[NFC] Bail out for scalable vectors before calling getNumElements

[ctetreau]

Summary

Move the bail out logic to before constructing the Result and Lane
vectors. This is both potentially faster, and avoids calling
getNumElements on a potentially scalable vector

[fpetrogalli]

llvm/lib/Analysis/ConstantFolding.cpp

2697

Given that from now on the vector is supposed to have a fixed size, maybe it is also worth defining auto FixedVTy = cast<FixedVectorType>(VTy) and use FixedVTy in place of VTy all through the end of the function?

[ctetreau]

address code review issues

[fpetrogalli]

LGTM, thanks!

llvm/lib/Analysis/ConstantFolding.cpp

 
 static Constant *ConstantFoldVectorCall(StringRef Name,
                                         Intrinsic::ID IntrinsicID,
                                         VectorType *VTy,
                                         ArrayRef<Constant *> Operands,
                                         const DataLayout &DL,
                                         const TargetLibraryInfo *TLI,
                                         const CallBase *Call) {
-  SmallVector<Constant *, 4> Result(VTy->getNumElements());
-  SmallVector<Constant *, 4> Lane(Operands.size());
-  Type *Ty = VTy->getElementType();
-
   // Do not iterate on scalable vector. The number of elements is unknown at
   // compile-time.
   if (isa<ScalableVectorType>(VTy))
     return nullptr;
 
+  auto *FVTy = cast<FixedVectorType>(VTy);
+
+  SmallVector<Constant *, 4> Result(FVTy->getNumElements());
+  SmallVector<Constant *, 4> Lane(Operands.size());
+  Type *Ty = FVTy->getElementType();
+
   if (IntrinsicID == Intrinsic::masked_load) {
     auto *SrcPtr = Operands[0];
     auto *Mask = Operands[2];
     auto *Passthru = Operands[3];
 
-    Constant *VecData = ConstantFoldLoadFromConstPtr(SrcPtr, VTy, DL);
+    Constant *VecData = ConstantFoldLoadFromConstPtr(SrcPtr, FVTy, DL);
 
     SmallVector<Constant *, 32> NewElements;
-    for (unsigned I = 0, E = VTy->getNumElements(); I != E; ++I) {
+    for (unsigned I = 0, E = FVTy->getNumElements(); I != E; ++I) {
       auto *MaskElt = Mask->getAggregateElement(I);
       if (!MaskElt)
         break;
       auto *PassthruElt = Passthru->getAggregateElement(I);
       auto *VecElt = VecData ? VecData->getAggregateElement(I) : nullptr;
       if (isa<UndefValue>(MaskElt)) {
         if (PassthruElt)
           NewElements.push_back(PassthruElt);
for (unsigned I = 0, E = FVTy->getNumElements(); I != E; ++I) {
       } else if (MaskElt->isOneValue()) {
         if (!VecElt)
           return nullptr;
         NewElements.push_back(VecElt);
       } else {
         return nullptr;
       }
     }
-    if (NewElements.size() != VTy->getNumElements())
+    if (NewElements.size() != FVTy->getNumElements())
       return nullptr;
     return ConstantVector::get(NewElements);
   }
 
-  for (unsigned I = 0, E = VTy->getNumElements(); I != E; ++I) {
+  for (unsigned I = 0, E = FVTy->getNumElements(); I != E; ++I) {
     // Gather a column of constants.
     for (unsigned J = 0, JE = Operands.size(); J != JE; ++J) {
       // Some intrinsics use a scalar type for certain arguments.
       if (hasVectorInstrinsicScalarOpd(IntrinsicID, J)) {
         Lane[J] = Operands[J];
         continue;
       }