Supporting inline memmove isn't going to be worthwhile. The only way to avoid

violating a dependency is to emit all loads prior to stores.  This would likely
cause a great deal of spillage offsetting any potential gains.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144585 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 9 insertions, 16 deletions

diff --git a/lib/Target/ARM/ARMFastISel.cpp b/lib/Target/ARM/ARMFastISel.cpp
index 67ba14b..4df084f 100644
--- a/lib/Target/ARM/ARMFastISel.cpp
+++ b/lib/Target/ARM/ARMFastISel.cpp
@@ -185,9 +185,8 @@ class ARMFastISel : public FastISel {
     bool ARMEmitStore(EVT VT, unsigned SrcReg, Address &Addr);
     bool ARMComputeAddress(const Value *Obj, Address &Addr);
     void ARMSimplifyAddress(Address &Addr, EVT VT, bool useAM3);
-    bool ARMIsMemXferSmall(uint64_t Len);
-    bool ARMTryEmitSmallMemXfer(Address Dest, Address Src, uint64_t Len,
-                                bool isMemCpy);
+    bool ARMIsMemCpySmall(uint64_t Len);
+    bool ARMTryEmitSmallMemCpy(Address Dest, Address Src, uint64_t Len);
     unsigned ARMEmitIntExt(EVT SrcVT, unsigned SrcReg, EVT DestVT, bool isZExt);
     unsigned ARMMaterializeFP(const ConstantFP *CFP, EVT VT);
     unsigned ARMMaterializeInt(const Constant *C, EVT VT);
@@ -2196,19 +2195,13 @@ bool ARMFastISel::SelectCall(const Instruction *I,
   return true;
 }
 
-bool ARMFastISel::ARMIsMemXferSmall(uint64_t Len) {
+bool ARMFastISel::ARMIsMemCpySmall(uint64_t Len) {
   return Len <= 16;
 }
 
-bool ARMFastISel::ARMTryEmitSmallMemXfer(Address Dest, Address Src, uint64_t Len,
-                                         bool isMemCpy) {
-  // FIXME: Memmove's require a little more care because their source and
-  // destination may overlap.
-  if (!isMemCpy)
-    return false;
-
+bool ARMFastISel::ARMTryEmitSmallMemCpy(Address Dest, Address Src, uint64_t Len) {
   // Make sure we don't bloat code by inlining very large memcpy's.
-  if (!ARMIsMemXferSmall(Len))
+  if (!ARMIsMemCpySmall(Len))
     return false;
 
   // We don't care about alignment here since we just emit integer accesses.
@@ -2254,15 +2247,15 @@ bool ARMFastISel::SelectIntrinsicCall(const IntrinsicInst &I) {
     // we would emit dead code because we don't currently handle memmoves.
     bool isMemCpy = (I.getIntrinsicID() == Intrinsic::memcpy);
     if (isa<ConstantInt>(MTI.getLength()) && isMemCpy) {
-      // Small memcpy/memmove's are common enough that we want to do them
-      // without a call if possible.
+      // Small memcpy's are common enough that we want to do them without a call
+      // if possible.
       uint64_t Len = cast<ConstantInt>(MTI.getLength())->getZExtValue();
-      if (ARMIsMemXferSmall(Len)) {
+      if (ARMIsMemCpySmall(Len)) {
         Address Dest, Src;
         if (!ARMComputeAddress(MTI.getRawDest(), Dest) ||
             !ARMComputeAddress(MTI.getRawSource(), Src))
           return false;
-        if (ARMTryEmitSmallMemXfer(Dest, Src, Len, isMemCpy))
+        if (ARMTryEmitSmallMemCpy(Dest, Src, Len))
           return true;
       }
     }