[RISCV] Create the correct mask type when lowering EXTRACT_VECTOR_ELT

[frasercrmck]

Summary

This particular case was creating a VMSET_VL using the old
fixed-length type in order to pass a mask to other custom nodes
operating on the scalable container type. This kind of thing wasn't
caught for us; I only noticed when experimenting with odd-length
vectors, where it was trying to generate an invalid v3i1 MVT.

[craig.topper]

LGTM

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

case ISD::EXTRACT_VECTOR_ELT: {
     if (VecVT.isFixedLengthVector()) {
       ContainerVT = getContainerForFixedLengthVector(VecVT);
       Vec = convertToScalableVector(ContainerVT, Vec, DAG, Subtarget);
     }
 
     MVT XLenVT = Subtarget.getXLenVT();
 
     // Use a VL of 1 to avoid processing more elements than we need.
-    MVT MaskVT = MVT::getVectorVT(MVT::i1, VecVT.getVectorElementCount());
+    MVT MaskVT = MVT::getVectorVT(MVT::i1, ContainerVT.getVectorElementCount());
     SDValue VL = DAG.getConstant(1, DL, XLenVT);
     SDValue Mask = DAG.getNode(RISCVISD::VMSET_VL, DL, MaskVT, VL);
 
     // Unless the index is known to be 0, we must slide the vector down to get
     // the desired element into index 0.
     if (!isNullConstant(Idx)) {
       Vec = DAG.getNode(RISCVISD::VSLIDEDOWN_VL, DL, ContainerVT,
                         DAG.getUNDEF(ContainerVT), Vec, Idx, Mask, VL);