Change memcpy/memset/memmove to have dest and source alignments.

Note, this was reviewed (and more details are in) http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html

These intrinsics currently have an explicit alignment argument which is
required to be a constant integer.  It represents the alignment of the
source and dest, and so must be the minimum of those.

This change allows source and dest to each have their own alignments
by using the alignment attribute on their arguments.  The alignment
argument itself is removed.

There are a few places in the code for which the code needs to be
checked by an expert as to whether using only src/dest alignment is
safe.  For those places, they currently take the minimum of src/dest
alignments which matches the current behaviour.

For example, code which used to read:
  call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 500, i32 8, i1 false)
will now read:
  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 8 %dest, i8* align 8 %src, i32 500, i1 false)

For out of tree owners, I was able to strip alignment from calls using sed by replacing:
  (call.*llvm\.memset.*)i32\ [0-9]*\,\ i1 false\)
with:
  $1i1 false)

and similarly for memmove and memcpy.

I then added back in alignment to test cases which needed it.

A similar commit will be made to clang which actually has many differences in alignment as now
IRBuilder can generate different source/dest alignments on calls.

In IRBuilder itself, a new argument was added.  Instead of calling:
  CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, /* isVolatile */ false)
you now call
  CreateMemCpy(Dst, Src, getInt64(Size), DstAlign, SrcAlign, /* isVolatile */ false)

There is a temporary class (IntegerAlignment) which takes the source alignment and rejects
implicit conversion from bool.  This is to prevent isVolatile here from passing its default
parameter to the source alignment.

Note, changes in future can now be made to codegen.  I didn't change anything here, but this
change should enable better memcpy code sequences.

Reviewed by Hal Finkel.

llvm-svn: 253511

1 files changed, 18 insertions, 13 deletions

diff --git a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
index f80b07bf219..2db2b802f08 100644
--- a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
+++ b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
@@ -229,7 +229,8 @@ public:
 
   void addMemSet(int64_t OffsetFromFirst, MemSetInst *MSI) {
     int64_t Size = cast<ConstantInt>(MSI->getLength())->getZExtValue();
-    addRange(OffsetFromFirst, Size, MSI->getDest(), MSI->getAlignment(), MSI);
+    addRange(OffsetFromFirst, Size, MSI->getDest(), MSI->getDestAlignment(),
+             MSI);
   }
 
   void addRange(int64_t Start, int64_t Size, Value *Ptr,
@@ -819,20 +820,17 @@ bool MemCpyOpt::processMemCpyMemCpyDependence(MemCpyInst *M, MemCpyInst *MDep) {
 
   // If all checks passed, then we can transform M.
 
-  // Make sure to use the lesser of the alignment of the source and the dest
-  // since we're changing where we're reading from, but don't want to increase
-  // the alignment past what can be read from or written to.
   // TODO: Is this worth it if we're creating a less aligned memcpy? For
   // example we could be moving from movaps -> movq on x86.
-  unsigned Align = std::min(MDep->getAlignment(), M->getAlignment());
-
   IRBuilder<> Builder(M);
   if (UseMemMove)
     Builder.CreateMemMove(M->getRawDest(), MDep->getRawSource(), M->getLength(),
-                          Align, M->isVolatile());
+                          M->getDestAlignment(), MDep->getSrcAlignment(),
+                          M->isVolatile());
   else
     Builder.CreateMemCpy(M->getRawDest(), MDep->getRawSource(), M->getLength(),
-                         Align, M->isVolatile());
+                         M->getDestAlignment(), MDep->getSrcAlignment(),
+                         M->isVolatile());
 
   // Remove the instruction we're replacing.
   MD->removeInstruction(M);
@@ -878,7 +876,7 @@ bool MemCpyOpt::processMemSetMemCpyDependence(MemCpyInst *MemCpy,
   // If Dest is aligned, and SrcSize is constant, use the minimum alignment
   // of the sum.
   const unsigned DestAlign =
-      std::max(MemSet->getAlignment(), MemCpy->getAlignment());
+      std::max(MemSet->getDestAlignment(), MemCpy->getDestAlignment());
   if (DestAlign > 1)
     if (ConstantInt *SrcSizeC = dyn_cast<ConstantInt>(SrcSize))
       Align = MinAlign(SrcSizeC->getZExtValue(), DestAlign);
@@ -935,7 +933,7 @@ bool MemCpyOpt::performMemCpyToMemSetOptzn(MemCpyInst *MemCpy,
 
   IRBuilder<> Builder(MemCpy);
   Builder.CreateMemSet(MemCpy->getRawDest(), MemSet->getOperand(1),
-                       CopySize, MemCpy->getAlignment());
+                       CopySize, MemCpy->getDestAlignment());
   return true;
 }
 
@@ -961,7 +959,7 @@ bool MemCpyOpt::processMemCpy(MemCpyInst *M) {
       if (Value *ByteVal = isBytewiseValue(GV->getInitializer())) {
         IRBuilder<> Builder(M);
         Builder.CreateMemSet(M->getRawDest(), ByteVal, M->getLength(),
-                             M->getAlignment(), false);
+                             M->getDestAlignment(), false);
         MD->removeInstruction(M);
         M->eraseFromParent();
         ++NumCpyToSet;
@@ -990,8 +988,11 @@ bool MemCpyOpt::processMemCpy(MemCpyInst *M) {
   //   d) memcpy from a just-memset'd source can be turned into memset.
   if (DepInfo.isClobber()) {
     if (CallInst *C = dyn_cast<CallInst>(DepInfo.getInst())) {
+      // FIXME: Can we pass in either of dest/src alignment here instead of
+      // convervatively taking the minimum?
+      unsigned Align = std::min(M->getDestAlignment(), M->getSrcAlignment());
       if (performCallSlotOptzn(M, M->getDest(), M->getSource(),
-                               CopySize->getZExtValue(), M->getAlignment(),
+                               CopySize->getZExtValue(), Align,
                                C)) {
         MD->removeInstruction(M);
         M->eraseFromParent();
@@ -1108,7 +1109,11 @@ bool MemCpyOpt::processByValArgument(CallSite CS, unsigned ArgNo) {
       getAnalysis<AssumptionCacheTracker>().getAssumptionCache(
           *CS->getParent()->getParent());
   DominatorTree &DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree();
-  if (MDep->getAlignment() < ByValAlign &&
+  // FIXME: Can we use either of dest/src alignment here instead of
+  // convervatively taking the minimum?
+  unsigned MinAlign = std::min(MDep->getDestAlignment(),
+                               MDep->getSrcAlignment());
+  if (MinAlign < ByValAlign &&
       getOrEnforceKnownAlignment(MDep->getSource(), ByValAlign, DL,
                                  CS.getInstruction(), &AC, &DT) < ByValAlign)
     return false;