[InstCombine] Don't transform bitcasts between x86_mmx and v1i64 into insertelement/extractelement

[craig.topper]

Summary

x86_mmx is conceptually a vector already. Don't introduce an extra conversion between it and scalar i64.

I'm using VectorType::isValidElementType which checks for floating point, integer, and pointers to hopefully make this more readable than just blacklisting x86_mmx.

[efriedma]

LGTM

[RKSimon]

LGTM

llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp

if (SrcElTy == DstElTy) {
         if (SrcPTy->getAddressSpace() == 0 || !CanBeNull)
           GEP->setIsInBounds();
       }
       return GEP;
     }
   }
 
   if (VectorType *DestVTy = dyn_cast<VectorType>(DestTy)) {
-    if (DestVTy->getNumElements() == 1 && !SrcTy->isVectorTy()) {
+    if (DestVTy->getNumElements() == 1 &&
+        VectorType::isValidElementType(SrcTy)) {
       Value *Elem = Builder.CreateBitCast(Src, DestVTy->getElementType());
       return InsertElementInst::Create(UndefValue::get(DestTy), Elem,
                      Constant::getNullValue(Type::getInt32Ty(CI.getContext())));
       // FIXME: Canonicalize bitcast(insertelement) -> insertelement(bitcast)
     }
 
     if (isa<IntegerType>(SrcTy)) {
       // If this is a cast from an integer to vector, check to see if the input
if (isa<IntegerType>(SrcTy)) {
         return replaceInstUsesWith(CI, V);
     }
   }
 
   if (VectorType *SrcVTy = dyn_cast<VectorType>(SrcTy)) {
     if (SrcVTy->getNumElements() == 1) {
       // If our destination is not a vector, then make this a straight
       // scalar-scalar cast.
-      if (!DestTy->isVectorTy()) {
+      if (VectorType::isValidElementType(DestTy)) {
         Value *Elem =
           Builder.CreateExtractElement(Src,
                      Constant::getNullValue(Type::getInt32Ty(CI.getContext())));
         return CastInst::Create(Instruction::BitCast, Elem, DestTy);
       }
 
       // Otherwise, see if our source is an insert. If so, then use the scalar
       // component directly:

llvm/test/Transforms/InstCombine/bitcast-vec-canon.ll

 ; CHECK-NEXT:    ret <1 x i64> [[C]]
 ;
   %c = bitcast i64 %y to <1 x i64>
   ret <1 x i64> %c
 }
 
 define x86_mmx @e(<1 x i64> %y) {
 ; CHECK-LABEL: @e(
-; CHECK-NEXT:    [[TMP1:%.*]] = extractelement <1 x i64> %y, i32 0
-; CHECK-NEXT:    [[C:%.*]] = bitcast i64 [[TMP1]] to x86_mmx
+; CHECK-NEXT:    [[C:%.*]] = bitcast <1 x i64> %y to x86_mmx
 ; CHECK-NEXT:    ret x86_mmx [[C]]
 ;
   %c = bitcast <1 x i64> %y to x86_mmx
   ret x86_mmx %c
 }
 
 define <1 x i64> @f(x86_mmx %y) {
 ; CHECK-LABEL: @f(
-; CHECK-NEXT:    [[TMP1:%.*]] = bitcast x86_mmx %y to i64
-; CHECK-NEXT:    [[C:%.*]] = insertelement <1 x i64> undef, i64 [[TMP1]], i32 0
+; CHECK-NEXT:    [[C:%.*]] = bitcast x86_mmx %y to <1 x i64>
 ; CHECK-NEXT:    ret <1 x i64> [[C]]
 ;
   %c = bitcast x86_mmx %y to <1 x i64>
   ret <1 x i64> %c
 }