#include "llvm/Support/Debug.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Transforms/Utils/Local.h"
-#include <list>
+#include <algorithm>
using namespace llvm;
#define DEBUG_TYPE "memcpyopt"
return Offset;
}
-/// IsPointerOffset - Return true if Ptr1 is provably equal to Ptr2 plus a
-/// constant offset, and return that constant offset. For example, Ptr1 might
-/// be &A[42], and Ptr2 might be &A[40]. In this case offset would be -8.
+/// Return true if Ptr1 is provably equal to Ptr2 plus a constant offset, and
+/// return that constant offset. For example, Ptr1 might be &A[42], and Ptr2
+/// might be &A[40]. In this case offset would be -8.
static bool IsPointerOffset(Value *Ptr1, Value *Ptr2, int64_t &Offset,
const DataLayout &DL) {
Ptr1 = Ptr1->stripPointerCasts();
}
-/// MemsetRange - Represents a range of memset'd bytes with the ByteVal value.
+/// Represents a range of memset'd bytes with the ByteVal value.
/// This allows us to analyze stores like:
/// store 0 -> P+1
/// store 0 -> P+0
namespace {
class MemsetRanges {
- /// Ranges - A sorted list of the memset ranges. We use std::list here
- /// because each element is relatively large and expensive to copy.
- std::list<MemsetRange> Ranges;
- typedef std::list<MemsetRange>::iterator range_iterator;
+ /// A sorted list of the memset ranges.
+ SmallVector<MemsetRange, 8> Ranges;
+ typedef SmallVectorImpl<MemsetRange>::iterator range_iterator;
const DataLayout &DL;
public:
MemsetRanges(const DataLayout &DL) : DL(DL) {}
- typedef std::list<MemsetRange>::const_iterator const_iterator;
+ typedef SmallVectorImpl<MemsetRange>::const_iterator const_iterator;
const_iterator begin() const { return Ranges.begin(); }
const_iterator end() const { return Ranges.end(); }
bool empty() const { return Ranges.empty(); }
} // end anon namespace
-/// addRange - Add a new store to the MemsetRanges data structure. This adds a
+/// Add a new store to the MemsetRanges data structure. This adds a
/// new range for the specified store at the specified offset, merging into
/// existing ranges as appropriate.
-///
-/// Do a linear search of the ranges to see if this can be joined and/or to
-/// find the insertion point in the list. We keep the ranges sorted for
-/// simplicity here. This is a linear search of a linked list, which is ugly,
-/// however the number of ranges is limited, so this won't get crazy slow.
void MemsetRanges::addRange(int64_t Start, int64_t Size, Value *Ptr,
unsigned Alignment, Instruction *Inst) {
int64_t End = Start+Size;
- range_iterator I = Ranges.begin(), E = Ranges.end();
- while (I != E && Start > I->End)
- ++I;
+ range_iterator I = std::lower_bound(Ranges.begin(), Ranges.end(), Start,
+ [](const MemsetRange &LHS, int64_t RHS) { return LHS.End < RHS; });
// 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 == E || End < I->Start) {
+ if (I == Ranges.end() || End < I->Start) {
MemsetRange &R = *Ranges.insert(I, MemsetRange());
R.Start = Start;
R.End = End;
if (End > I->End) {
I->End = End;
range_iterator NextI = I;
- while (++NextI != E && End >= NextI->Start) {
+ 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)
AU.addPreserved<MemoryDependenceAnalysis>();
}
- // Helper fuctions
+ // Helper functions
bool processStore(StoreInst *SI, BasicBlock::iterator &BBI);
bool processMemSet(MemSetInst *SI, BasicBlock::iterator &BBI);
bool processMemCpy(MemCpyInst *M);
char MemCpyOpt::ID = 0;
}
-// createMemCpyOptPass - The public interface to this file...
+/// The public interface to this file...
FunctionPass *llvm::createMemCpyOptPass() { return new MemCpyOpt(); }
INITIALIZE_PASS_BEGIN(MemCpyOpt, "memcpyopt", "MemCpy Optimization",
INITIALIZE_PASS_END(MemCpyOpt, "memcpyopt", "MemCpy Optimization",
false, false)
-/// tryMergingIntoMemset - When scanning forward over instructions, we look for
-/// some other patterns to fold away. In particular, this looks for stores to
-/// neighboring locations of memory. If it sees enough consecutive ones, it
-/// attempts to merge them together into a memcpy/memset.
+/// When scanning forward over instructions, we look for some other patterns to
+/// fold away. In particular, this looks for stores to neighboring locations of
+/// memory. If it sees enough consecutive ones, it attempts to merge them
+/// together into a memcpy/memset.
Instruction *MemCpyOpt::tryMergingIntoMemset(Instruction *StartInst,
Value *StartPtr, Value *ByteVal) {
const DataLayout &DL = StartInst->getModule()->getDataLayout();
// Check that nothing touches the dest of the "copy" between
// the call and the store.
AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
- AliasAnalysis::Location StoreLoc = AA.getLocation(SI);
+ MemoryLocation StoreLoc = MemoryLocation::get(SI);
for (BasicBlock::iterator I = --BasicBlock::iterator(SI),
E = C; I != E; --I) {
- if (AA.getModRefInfo(&*I, StoreLoc) != AliasAnalysis::NoModRef) {
+ if (AA.getModRefInfo(&*I, StoreLoc) != MRI_NoModRef) {
C = nullptr;
break;
}
}
-/// performCallSlotOptzn - takes a memcpy and a call that it depends on,
+/// Takes a memcpy and a call that it depends on,
/// and checks for the possibility of a call slot optimization by having
/// the call write its result directly into the destination of the memcpy.
bool MemCpyOpt::performCallSlotOptzn(Instruction *cpy,
// 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.
AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
- AliasAnalysis::ModRefResult MR = AA.getModRefInfo(C, cpyDest, srcSize);
+ ModRefInfo MR = AA.getModRefInfo(C, cpyDest, srcSize);
// If necessary, perform additional analysis.
- if (MR != AliasAnalysis::NoModRef)
+ if (MR != MRI_NoModRef)
MR = AA.callCapturesBefore(C, cpyDest, srcSize, &DT);
- if (MR != AliasAnalysis::NoModRef)
+ if (MR != MRI_NoModRef)
return false;
// All the checks have passed, so do the transformation.
return true;
}
-/// processMemCpyMemCpyDependence - We've found that the (upward scanning)
-/// memory dependence of memcpy 'M' is the memcpy 'MDep'. Try to simplify M to
-/// copy from MDep's input if we can.
-///
+/// We've found that the (upward scanning) memory dependence of memcpy 'M' is
+/// the memcpy 'MDep'. Try to simplify M to copy from MDep's input if we can.
bool MemCpyOpt::processMemCpyMemCpyDependence(MemCpyInst *M, MemCpyInst *MDep) {
// We can only transforms memcpy's where the dest of one is the source of the
// other.
//
// NOTE: This is conservative, it will stop on any read from the source loc,
// not just the defining memcpy.
- MemDepResult SourceDep =
- MD->getPointerDependencyFrom(AA.getLocationForSource(MDep),
- false, M, M->getParent());
+ MemDepResult SourceDep = MD->getPointerDependencyFrom(
+ MemoryLocation::getForSource(MDep), false, M, M->getParent());
if (!SourceDep.isClobber() || SourceDep.getInst() != MDep)
return false;
// source and dest might overlap. We still want to eliminate the intermediate
// value, but we have to generate a memmove instead of memcpy.
bool UseMemMove = false;
- if (!AA.isNoAlias(AA.getLocationForDest(M), AA.getLocationForSource(MDep)))
+ if (!AA.isNoAlias(MemoryLocation::getForDest(M),
+ MemoryLocation::getForSource(MDep)))
UseMemMove = true;
// If all checks passed, then we can transform M.
if (MemSet->getDest() != MemCpy->getDest())
return false;
+ // Check that there are no other dependencies on the memset destination.
+ MemDepResult DstDepInfo = MD->getPointerDependencyFrom(
+ MemoryLocation::getForDest(MemSet), false, MemCpy, MemCpy->getParent());
+ if (DstDepInfo.getInst() != MemSet)
+ return false;
+
// Use the same i8* dest as the memcpy, killing the memset dest if different.
Value *Dest = MemCpy->getRawDest();
Value *DestSize = MemSet->getLength();
if (ConstantInt *SrcSizeC = dyn_cast<ConstantInt>(SrcSize))
Align = MinAlign(SrcSizeC->getZExtValue(), DestAlign);
- IRBuilder<> Builder(MemSet);
+ IRBuilder<> Builder(MemCpy);
// If the sizes have different types, zext the smaller one.
if (DestSize->getType() != SrcSize->getType()) {
return true;
}
-/// processMemCpy - perform simplification of memcpy's. If we have memcpy A
+/// Perform simplification of memcpy's. If we have memcpy A
/// which copies X to Y, and memcpy B which copies Y to Z, then we can rewrite
/// B to be a memcpy from X to Z (or potentially a memmove, depending on
/// circumstances). This allows later passes to remove the first memcpy
}
}
- AliasAnalysis::Location SrcLoc = AliasAnalysis::getLocationForSource(M);
+ MemoryLocation SrcLoc = MemoryLocation::getForSource(M);
MemDepResult SrcDepInfo = MD->getPointerDependencyFrom(SrcLoc, true,
M, M->getParent());
return false;
}
-/// processMemMove - Transforms memmove calls to memcpy calls when the src/dst
-/// are guaranteed not to alias.
+/// Transforms memmove calls to memcpy calls when the src/dst are guaranteed
+/// not to alias.
bool MemCpyOpt::processMemMove(MemMoveInst *M) {
AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
return false;
// See if the pointers alias.
- if (!AA.isNoAlias(AA.getLocationForDest(M), AA.getLocationForSource(M)))
+ if (!AA.isNoAlias(MemoryLocation::getForDest(M),
+ MemoryLocation::getForSource(M)))
return false;
DEBUG(dbgs() << "MemCpyOpt: Optimizing memmove -> memcpy: " << *M << "\n");
return true;
}
-/// processByValArgument - This is called on every byval argument in call sites.
+/// This is called on every byval argument in call sites.
bool MemCpyOpt::processByValArgument(CallSite CS, unsigned ArgNo) {
const DataLayout &DL = CS.getCaller()->getParent()->getDataLayout();
// Find out what feeds this byval argument.
Value *ByValArg = CS.getArgument(ArgNo);
Type *ByValTy = cast<PointerType>(ByValArg->getType())->getElementType();
uint64_t ByValSize = DL.getTypeAllocSize(ByValTy);
- MemDepResult DepInfo =
- MD->getPointerDependencyFrom(AliasAnalysis::Location(ByValArg, ByValSize),
- true, CS.getInstruction(),
- CS.getInstruction()->getParent());
+ MemDepResult DepInfo = MD->getPointerDependencyFrom(
+ MemoryLocation(ByValArg, ByValSize), true, CS.getInstruction(),
+ CS.getInstruction()->getParent());
if (!DepInfo.isClobber())
return false;
// NOTE: This is conservative, it will stop on any read from the source loc,
// not just the defining memcpy.
MemDepResult SourceDep =
- MD->getPointerDependencyFrom(AliasAnalysis::getLocationForSource(MDep),
- false, CS.getInstruction(), MDep->getParent());
+ MD->getPointerDependencyFrom(MemoryLocation::getForSource(MDep), false,
+ CS.getInstruction(), MDep->getParent());
if (!SourceDep.isClobber() || SourceDep.getInst() != MDep)
return false;
return true;
}
-/// iterateOnFunction - Executes one iteration of MemCpyOpt.
+/// Executes one iteration of MemCpyOpt.
bool MemCpyOpt::iterateOnFunction(Function &F) {
bool MadeChange = false;
return MadeChange;
}
-// MemCpyOpt::runOnFunction - This is the main transformation entry point for a
-// function.
-//
+/// This is the main transformation entry point for a function.
bool MemCpyOpt::runOnFunction(Function &F) {
if (skipOptnoneFunction(F))
return false;