).
  static_assert(!TakesParamByValue || std::is_same<ArgType, T>::value,
                "ArgType must be 'T' when taking by value!");
  if (!TakesParamByValue && this->isReferenceToRange(EltPtr, I, this->end()))
    ++EltPtr;

  *I = ::std::forward<ArgType>(*EltPtr);
  return I;
}

772public:
iterator insert(iterator I, T &&Elt) {
  return insert_one_impl(I, this->forward_value_param(std::move(Elt)));
}

iterator insert(iterator I, const T &Elt) {
  return insert_one_impl(I, this->forward_value_param(Elt));
}

iterator insert(iterator I, size_type NumToInsert, ValueParamT Elt) {
  // Convert iterator to elt# to avoid invalidating iterator when we reserve()
  size_t InsertElt = I - this->begin();

  if (I == this->end()) {  // Important special case for empty vector.
    append(NumToInsert, Elt);
    return this->begin()+InsertElt;
  }

  assert(this->isReferenceToStorage(I) && "Insertion iterator is out of bounds.")(static_cast<void> (0));

  // Ensure there is enough space, and get the (maybe updated) address of
  // Elt.
  const T *EltPtr = this->reserveForParamAndGetAddress(Elt, NumToInsert);

  // Uninvalidate the iterator.
  I = this->begin()+InsertElt;

  // If there are more elements between the insertion point and the end of the
  // range than there are being inserted, we can use a simple approach to
  // insertion.  Since we already reserved space, we know that this won't
  // reallocate the vector.
  if (size_t(this->end()-I) >= NumToInsert) {
    T *OldEnd = this->end();
    append(std::move_iterator<iterator>(this->end() - NumToInsert),
           std::move_iterator<iterator>(this->end()));

    // Copy the existing elements that get replaced.
    std::move_backward(I, OldEnd-NumToInsert, OldEnd);

    // If we just moved the element we're inserting, be sure to update
    // the reference (never happens if TakesParamByValue).
    if (!TakesParamByValue && I <= EltPtr && EltPtr < this->end())
      EltPtr += NumToInsert;

    std::fill_n(I, NumToInsert, *EltPtr);
    return I;
  }

  // Otherwise, we're inserting more elements than exist already, and we're
  // not inserting at the end.

  // Move over the elements that we're about to overwrite.
  T *OldEnd = this->end();
  this->set_size(this->size() + NumToInsert);
  size_t NumOverwritten = OldEnd-I;
  this->uninitialized_move(I, OldEnd, this->end()-NumOverwritten);

  // If we just moved the element we're inserting, be sure to update
  // the reference (never happens if TakesParamByValue).
  if (!TakesParamByValue && I <= EltPtr && EltPtr < this->end())
    EltPtr += NumToInsert;

  // Replace the overwritten part.
  std::fill_n(I, NumOverwritten, *EltPtr);

  // Insert the non-overwritten middle part.
  std::uninitialized_fill_n(OldEnd, NumToInsert - NumOverwritten, *EltPtr);
  return I;
}

template <typename ItTy,
          typename = std::enable_if_t<std::is_convertible<
              typename std::iterator_traits<ItTy>::iterator_category,
              std::input_iterator_tag>::value>>
iterator insert(iterator I, ItTy From, ItTy To) {
  // Convert iterator to elt# to avoid invalidating iterator when we reserve()
  size_t InsertElt = I - this->begin();

  if (I == this->end()) {  // Important special case for empty vector.
    append(From, To);
    return this->begin()+InsertElt;
  }

  assert(this->isReferenceToStorage(I) && "Insertion iterator is out of bounds.")(static_cast<void> (0));

  // Check that the reserve that follows doesn't invalidate the iterators.
  this->assertSafeToAddRange(From, To);

  size_t NumToInsert = std::distance(From, To);

  // Ensure there is enough space.
  reserve(this->size() + NumToInsert);

  // Uninvalidate the iterator.
  I = this->begin()+InsertElt;

  // If there are more elements between the insertion point and the end of the
  // range than there are being inserted, we can use a simple approach to
  // insertion.  Since we already reserved space, we know that this won't
  // reallocate the vector.
  if (size_t(this->end()-I) >= NumToInsert) {
    T *OldEnd = this->end();
    append(std::move_iterator<iterator>(this->end() - NumToInsert),
           std::move_iterator<iterator>(this->end()));

    // Copy the existing elements that get replaced.
    std::move_backward(I, OldEnd-NumToInsert, OldEnd);

    std::copy(From, To, I);
    return I;
  }

  // Otherwise, we're inserting more elements than exist already, and we're
  // not inserting at the end.

  // Move over the elements that we're about to overwrite.
  T *OldEnd = this->end();
  this->set_size(this->size() + NumToInsert);
  size_t NumOverwritten = OldEnd-I;
  this->uninitialized_move(I, OldEnd, this->end()-NumOverwritten);

  // Replace the overwritten part.
  for (T *J = I; NumOverwritten > 0; --NumOverwritten) {
    *J = *From;
    ++J; ++From;
  }

  // Insert the non-overwritten middle part.
  this->uninitialized_copy(From, To, OldEnd);
  return I;
}

void insert(iterator I, std::initializer_list<T> IL) {
  insert(I, IL.begin(), IL.end());
}

template <typename... ArgTypes> reference emplace_back(ArgTypes &&... Args) {
  if (LLVM_UNLIKELY(this->size() >= this->capacity())__builtin_expect((bool)(this->size() >= this->capacity
()), false))
    return this->growAndEmplaceBack(std::forward<ArgTypes>(Args)...);

  ::new ((void *)this->end()) T(std::forward<ArgTypes>(Args)...);
  this->set_size(this->size() + 1);
  return this->back();
}

SmallVectorImpl &operator=(const SmallVectorImpl &RHS);

SmallVectorImpl &operator=(SmallVectorImpl &&RHS);

bool operator==(const SmallVectorImpl &RHS) const {
  if (this->size() != RHS.size()) return false;
  return std::equal(this->begin(), this->end(), RHS.begin());
}
bool operator!=(const SmallVectorImpl &RHS) const {
  return !(*this == RHS);
}

bool operator<(const SmallVectorImpl &RHS) const {
  return std::lexicographical_compare(this->begin(), this->end(),
                                      RHS.begin(), RHS.end());
}
933};

935template <typename T>
936void SmallVectorImpl<T>::swap(SmallVectorImpl<T> &RHS) {
if (this == &RHS) return;

// We can only avoid copying elements if neither vector is small.
if (!this->isSmall() && !RHS.isSmall()) {
  std::swap(this->BeginX, RHS.BeginX);
  std::swap(this->Size, RHS.Size);
  std::swap(this->Capacity, RHS.Capacity);
  return;
}
this->reserve(RHS.size());
RHS.reserve(this->size());

// Swap the shared elements.
size_t NumShared = this->size();
if (NumShared > RHS.size()) NumShared = RHS.size();
for (size_type i = 0; i != NumShared; ++i)
  std::swap((*this)[i], RHS[i]);

// Copy over the extra elts.
if (this->size() > RHS.size()) {
  size_t EltDiff = this->size() - RHS.size();
  this->uninitialized_copy(this->begin()+NumShared, this->end(), RHS.end());
  RHS.set_size(RHS.size() + EltDiff);
  this->destroy_range(this->begin()+NumShared, this->end());
  this->set_size(NumShared);
} else if (RHS.size() > this->size()) {
  size_t EltDiff = RHS.size() - this->size();
  this->uninitialized_copy(RHS.begin()+NumShared, RHS.end(), this->end());
  this->set_size(this->size() + EltDiff);
  this->destroy_range(RHS.begin()+NumShared, RHS.end());
  RHS.set_size(NumShared);
}
969}

971template <typename T>
972SmallVectorImpl<T> &SmallVectorImpl<T>::
operator=(const SmallVectorImpl<T> &RHS) {
// Avoid self-assignment.
if (this == &RHS) return *this;

// If we already have sufficient space, assign the common elements, then
// destroy any excess.
size_t RHSSize = RHS.size();
size_t CurSize = this->size();
if (CurSize >= RHSSize) {
  // Assign common elements.
  iterator NewEnd;
  if (RHSSize)
    NewEnd = std::copy(RHS.begin(), RHS.begin()+RHSSize, this->begin());
  else
    NewEnd = this->begin();

  // Destroy excess elements.
  this->destroy_range(NewEnd, this->end());

  // Trim.
  this->set_size(RHSSize);
  return *this;
}

// If we have to grow to have enough elements, destroy the current elements.
// This allows us to avoid copying them during the grow.
// FIXME: don't do this if they're efficiently moveable.
if (this->capacity() < RHSSize) {
  // Destroy current elements.
  this->clear();
  CurSize = 0;
  this->grow(RHSSize);
} else if (CurSize) {
  // Otherwise, use assignment for the already-constructed elements.
  std::copy(RHS.begin(), RHS.begin()+CurSize, this->begin());
}

// Copy construct the new elements in place.
this->uninitialized_copy(RHS.begin()+CurSize, RHS.end(),
                         this->begin()+CurSize);

// Set end.
this->set_size(RHSSize);
return *this;
1017}

1019template <typename T>
1020SmallVectorImpl<T> &SmallVectorImpl<T>::operator=(SmallVectorImpl<T> &&RHS) {
// Avoid self-assignment.
if (this == &RHS) return *this;

// If the RHS isn't small, clear this vector and then steal its buffer.
if (!RHS.isSmall()) {
  this->destroy_range(this->begin(), this->end());
  if (!this->isSmall()) free(this->begin());
  this->BeginX = RHS.BeginX;
  this->Size = RHS.Size;
  this->Capacity = RHS.Capacity;
  RHS.resetToSmall();
  return *this;
}

// If we already have sufficient space, assign the common elements, then
// destroy any excess.
size_t RHSSize = RHS.size();
size_t CurSize = this->size();
if (CurSize >= RHSSize) {
  // Assign common elements.
  iterator NewEnd = this->begin();
  if (RHSSize)
    NewEnd = std::move(RHS.begin(), RHS.end(), NewEnd);

  // Destroy excess elements and trim the bounds.
  this->destroy_range(NewEnd, this->end());
  this->set_size(RHSSize);

  // Clear the RHS.
  RHS.clear();

  return *this;
}

// If we have to grow to have enough elements, destroy the current elements.
// This allows us to avoid copying them during the grow.
// FIXME: this may not actually make any sense if we can efficiently move
// elements.
if (this->capacity() < RHSSize) {
  // Destroy current elements.
  this->clear();
  CurSize = 0;
  this->grow(RHSSize);
} else if (CurSize) {
  // Otherwise, use assignment for the already-constructed elements.
  std::move(RHS.begin(), RHS.begin()+CurSize, this->begin());
}

// Move-construct the new elements in place.
this->uninitialized_move(RHS.begin()+CurSize, RHS.end(),
                         this->begin()+CurSize);

// Set end.
this->set_size(RHSSize);

RHS.clear();
return *this;
1078}

1080/// Storage for the SmallVector elements.  This is specialized for the N=0 case
1081/// to avoid allocating unnecessary storage.
1082template <typename T, unsigned N>
1083struct SmallVectorStorage {
alignas(T) char InlineElts[N * sizeof(T)];
1085};

1087/// We need the storage to be properly aligned even for small-size of 0 so that
1088/// the pointer math in \a SmallVectorTemplateCommon::getFirstEl() is
1089/// well-defined.
1090template <typename T> struct alignas(T) SmallVectorStorage<T, 0> {};

1092/// Forward declaration of SmallVector so that
1093/// calculateSmallVectorDefaultInlinedElements can reference
1094/// `sizeof(SmallVector<T, 0>)`.
1095template <typename T, unsigned N> class LLVM_GSL_OWNER[[gsl::Owner]] SmallVector;

1097/// Helper class for calculating the default number of inline elements for
1098/// `SmallVector<T>`.
1099///
1100/// This should be migrated to a constexpr function when our minimum
1101/// compiler support is enough for multi-statement constexpr functions.
1102template <typename T> struct CalculateSmallVectorDefaultInlinedElements {
// Parameter controlling the default number of inlined elements
// for `SmallVector<T>`.
//
// The default number of inlined elements ensures that
// 1. There is at least one inlined element.
// 2. `sizeof(SmallVector<T>) <= kPreferredSmallVectorSizeof` unless
// it contradicts 1.
static constexpr size_t kPreferredSmallVectorSizeof = 64;

// static_assert that sizeof(T) is not "too big".
//
// Because our policy guarantees at least one inlined element, it is possible
// for an arbitrarily large inlined element to allocate an arbitrarily large
// amount of inline storage. We generally consider it an antipattern for a
// SmallVector to allocate an excessive amount of inline storage, so we want
// to call attention to these cases and make sure that users are making an
// intentional decision if they request a lot of inline storage.
//
// We want this assertion to trigger in pathological cases, but otherwise
// not be too easy to hit. To accomplish that, the cutoff is actually somewhat
// larger than kPreferredSmallVectorSizeof (otherwise,
// `SmallVector<SmallVector<T>>` would be one easy way to trip it, and that
// pattern seems useful in practice).
//
// One wrinkle is that this assertion is in theory non-portable, since
// sizeof(T) is in general platform-dependent. However, we don't expect this
// to be much of an issue, because most LLVM development happens on 64-bit
// hosts, and therefore sizeof(T) is expected to *decrease* when compiled for
// 32-bit hosts, dodging the issue. The reverse situation, where development
// happens on a 32-bit host and then fails due to sizeof(T) *increasing* on a
// 64-bit host, is expected to be very rare.
static_assert(
    sizeof(T) <= 256,
    "You are trying to use a default number of inlined elements for "
    "`SmallVector<T>` but `sizeof(T)` is really big! Please use an "
    "explicit number of inlined elements with `SmallVector<T, N>` to make "
    "sure you really want that much inline storage.");

// Discount the size of the header itself when calculating the maximum inline
// bytes.
static constexpr size_t PreferredInlineBytes =
    kPreferredSmallVectorSizeof - sizeof(SmallVector<T, 0>);
static constexpr size_t NumElementsThatFit = PreferredInlineBytes / sizeof(T);
static constexpr size_t value =
    NumElementsThatFit == 0 ? 1 : NumElementsThatFit;
1148};

1150/// This is a 'vector' (really, a variable-sized array), optimized
1151/// for the case when the array is small.  It contains some number of elements
1152/// in-place, which allows it to avoid heap allocation when the actual number of
1153/// elements is below that threshold.  This allows normal "small" cases to be
1154/// fast without losing generality for large inputs.
1155///
1156/// \note
1157/// In the absence of a well-motivated choice for the number of inlined
1158/// elements \p N, it is recommended to use \c SmallVector<T> (that is,
1159/// omitting the \p N). This will choose a default number of inlined elements
1160/// reasonable for allocation on the stack (for example, trying to keep \c
1161/// sizeof(SmallVector<T>) around 64 bytes).
1162///
1163/// \warning This does not attempt to be exception safe.
1164///
1165/// \see https://llvm.org/docs/ProgrammersManual.html#llvm-adt-smallvector-h
1166template <typename T,
        unsigned N = CalculateSmallVectorDefaultInlinedElements<T>::value>
1168class LLVM_GSL_OWNER[[gsl::Owner]] SmallVector : public SmallVectorImpl<T>,
                                 SmallVectorStorage<T, N> {
1170public:
SmallVector() : SmallVectorImpl<T>(N) {}

~SmallVector() {
  // Destroy the constructed elements in the vector.
  this->destroy_range(this->begin(), this->end());
}

explicit SmallVector(size_t Size, const T &Value = T())
  : SmallVectorImpl<T>(N) {
  this->assign(Size, Value);
}

template <typename ItTy,
          typename = std::enable_if_t<std::is_convertible<
              typename std::iterator_traits<ItTy>::iterator_category,
              std::input_iterator_tag>::value>>
SmallVector(ItTy S, ItTy E) : SmallVectorImpl<T>(N) {
  this->append(S, E);
}

template <typename RangeTy>
explicit SmallVector(const iterator_range<RangeTy> &R)
    : SmallVectorImpl<T>(N) {
  this->append(R.begin(), R.end());
}

SmallVector(std::initializer_list<T> IL) : SmallVectorImpl<T>(N) {
  this->assign(IL);
}

SmallVector(const SmallVector &RHS) : SmallVectorImpl<T>(N) {
  if (!RHS.empty())
    SmallVectorImpl<T>::operator=(RHS);
}

SmallVector &operator=(const SmallVector &RHS) {
  SmallVectorImpl<T>::operator=(RHS);
  return *this;
}

SmallVector(SmallVector &&RHS) : SmallVectorImpl<T>(N) {
  if (!RHS.empty())
    SmallVectorImpl<T>::operator=(::std::move(RHS));
}

SmallVector(SmallVectorImpl<T> &&RHS) : SmallVectorImpl<T>(N) {
  if (!RHS.empty())
    SmallVectorImpl<T>::operator=(::std::move(RHS));
}

SmallVector &operator=(SmallVector &&RHS) {
  SmallVectorImpl<T>::operator=(::std::move(RHS));
  return *this;
}

SmallVector &operator=(SmallVectorImpl<T> &&RHS) {
  SmallVectorImpl<T>::operator=(::std::move(RHS));
  return *this;
}

SmallVector &operator=(std::initializer_list<T> IL) {
  this->assign(IL);
  return *this;
}
1235};

1237template <typename T, unsigned N>
1238inline size_t capacity_in_bytes(const SmallVector<T, N> &X) {
return X.capacity_in_bytes();
1240}

1242/// Given a range of type R, iterate the entire range and return a
1243/// SmallVector with elements of the vector.  This is useful, for example,
1244/// when you want to iterate a range and then sort the results.
1245template <unsigned Size, typename R>
1246SmallVector<typename std::remove_const<typename std::remove_reference<
              decltype(*std::begin(std::declval<R &>()))>::type>::type,
          Size>
1249to_vector(R &&Range) {
return {std::begin(Range), std::end(Range)};
1251}

1253} // end namespace llvm

1255namespace std {

/// Implement std::swap in terms of SmallVector swap.
template<typename T>
inline void
swap(llvm::SmallVectorImpl<T> &LHS, llvm::SmallVectorImpl<T> &RHS) {
  LHS.swap(RHS);
}

/// Implement std::swap in terms of SmallVector swap.
template<typename T, unsigned N>
inline void
swap(llvm::SmallVector<T, N> &LHS, llvm::SmallVector<T, N> &RHS) {
  LHS.swap(RHS);
}

1271} // end namespace std

1273#endif // LLVM_ADT_SMALLVECTOR_H