[ARM] Mve vector shuffles

[dmgreen]

Summary

This patch tries to add the necessary shuffle vector and buildvector support for ARM MVE. It essentially adds support for VDUP, VREVs and some VMOVs, which are often required by other code (like the upcoming DIV patch).

This mostly uses the same code as for Neon that already generated NEONvdup/NEONvduplane/NEONvrev's. These have been renamed to ARMvdup/etc and moved to ARMInstrInfo as they are common to both architectures. Most of the selection code seems to be applicable to both, but NEON does have some more instructions making some parts specific.

Most code originally by David Sherwood. I've put the patterns near the instructions they effect, but not in them.

[SjoerdMeijer]

llvm/lib/Target/ARM/ARMISelLowering.cpp

6888

a proper nit: perhaps Mve -> MVE for consistency

7191

There's a lot of prior art here of magic constants and shifts etc., but this doesn't mean much to me.
It probably makes sense when you're familiar with querying the shuffle table.
Not sure if it is worth a comment, but from a quick look at code here, that might not be even necessary.

llvm/test/CodeGen/Thumb2/mve-shuffle.ll

2

Do we need tests when we don't have HasMVEInt? Or is that not really useful (or done elsewhere already)?

[dmgreen]

llvm/test/CodeGen/Thumb2/mve-shuffle.ll

2

Yeah, This certainly sounds like something that should work. We may need to make some adjustments first to the calling convention and what is a legal type.

[dmgreen]

Rebase and change capitalisation. I have not yet added the +mve run line.

[SjoerdMeijer]

We may need to make some adjustments first to the calling convention and what is a legal type.

Okay, if that turns out to be the case, I guess that's probably best addressed in a separate patch.

I hope that all the stuff related to passing/returning vectors is okay.
The only place I might expect some subtleties is here:

@vdup_f16(half* %src1, half* %src2)

which just passes a half, because I think so far we expect the IR to be prepared slightly differently (for inspiration, see e.g. test/CodeGen/ARM/fp16-instructions.ll, although I expect it will mostly work).

[dmgreen]

OK. Now that D60708 is in this is a rebase on what that became. Some of this patch has changed a little as a result and in getting +mve to work. It appears that +mve without a fp is currently broken, as it is add registers without making the operations illegal.

The changes to addMVEVectorTypes have also been moved here, because of how much they have changed. I also needed to fix an fp16 bug in one of the buildvector routines and added a couple of extra patterns.

The tests now test +mve,+fullfp16 and +mve.fp, but not yet +mve which can be added later once that works sensibly.

[SjoerdMeijer]

Ok, sounds reasonable, one step at a time.

[dmgreen]

Oh, and the patterns have been moved to use HasMVEInt, even if they are float operations, as we still have the instructions required for them.

llvm/lib/Target/ARM/ARMISelLowering.cpp

void ARMTargetLowering::addQRTypeForNEON(MVT VT) {
   addTypeForNEON(VT, MVT::v2f64, MVT::v4i32);
 }
 
 void ARMTargetLowering::addMVEITypes() {
   const MVT iTypes[] = { MVT::v16i8, MVT::v8i16, MVT::v4i32 };
 
   for (auto VT : iTypes) {
     addRegisterClass(VT, &ARM::QPRRegClass);
+    setOperationAction(ISD::VECTOR_SHUFFLE, VT, Custom);
+    setOperationAction(ISD::INSERT_VECTOR_ELT, VT, Custom);
+    setOperationAction(ISD::EXTRACT_VECTOR_ELT, VT, Custom);
+    setOperationAction(ISD::BUILD_VECTOR, VT, Custom);
   }
 
   addRegisterClass(MVT::v2i64, &ARM::QPRRegClass);
   for (unsigned Opc = 0; Opc < ISD::BUILTIN_OP_END; ++Opc)
     setOperationAction(Opc, MVT::v2i64, Expand);
   setOperationAction(ISD::BITCAST, MVT::v2i64, Legal);
   setOperationAction(ISD::LOAD, MVT::v2i64, Legal);
   setOperationAction(ISD::STORE, MVT::v2i64, Legal);
 }
 
 void ARMTargetLowering::addMVEFPTypes() {
   const MVT fTypes[] = { MVT::v4f32, MVT::v8f16 };
 
   for (MVT VT : fTypes) {
     addRegisterClass(VT, &ARM::QPRRegClass);
+    setOperationAction(ISD::VECTOR_SHUFFLE, VT, Custom);
+    setOperationAction(ISD::INSERT_VECTOR_ELT, VT, Custom);
+    setOperationAction(ISD::EXTRACT_VECTOR_ELT, VT, Custom);
+    setOperationAction(ISD::BUILD_VECTOR, VT, Custom);
 
     setOperationAction(ISD::FMINNUM, VT, Legal);
     setOperationAction(ISD::FMAXNUM, VT, Legal);
   }
 
   addRegisterClass(MVT::v2f64, &ARM::QPRRegClass);
   for (unsigned Opc = 0; Opc < ISD::BUILTIN_OP_END; ++Opc)
     setOperationAction(Opc, MVT::v2f64, Expand);
SDValue ARMTargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
 
   APInt SplatBits, SplatUndef;
   unsigned SplatBitSize;
   bool HasAnyUndefs;
   if (BVN->isConstantSplat(SplatBits, SplatUndef, SplatBitSize, HasAnyUndefs)) {
     if (SplatUndef.isAllOnesValue())
       return DAG.getUNDEF(VT);
 
-    if (SplatBitSize <= 64) {
+    if (ST->hasNEON() && SplatBitSize <= 64) {
       // Check if an immediate VMOV works.
       EVT VmovVT;
       SDValue Val = isNEONModifiedImm(SplatBits.getZExtValue(),
                                       SplatUndef.getZExtValue(), SplatBitSize,
                                       DAG, dl, VmovVT, VT.is128BitVector(),
                                       VMOVModImm);
       if (Val.getNode()) {
         SDValue Vmov = DAG.getNode(ARMISD::VMOVIMM, dl, VmovVT, Val);
SDValue ARMTargetLowering::LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
 
   // Empirical tests suggest this is rarely worth it for vectors of length <= 2.
   if (NumElts >= 4) {
     SDValue shuffle = ReconstructShuffle(Op, DAG);
     if (shuffle != SDValue())
       return shuffle;
   }
 
-  if (VT.is128BitVector() && VT != MVT::v2f64 && VT != MVT::v4f32) {
+  if (ST->hasNEON() && VT.is128BitVector() && VT != MVT::v2f64 && VT != MVT::v4f32) {
     // If we haven't found an efficient lowering, try splitting a 128-bit vector
     // into two 64-bit vectors; we might discover a better way to lower it.
     SmallVector<SDValue, 64> Ops(Op->op_begin(), Op->op_begin() + NumElts);
     EVT ExtVT = VT.getVectorElementType();
     EVT HVT = EVT::getVectorVT(*DAG.getContext(), ExtVT, NumElts / 2);
     SDValue Lower =
         DAG.getBuildVector(HVT, dl, makeArrayRef(&Ops[0], NumElts / 2));
     if (Lower.getOpcode() == ISD::BUILD_VECTOR)
       Lower = LowerBUILD_VECTOR(Lower, DAG, ST);
     SDValue Upper = DAG.getBuildVector(
         HVT, dl, makeArrayRef(&Ops[NumElts / 2], NumElts / 2));
     if (Upper.getOpcode() == ISD::BUILD_VECTOR)
       Upper = LowerBUILD_VECTOR(Upper, DAG, ST);
     if (Lower && Upper)
       return DAG.getNode(ISD::CONCAT_VECTORS, dl, VT, Lower, Upper);
   }
 
   // Vectors with 32- or 64-bit elements can be built by directly assigning
   // the subregisters.  Lower it to an ARMISD::BUILD_VECTOR so the operands
   // will be legalized.
-  if (EltSize >= 32) {
+  if (ST->hasNEON() && EltSize >= 32) {
     // Do the expansion with floating-point types, since that is what the VFP
     // registers are defined to use, and since i64 is not legal.
     EVT EltVT = EVT::getFloatingPointVT(EltSize);
     EVT VecVT = EVT::getVectorVT(*DAG.getContext(), EltVT, NumElts);
     SmallVector<SDValue, 8> Ops;
     for (unsigned i = 0; i < NumElts; ++i)
       Ops.push_back(DAG.getNode(ISD::BITCAST, dl, EltVT, Op.getOperand(i)));
     SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, Ops);
SDValue ARMTargetLowering::ReconstructShuffle(SDValue Op,
   for (unsigned i = 0; i < Sources.size(); ++i)
     ShuffleOps[i] = Sources[i].ShuffleVec;
 
   SDValue Shuffle = DAG.getVectorShuffle(ShuffleVT, dl, ShuffleOps[0],
                                          ShuffleOps[1], Mask);
   return DAG.getNode(ISD::BITCAST, dl, VT, Shuffle);
 }
 
+enum ShuffleOpCodes {
+  OP_COPY = 0, // Copy, used for things like <u,u,u,3> to say it is <0,1,2,3>
+  OP_VREV,
+  OP_VDUP0,
+  OP_VDUP1,
+  OP_VDUP2,
+  OP_VDUP3,
+  OP_VEXT1,
+  OP_VEXT2,
+  OP_VEXT3,
+  OP_VUZPL, // VUZP, left result
+  OP_VUZPR, // VUZP, right result
+  OP_VZIPL, // VZIP, left result
+  OP_VZIPR, // VZIP, right result
+  OP_VTRNL, // VTRN, left result
+  OP_VTRNR  // VTRN, right result
+};
+
+static bool isLegalMveShuffleOp(unsigned PFEntry) {
+  unsigned OpNum = (PFEntry >> 26) & 0x0F;
+  switch (OpNum) {
+  case OP_COPY:
+  case OP_VREV:
+  case OP_VDUP0:
+  case OP_VDUP1:
+  case OP_VDUP2:
+  case OP_VDUP3:
+    return true;
+  }
+  return false;
+}
+
 /// isShuffleMaskLegal - Targets can use this to indicate that they only
 /// support *some* VECTOR_SHUFFLE operations, those with specific masks.
 /// By default, if a target supports the VECTOR_SHUFFLE node, all mask values
 /// are assumed to be legal.
 bool ARMTargetLowering::isShuffleMaskLegal(ArrayRef<int> M, EVT VT) const {
   if (VT.getVectorNumElements() == 4 &&
       (VT.is128BitVector() || VT.is64BitVector())) {
     unsigned PFIndexes[4];
     for (unsigned i = 0; i != 4; ++i) {
       if (M[i] < 0)
         PFIndexes[i] = 8;
       else
         PFIndexes[i] = M[i];
     }
 
     // Compute the index in the perfect shuffle table.
     unsigned PFTableIndex =
       PFIndexes[0]*9*9*9+PFIndexes[1]*9*9+PFIndexes[2]*9+PFIndexes[3];
     unsigned PFEntry = PerfectShuffleTable[PFTableIndex];
     unsigned Cost = (PFEntry >> 30);
 
-    if (Cost <= 4)
+    if (Cost <= 4 && (Subtarget->hasNEON() || isLegalMveShuffleOp(PFEntry)))
       return true;
   }
 
   bool ReverseVEXT, isV_UNDEF;
   unsigned Imm, WhichResult;
 
   unsigned EltSize = VT.getScalarSizeInBits();
-  return (EltSize >= 32 ||
-          ShuffleVectorSDNode::isSplatMask(&M[0], VT) ||
-          isVREVMask(M, VT, 64) ||
-          isVREVMask(M, VT, 32) ||
-          isVREVMask(M, VT, 16) ||
-          isVEXTMask(M, VT, ReverseVEXT, Imm) ||
+  if (EltSize >= 32 ||
+      isVREVMask(M, VT, 16))
+    return true;
+  else if (Subtarget->hasNEON() &&
+           (isVEXTMask(M, VT, ReverseVEXT, Imm) ||
             isVTBLMask(M, VT) ||
-          isNEONTwoResultShuffleMask(M, VT, WhichResult, isV_UNDEF) ||
-          ((VT == MVT::v8i16 || VT == MVT::v16i8) && isReverseMask(M, VT)));
+            isNEONTwoResultShuffleMask(M, VT, WhichResult, isV_UNDEF)))
+    return true;
+  else if (Subtarget->hasNEON() && (VT == MVT::v8i16 || VT == MVT::v16i8) &&
+           isReverseMask(M, VT))
+    return true;
+  else
+    return false;
 }
 
 /// GeneratePerfectShuffle - Given an entry in the perfect-shuffle table, emit
 /// the specified operations to build the shuffle.
 static SDValue GeneratePerfectShuffle(unsigned PFEntry, SDValue LHS,
                                       SDValue RHS, SelectionDAG &DAG,
                                       const SDLoc &dl) {
   unsigned OpNum = (PFEntry >> 26) & 0x0F;
   unsigned LHSID = (PFEntry >> 13) & ((1 << 13)-1);
   unsigned RHSID = (PFEntry >>  0) & ((1 << 13)-1);
 
-  enum {
-    OP_COPY = 0, // Copy, used for things like <u,u,u,3> to say it is <0,1,2,3>
-    OP_VREV,
-    OP_VDUP0,
-    OP_VDUP1,
-    OP_VDUP2,
-    OP_VDUP3,
-    OP_VEXT1,
-    OP_VEXT2,
-    OP_VEXT3,
-    OP_VUZPL, // VUZP, left result
-    OP_VUZPR, // VUZP, right result
-    OP_VZIPL, // VZIP, left result
-    OP_VZIPR, // VZIP, right result
-    OP_VTRNL, // VTRN, left result
-    OP_VTRNR  // VTRN, right result
-  };
-
   if (OpNum == OP_COPY) {
     if (LHSID == (1*9+2)*9+3) return LHS;
     assert(LHSID == ((4*9+5)*9+6)*9+7 && "Illegal OP_COPY!");
     return RHS;
   }
 
   SDValue OpLHS, OpRHS;
   OpLHS = GeneratePerfectShuffle(PerfectShuffleTable[LHSID], LHS, RHS, DAG, dl);
static SDValue LowerReverse_VECTOR_SHUFFLEv16i8_v8i16(SDValue Op,
   // For a v16i8 type: After the VREV, we have got <8, ...15, 8, ..., 0>. Now,
   // extract the first 8 bytes into the top double word and the last 8 bytes
   // into the bottom double word. The v8i16 case is similar.
   unsigned ExtractNum = (VT == MVT::v16i8) ? 8 : 4;
   return DAG.getNode(ARMISD::VEXT, DL, VT, OpLHS, OpLHS,
                      DAG.getConstant(ExtractNum, DL, MVT::i32));
 }
 
-static SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) {
+static SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG,
+                                   const ARMSubtarget *ST) {
   SDValue V1 = Op.getOperand(0);
   SDValue V2 = Op.getOperand(1);
   SDLoc dl(Op);
   EVT VT = Op.getValueType();
   ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(Op.getNode());
 
   // Convert shuffles that are directly supported on NEON to target-specific
   // DAG nodes, instead of keeping them as shuffles and matching them again
if (SVN->isSplat()) {
           return DAG.getNode(ARMISD::VDUP, dl, VT, V1.getOperand(0));
       }
       return DAG.getNode(ARMISD::VDUPLANE, dl, VT, V1,
                          DAG.getConstant(Lane, dl, MVT::i32));
     }
 
     bool ReverseVEXT = false;
     unsigned Imm = 0;
-    if (isVEXTMask(ShuffleMask, VT, ReverseVEXT, Imm)) {
+    if (ST->hasNEON() && isVEXTMask(ShuffleMask, VT, ReverseVEXT, Imm)) {
       if (ReverseVEXT)
         std::swap(V1, V2);
       return DAG.getNode(ARMISD::VEXT, dl, VT, V1, V2,
                          DAG.getConstant(Imm, dl, MVT::i32));
     }
 
     if (isVREVMask(ShuffleMask, VT, 64))
       return DAG.getNode(ARMISD::VREV64, dl, VT, V1);
     if (isVREVMask(ShuffleMask, VT, 32))
       return DAG.getNode(ARMISD::VREV32, dl, VT, V1);
     if (isVREVMask(ShuffleMask, VT, 16))
       return DAG.getNode(ARMISD::VREV16, dl, VT, V1);
 
-    if (V2->isUndef() && isSingletonVEXTMask(ShuffleMask, VT, Imm)) {
+    if (ST->hasNEON() && V2->isUndef() && isSingletonVEXTMask(ShuffleMask, VT, Imm)) {
       return DAG.getNode(ARMISD::VEXT, dl, VT, V1, V1,
                          DAG.getConstant(Imm, dl, MVT::i32));
     }
 
     // Check for Neon shuffles that modify both input vectors in place.
     // If both results are used, i.e., if there are two shuffles with the same
     // source operands and with masks corresponding to both results of one of
     // these operations, DAG memoization will ensure that a single node is
     // used for both shuffles.
     unsigned WhichResult = 0;
     bool isV_UNDEF = false;
+    if (ST->hasNEON()) {
       if (unsigned ShuffleOpc = isNEONTwoResultShuffleMask(
               ShuffleMask, VT, WhichResult, isV_UNDEF)) {
         if (isV_UNDEF)
           V2 = V1;
         return DAG.getNode(ShuffleOpc, dl, DAG.getVTList(VT, VT), V1, V2)
             .getValue(WhichResult);
       }
+    }
 
     // Also check for these shuffles through CONCAT_VECTORS: we canonicalize
     // shuffles that produce a result larger than their operands with:
     //   shuffle(concat(v1, undef), concat(v2, undef))
     // ->
     //   shuffle(concat(v1, v2), undef)
     // because we can access quad vectors (see PerformVECTOR_SHUFFLECombine).
     //
     // This is useful in the general case, but there are special cases where
     // native shuffles produce larger results: the two-result ops.
     //
     // Look through the concat when lowering them:
     //   shuffle(concat(v1, v2), undef)
     // ->
     //   concat(VZIP(v1, v2):0, :1)
     //
-    if (V1->getOpcode() == ISD::CONCAT_VECTORS && V2->isUndef()) {
+    if (ST->hasNEON() && V1->getOpcode() == ISD::CONCAT_VECTORS && V2->isUndef()) {
       SDValue SubV1 = V1->getOperand(0);
       SDValue SubV2 = V1->getOperand(1);
       EVT SubVT = SubV1.getValueType();
 
       // We expect these to have been canonicalized to -1.
       assert(llvm::all_of(ShuffleMask, [&](int i) {
         return i < (int)VT.getVectorNumElements();
       }) && "Unexpected shuffle index into UNDEF operand!");
if (NumElts == 4) {
     }
 
     // Compute the index in the perfect shuffle table.
     unsigned PFTableIndex =
       PFIndexes[0]*9*9*9+PFIndexes[1]*9*9+PFIndexes[2]*9+PFIndexes[3];
     unsigned PFEntry = PerfectShuffleTable[PFTableIndex];
     unsigned Cost = (PFEntry >> 30);
 
-    if (Cost <= 4)
+    if (Cost <= 4) {
+      if (ST->hasNEON())
+        return GeneratePerfectShuffle(PFEntry, V1, V2, DAG, dl);
+      else if (isLegalMveShuffleOp(PFEntry)) {
+        unsigned LHSID = (PFEntry >> 13) & ((1 << 13)-1);
+        unsigned RHSID = (PFEntry >>  0) & ((1 << 13)-1);
+        unsigned PFEntryLHS = PerfectShuffleTable[LHSID];
+        unsigned PFEntryRHS = PerfectShuffleTable[RHSID];
+        if (isLegalMveShuffleOp(PFEntryLHS) && isLegalMveShuffleOp(PFEntryRHS))
           return GeneratePerfectShuffle(PFEntry, V1, V2, DAG, dl);
       }
+    }
+  }
 
   // Implement shuffles with 32- or 64-bit elements as ARMISD::BUILD_VECTORs.
   if (EltSize >= 32) {
     // Do the expansion with floating-point types, since that is what the VFP
     // registers are defined to use, and since i64 is not legal.
     EVT EltVT = EVT::getFloatingPointVT(EltSize);
     EVT VecVT = EVT::getVectorVT(*DAG.getContext(), EltVT, NumElts);
     V1 = DAG.getNode(ISD::BITCAST, dl, VecVT, V1);
     V2 = DAG.getNode(ISD::BITCAST, dl, VecVT, V2);
     SmallVector<SDValue, 8> Ops;
     for (unsigned i = 0; i < NumElts; ++i) {
       if (ShuffleMask[i] < 0)
         Ops.push_back(DAG.getUNDEF(EltVT));
       else
         Ops.push_back(DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl, EltVT,
                                   ShuffleMask[i] < (int)NumElts ? V1 : V2,
                                   DAG.getConstant(ShuffleMask[i] & (NumElts-1),
                                                   dl, MVT::i32)));
     }
     SDValue Val = DAG.getNode(ARMISD::BUILD_VECTOR, dl, VecVT, Ops);
     return DAG.getNode(ISD::BITCAST, dl, VT, Val);
   }
 
-  if ((VT == MVT::v8i16 || VT == MVT::v16i8) && isReverseMask(ShuffleMask, VT))
+  if (ST->hasNEON() && (VT == MVT::v8i16 || VT == MVT::v16i8) && isReverseMask(ShuffleMask, VT))
     return LowerReverse_VECTOR_SHUFFLEv16i8_v8i16(Op, DAG);
 
-  if (VT == MVT::v8i8)
+  if (ST->hasNEON() && VT == MVT::v8i8)
     if (SDValue NewOp = LowerVECTOR_SHUFFLEv8i8(Op, ShuffleMask, DAG))
       return NewOp;
 
   return SDValue();
 }
 
 static SDValue LowerINSERT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG) {
   // INSERT_VECTOR_ELT is legal only for immediate indexes.
SDValue ARMTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::SRA_PARTS:     return LowerShiftRightParts(Op, DAG);
   case ISD::CTTZ:
   case ISD::CTTZ_ZERO_UNDEF: return LowerCTTZ(Op.getNode(), DAG, Subtarget);
   case ISD::CTPOP:         return LowerCTPOP(Op.getNode(), DAG, Subtarget);
   case ISD::SETCC:         return LowerVSETCC(Op, DAG);
   case ISD::SETCCCARRY:    return LowerSETCCCARRY(Op, DAG);
   case ISD::ConstantFP:    return LowerConstantFP(Op, DAG, Subtarget);
   case ISD::BUILD_VECTOR:  return LowerBUILD_VECTOR(Op, DAG, Subtarget);
-  case ISD::VECTOR_SHUFFLE: return LowerVECTOR_SHUFFLE(Op, DAG);
+  case ISD::VECTOR_SHUFFLE: return LowerVECTOR_SHUFFLE(Op, DAG, Subtarget);
   case ISD::INSERT_VECTOR_ELT: return LowerINSERT_VECTOR_ELT(Op, DAG);
   case ISD::EXTRACT_VECTOR_ELT: return LowerEXTRACT_VECTOR_ELT(Op, DAG);
   case ISD::CONCAT_VECTORS: return LowerCONCAT_VECTORS(Op, DAG);
   case ISD::FLT_ROUNDS_:   return LowerFLT_ROUNDS_(Op, DAG);
   case ISD::MUL:           return LowerMUL(Op, DAG);
   case ISD::SDIV:
     if (Subtarget->isTargetWindows() && !Op.getValueType().isVector())
       return LowerDIV_Windows(Op, DAG, /* Signed */ true);

llvm/lib/Target/ARM/ARMInstrInfo.td

 
 def ARMsmulwb       : SDNode<"ARMISD::SMULWB", SDTIntBinOp, []>;
 def ARMsmulwt       : SDNode<"ARMISD::SMULWT", SDTIntBinOp, []>;
 def ARMsmlalbb      : SDNode<"ARMISD::SMLALBB", SDT_LongMac, []>;
 def ARMsmlalbt      : SDNode<"ARMISD::SMLALBT", SDT_LongMac, []>;
 def ARMsmlaltb      : SDNode<"ARMISD::SMLALTB", SDT_LongMac, []>;
 def ARMsmlaltt      : SDNode<"ARMISD::SMLALTT", SDT_LongMac, []>;
 
+// Vector operations shared between NEON and MVE
+
+def ARMvdup      : SDNode<"ARMISD::VDUP", SDTypeProfile<1, 1, [SDTCisVec<0>]>>;
+
+// VDUPLANE can produce a quad-register result from a double-register source,
+// so the result is not constrained to match the source.
+def ARMvduplane  : SDNode<"ARMISD::VDUPLANE",
+                          SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>,
+                                               SDTCisVT<2, i32>]>>;
+
+def SDTARMVSHUF   : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0, 1>]>;
+def ARMvrev64    : SDNode<"ARMISD::VREV64", SDTARMVSHUF>;
+def ARMvrev32    : SDNode<"ARMISD::VREV32", SDTARMVSHUF>;
+def ARMvrev16    : SDNode<"ARMISD::VREV16", SDTARMVSHUF>;
+
+def SDTARMVGETLN  : SDTypeProfile<1, 2, [SDTCisVT<0, i32>, SDTCisInt<1>,
+                                         SDTCisVT<2, i32>]>;
+def ARMvgetlaneu : SDNode<"ARMISD::VGETLANEu", SDTARMVGETLN>;
+def ARMvgetlanes : SDNode<"ARMISD::VGETLANEs", SDTARMVGETLN>;
+
 //===----------------------------------------------------------------------===//
 // ARM Flag Definitions.
 
 class RegConstraint<string C> {
   string Constraints = C;
 }
 
 //===----------------------------------------------------------------------===//

llvm/lib/Target/ARM/ARMInstrMVE.td

 def VREV64_16 : t2VREV<"vrev64", "16", 0b01, 0b00>;
 def VREV64_32 : t2VREV<"vrev64", "32", 0b10, 0b00>;
 
 def VREV32_8  : t2VREV<"vrev32", "8", 0b00, 0b01>;
 def VREV32_16 : t2VREV<"vrev32", "16", 0b01, 0b01>;
 
 def VREV16_8  : t2VREV<"vrev16", "8", 0b00, 0b10>;
 
+let Predicates = [HasMVEInt] in {
+def : Pat<(v4i32 (ARMvrev64 (v4i32 MQPR:$src))),
+          (v4i32 (VREV64_32 (v4i32 MQPR:$src)))>;
+def : Pat<(v8i16 (ARMvrev64 (v8i16 MQPR:$src))),
+          (v8i16 (VREV64_16 (v8i16 MQPR:$src)))>;
+def : Pat<(v16i8 (ARMvrev64 (v16i8 MQPR:$src))),
+          (v16i8 (VREV64_8  (v16i8 MQPR:$src)))>;
+
+def : Pat<(v8i16 (ARMvrev32 (v8i16 MQPR:$src))),
+          (v8i16 (VREV32_16 (v8i16 MQPR:$src)))>;
+def : Pat<(v16i8 (ARMvrev32 (v16i8 MQPR:$src))),
+          (v16i8 (VREV32_8  (v16i8 MQPR:$src)))>;
+
+def : Pat<(v16i8 (ARMvrev16 (v16i8 MQPR:$src))),
+          (v16i8 (VREV16_8  (v16i8 MQPR:$src)))>;
+}
+
+let Predicates = [HasMVEFloat] in {
+def : Pat<(v4f32 (ARMvrev64 (v4f32 MQPR:$src))),
+          (v4f32 (VREV64_32 (v4f32 MQPR:$src)))>;
+def : Pat<(v8f16 (ARMvrev64 (v8f16 MQPR:$src))),
+          (v8f16 (VREV64_16 (v8f16 MQPR:$src)))>;
+def : Pat<(v8f16 (ARMvrev32 (v8f16 MQPR:$src))),
+          (v8f16 (VREV32_16 (v8f16 MQPR:$src)))>;
+}
+
 def VMVN : MVE_bit_arith<(outs MQPR:$Qd), (ins MQPR:$Qm),
                          "vmvn", "", "$Qd, $Qm", ""> {
   let Inst{28} = 0b1;
   let Inst{25-23} = 0b111;
   let Inst{21-16} = 0b110000;
   let Inst{12-6} = 0b0010111;
   let Inst{4} = 0b0;
   let Inst{0} = 0b0;
def VMOVt116 : VMOVt1regtovec<"16", 0b0, (ins MVEVectorIndex16:$Idx)> {
   let Inst{5} = 0b1;
 }
 def VMOVt132 : VMOVt1regtovec<"32", 0b0, (ins MVEVectorIndex32:$Idx)> {
   let Inst{22} = 0b0;
   let Inst{6-5} = 0b00;
   let Predicates = [HasFPRegsV8_1M];
 }
 
+let Predicates = [HasMVEInt] in {
+  def : Pat<(extractelt (v4i32 MQPR:$src), imm:$lane),
+            (COPY_TO_REGCLASS
+              (i32 (EXTRACT_SUBREG MQPR:$src, (SSubReg_f32_reg imm:$lane))), rGPR)>;
+  def : Pat<(insertelt (v4i32 MQPR:$src1), rGPR:$src2, imm:$lane),
+            (VMOVt132 MQPR:$src1, rGPR:$src2, imm:$lane)>;
+
+  def : Pat<(vector_insert (v16i8 MQPR:$src1), rGPR:$src2, imm:$lane),
+            (VMOVt18  MQPR:$src1, rGPR:$src2, imm:$lane)>;
+  def : Pat<(vector_insert (v8i16 MQPR:$src1), rGPR:$src2, imm:$lane),
+            (VMOVt116 MQPR:$src1, rGPR:$src2, imm:$lane)>;
+
+  def : Pat<(ARMvgetlanes (v16i8 MQPR:$src), imm:$lane),
+            (VMOVt1s8 MQPR:$src, imm:$lane)>;
+  def : Pat<(ARMvgetlanes (v8i16 MQPR:$src), imm:$lane),
+            (VMOVt1s16 MQPR:$src, imm:$lane)>;
+  def : Pat<(ARMvgetlaneu (v16i8 MQPR:$src), imm:$lane),
+            (VMOVt1u8 MQPR:$src, imm:$lane)>;
+  def : Pat<(ARMvgetlaneu (v8i16 MQPR:$src), imm:$lane),
+            (VMOVt1u16 MQPR:$src, imm:$lane)>;
+
+  def : Pat<(v16i8 (scalar_to_vector GPR:$src)),
+            (VMOVt18  (v16i8 (IMPLICIT_DEF)), rGPR:$src, (i32 0))>;
+  def : Pat<(v8i16 (scalar_to_vector GPR:$src)),
+            (VMOVt116 (v8i16 (IMPLICIT_DEF)), rGPR:$src, (i32 0))>;
+  def : Pat<(v4i32 (scalar_to_vector GPR:$src)),
+            (VMOVt132 (v4i32 (IMPLICIT_DEF)), rGPR:$src, (i32 0))>;
+}
+
+let Predicates = [HasMVEFloat] in {
+  def : Pat<(extractelt (v4f32 MQPR:$src), imm:$lane),
+            (COPY_TO_REGCLASS
+            (f32 (EXTRACT_SUBREG MQPR:$src, (SSubReg_f32_reg imm:$lane))), SPR)>;
+  def : Pat<(v4f32 (insertelt (v4f32 MQPR:$src1), (f32 SPR:$src2), imm:$lane)),
+            (INSERT_SUBREG (v4f32 (COPY_TO_REGCLASS MQPR:$src1, MQPR)), SPR:$src2,
+            (SSubReg_f32_reg imm:$lane))>;
+
+  def : Pat<(insertelt (v8f16 MQPR:$src1), HPR:$src2, imm:$lane),
+            (VMOVt116 MQPR:$src1, (COPY_TO_REGCLASS HPR:$src2, rGPR), imm:$lane)>;
+  def : Pat<(extractelt (v8f16 MQPR:$src), imm:$lane),
+            (COPY_TO_REGCLASS (VMOVt1u16 MQPR:$src, imm:$lane), HPR)>;
+
+  def : Pat<(v4f32 (scalar_to_vector SPR:$src)),
+            (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), SPR:$src, ssub_0)>;
+  def : Pat<(v8f16 (scalar_to_vector HPR:$src)),
+            (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), HPR:$src, ssub_0)>;
+}
+
 // end of mve_bit instructions
 
 // start of MVE Integer instructions
 
 class MVE_int<string iname, string suffix, bits<2> size, list<dag> pattern=[]>
   : MVE_p<(outs MQPR:$Qd), (ins MQPR:$Qn, MQPR:$Qm), NoItinerary,
           iname, suffix, "$Qd, $Qn, $Qm", vpred_r, "", pattern> {
   bits<4> Qd;
class VDUPt1<string suffix, bit B, bit E, list<dag> pattern=[]>
   let Inst{5} = E;
   let Inst{4-0} = 0b10000;
 }
 
 def VDUPt132 : VDUPt1<"32", 0b0, 0b0>;
 def VDUPt116 : VDUPt1<"16", 0b0, 0b1>;
 def VDUPt18  : VDUPt1<"8", 0b1, 0b0>;
 
+let Predicates = [HasMVEInt] in {
+  def : Pat<(v16i8 (ARMvdup (i32 rGPR:$elem))),
+            (VDUPt18  rGPR:$elem)>;
+  def : Pat<(v8i16 (ARMvdup (i32 rGPR:$elem))),
+            (VDUPt116 rGPR:$elem)>;
+  def : Pat<(v4i32 (ARMvdup (i32 rGPR:$elem))),
+            (VDUPt132 rGPR:$elem)>;
+
+  def : Pat<(v4i32 (ARMvduplane (v4i32 MQPR:$src), imm:$lane)),
+            (VDUPt132 (VMOVt1vtor32 MQPR:$src, imm:$lane))>;
+  // For the 16-bit and 8-bit vduplanes we don't care about the signedness
+  // of the lane move operation as we only want the lowest 8/16 bits anyway.
+  def : Pat<(v8i16 (ARMvduplane (v8i16 MQPR:$src), imm:$lane)),
+            (VDUPt116 (VMOVt1u16 MQPR:$src, imm:$lane))>;
+  def : Pat<(v16i8 (ARMvduplane (v16i8 MQPR:$src), imm:$lane)),
+            (VDUPt18  (VMOVt1u8 MQPR:$src, imm:$lane))>;
+}
+
+let Predicates = [HasMVEFloat] in {
+  def : Pat<(v4f32 (ARMvdup (f32 SPR:$elem))),
+            (v4f32 (VDUPt132 (i32 (COPY_TO_REGCLASS (f32 SPR:$elem),
+                                                    rGPR))))>;
+  def : Pat<(v8f16 (ARMvdup (f16 HPR:$elem))),
+            (v8f16 (VDUPt116 (i32 (COPY_TO_REGCLASS (f16 HPR:$elem),
+                                                    rGPR))))>;
+
+  def : Pat<(v4f32 (ARMvduplane (v4f32 MQPR:$src), imm:$lane)),
+            (VDUPt132 (VMOVt1vtor32 MQPR:$src, imm:$lane))>;
+  def : Pat<(v8f16 (ARMvduplane (v8f16 MQPR:$src), imm:$lane)),
+            (VDUPt116 (VMOVt1u16 MQPR:$src, imm:$lane))>;
+}
+
+
 class MVEIntSingleSrc<string iname, string suffix, bits<2> size,
                          list<dag> pattern=[]>
   : MVE_p<(outs MQPR:$Qd), (ins MQPR:$Qm), NoItinerary,
           iname, suffix, "$Qd, $Qm", vpred_r, "", pattern> {
   bits<4> Qd;
   bits<4> Qm;
 
   let Inst{22} = Qd{3};

llvm/lib/Target/ARM/ARMInstrNEON.td

 
 def NEONvqrshrns  : SDNode<"ARMISD::VQRSHRNs", SDTARMVSHX>;
 def NEONvqrshrnu  : SDNode<"ARMISD::VQRSHRNu", SDTARMVSHX>;
 def NEONvqrshrnsu : SDNode<"ARMISD::VQRSHRNsu", SDTARMVSHX>;
 
 def NEONvsli      : SDNode<"ARMISD::VSLI", SDTARMVSHINS>;
 def NEONvsri      : SDNode<"ARMISD::VSRI", SDTARMVSHINS>;
 
-def SDTARMVGETLN  : SDTypeProfile<1, 2, [SDTCisVT<0, i32>, SDTCisInt<1>,
-                                         SDTCisVT<2, i32>]>;
-def NEONvgetlaneu : SDNode<"ARMISD::VGETLANEu", SDTARMVGETLN>;
-def NEONvgetlanes : SDNode<"ARMISD::VGETLANEs", SDTARMVGETLN>;
-
 def SDTARMVMOVIMM : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisVT<1, i32>]>;
 def NEONvmovImm   : SDNode<"ARMISD::VMOVIMM", SDTARMVMOVIMM>;
 def NEONvmvnImm   : SDNode<"ARMISD::VMVNIMM", SDTARMVMOVIMM>;
 def NEONvmovFPImm : SDNode<"ARMISD::VMOVFPIMM", SDTARMVMOVIMM>;
 
 def SDTARMVORRIMM : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0, 1>,
                                            SDTCisVT<2, i32>]>;
 def NEONvorrImm   : SDNode<"ARMISD::VORRIMM", SDTARMVORRIMM>;
 def NEONvbicImm   : SDNode<"ARMISD::VBICIMM", SDTARMVORRIMM>;
 
 def NEONvbsl      : SDNode<"ARMISD::VBSL",
                            SDTypeProfile<1, 3, [SDTCisVec<0>,
                                                 SDTCisSameAs<0, 1>,
                                                 SDTCisSameAs<0, 2>,
                                                 SDTCisSameAs<0, 3>]>>;
 
-def NEONvdup      : SDNode<"ARMISD::VDUP", SDTypeProfile<1, 1, [SDTCisVec<0>]>>;
-
-// VDUPLANE can produce a quad-register result from a double-register source,
-// so the result is not constrained to match the source.
-def NEONvduplane  : SDNode<"ARMISD::VDUPLANE",
-                           SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisVec<1>,
-                                                SDTCisVT<2, i32>]>>;
-
 def SDTARMVEXT    : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0, 1>,
                                          SDTCisSameAs<0, 2>, SDTCisVT<3, i32>]>;
 def NEONvext      : SDNode<"ARMISD::VEXT", SDTARMVEXT>;
 
-def SDTARMVSHUF   : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0, 1>]>;
-def NEONvrev64    : SDNode<"ARMISD::VREV64", SDTARMVSHUF>;
-def NEONvrev32    : SDNode<"ARMISD::VREV32", SDTARMVSHUF>;
-def NEONvrev16    : SDNode<"ARMISD::VREV16", SDTARMVSHUF>;
-
 def SDTARMVSHUF2  : SDTypeProfile<2, 2, [SDTCisVec<0>, SDTCisSameAs<0, 1>,
                                          SDTCisSameAs<0, 2>,
                                          SDTCisSameAs<0, 3>]>;
 def NEONzip       : SDNode<"ARMISD::VZIP", SDTARMVSHUF2>;
 def NEONuzp       : SDNode<"ARMISD::VUZP", SDTARMVSHUF2>;
 def NEONtrn       : SDNode<"ARMISD::VTRN", SDTARMVSHUF2>;
 
 def SDTARMVMULL   : SDTypeProfile<1, 2, [SDTCisInt<0>, SDTCisInt<1>,

 
 //   VLD1DUP  : Vector Load (single element to all lanes)
 class VLD1DUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp,
               Operand AddrMode>
   : NLdSt<1, 0b10, 0b1100, op7_4, (outs VecListOneDAllLanes:$Vd),
           (ins AddrMode:$Rn),
           IIC_VLD1dup, "vld1", Dt, "$Vd, $Rn", "",
           [(set VecListOneDAllLanes:$Vd,
-                (Ty (NEONvdup (i32 (LoadOp AddrMode:$Rn)))))]>,
+                (Ty (ARMvdup (i32 (LoadOp AddrMode:$Rn)))))]>,
    Sched<[WriteVLD2]> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLD1DupInstruction";
 }
 def VLD1DUPd8  : VLD1DUP<{0,0,0,?}, "8", v8i8, extloadi8,
                          addrmode6dupalignNone>;
 def VLD1DUPd16 : VLD1DUP<{0,1,0,?}, "16", v4i16, extloadi16,
                          addrmode6dupalign16>;
 def VLD1DUPd32 : VLD1DUP<{1,0,0,?}, "32", v2i32, load,
                          addrmode6dupalign32>;
 
 let Predicates = [HasNEON] in {
-def : Pat<(v2f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
+def : Pat<(v2f32 (ARMvdup (f32 (load addrmode6dup:$addr)))),
           (VLD1DUPd32 addrmode6:$addr)>;
 }
 
 class VLD1QDUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp,
                Operand AddrMode>
   : NLdSt<1, 0b10, 0b1100, op7_4, (outs VecListDPairAllLanes:$Vd),
           (ins AddrMode:$Rn), IIC_VLD1dup,
           "vld1", Dt, "$Vd, $Rn", "",
           [(set VecListDPairAllLanes:$Vd,
-                (Ty (NEONvdup (i32 (LoadOp AddrMode:$Rn)))))]> {
+                (Ty (ARMvdup (i32 (LoadOp AddrMode:$Rn)))))]> {
   let Rm = 0b1111;
   let Inst{4} = Rn{4};
   let DecoderMethod = "DecodeVLD1DupInstruction";
 }
 
 def VLD1DUPq8  : VLD1QDUP<{0,0,1,0}, "8", v16i8, extloadi8,
                           addrmode6dupalignNone>;
 def VLD1DUPq16 : VLD1QDUP<{0,1,1,?}, "16", v8i16, extloadi16,
                           addrmode6dupalign16>;
 def VLD1DUPq32 : VLD1QDUP<{1,0,1,?}, "32", v4i32, load,
                           addrmode6dupalign32>;
 
 let Predicates = [HasNEON] in {
-def : Pat<(v4f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
+def : Pat<(v4f32 (ARMvdup (f32 (load addrmode6dup:$addr)))),
           (VLD1DUPq32 addrmode6:$addr)>;
 }
 
 let mayLoad = 1, hasSideEffects = 0, hasExtraDefRegAllocReq = 1 in {
 // ...with address register writeback:
 multiclass VLD1DUPWB<bits<4> op7_4, string Dt, Operand AddrMode> {
   def _fixed : NLdSt<1, 0b10, 0b1100, op7_4,
                      (outs VecListOneDAllLanes:$Vd, GPR:$wb),

 }
 class VST1QLNPseudo<ValueType Ty, PatFrag StoreOp, SDNode ExtractOp>
   : VSTQLNPseudo<IIC_VST1ln>, Sched<[WriteVST1]> {
   let Pattern = [(StoreOp (ExtractOp (Ty QPR:$src), imm:$lane),
                           addrmode6:$addr)];
 }
 
 def VST1LNd8  : VST1LN<0b0000, {?,?,?,0}, "8", v8i8, truncstorei8,
-                       NEONvgetlaneu, addrmode6> {
+                       ARMvgetlaneu, addrmode6> {
   let Inst{7-5} = lane{2-0};
 }
 def VST1LNd16 : VST1LN<0b0100, {?,?,0,?}, "16", v4i16, truncstorei16,
-                       NEONvgetlaneu, addrmode6> {
+                       ARMvgetlaneu, addrmode6> {
   let Inst{7-6} = lane{1-0};
   let Inst{4}   = Rn{4};
 }
 
 def VST1LNd32 : VST1LN<0b1000, {?,0,?,?}, "32", v2i32, store, extractelt,
                        addrmode6oneL32> {
   let Inst{7}   = lane{0};
   let Inst{5-4} = Rn{5-4};
 }
 
-def VST1LNq8Pseudo  : VST1QLNPseudo<v16i8, truncstorei8, NEONvgetlaneu>;
-def VST1LNq16Pseudo : VST1QLNPseudo<v8i16, truncstorei16, NEONvgetlaneu>;
+def VST1LNq8Pseudo  : VST1QLNPseudo<v16i8, truncstorei8, ARMvgetlaneu>;
+def VST1LNq16Pseudo : VST1QLNPseudo<v8i16, truncstorei16, ARMvgetlaneu>;
 def VST1LNq32Pseudo : VST1QLNPseudo<v4i32, store, extractelt>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(store (extractelt (v2f32 DPR:$src), imm:$lane), addrmode6:$addr),
           (VST1LNd32 addrmode6:$addr, DPR:$src, imm:$lane)>;
 def : Pat<(store (extractelt (v4f32 QPR:$src), imm:$lane), addrmode6:$addr),
           (VST1LNq32Pseudo addrmode6:$addr, QPR:$src, imm:$lane)>;
 

 }
 class VST1QLNWBPseudo<ValueType Ty, PatFrag StoreOp, SDNode ExtractOp>
   : VSTQLNWBPseudo<IIC_VST1lnu>, Sched<[WriteVST1]> {
   let Pattern = [(set GPR:$wb, (StoreOp (ExtractOp (Ty QPR:$src), imm:$lane),
                                         addrmode6:$addr, am6offset:$offset))];
 }
 
 def VST1LNd8_UPD  : VST1LNWB<0b0000, {?,?,?,0}, "8", v8i8, post_truncsti8,
-                             NEONvgetlaneu, addrmode6> {
+                             ARMvgetlaneu, addrmode6> {
   let Inst{7-5} = lane{2-0};
 }
 def VST1LNd16_UPD : VST1LNWB<0b0100, {?,?,0,?}, "16", v4i16, post_truncsti16,
-                             NEONvgetlaneu, addrmode6> {
+                             ARMvgetlaneu, addrmode6> {
   let Inst{7-6} = lane{1-0};
   let Inst{4}   = Rn{4};
 }
 def VST1LNd32_UPD : VST1LNWB<0b1000, {?,0,?,?}, "32", v2i32, post_store,
                              extractelt, addrmode6oneL32> {
   let Inst{7}   = lane{0};
   let Inst{5-4} = Rn{5-4};
 }
 
-def VST1LNq8Pseudo_UPD  : VST1QLNWBPseudo<v16i8, post_truncsti8, NEONvgetlaneu>;
-def VST1LNq16Pseudo_UPD : VST1QLNWBPseudo<v8i16, post_truncsti16,NEONvgetlaneu>;
+def VST1LNq8Pseudo_UPD  : VST1QLNWBPseudo<v16i8, post_truncsti8, ARMvgetlaneu>;
+def VST1LNq16Pseudo_UPD : VST1QLNWBPseudo<v8i16, post_truncsti16,ARMvgetlaneu>;
 def VST1LNq32Pseudo_UPD : VST1QLNWBPseudo<v4i32, post_store, extractelt>;
 
 let mayStore = 1, hasSideEffects = 0, hasExtraSrcRegAllocReq = 1 in {
 
 //   VST2LN   : Vector Store (single 2-element structure from one lane)
 class VST2LN<bits<4> op11_8, bits<4> op7_4, string Dt>
   : NLdStLn<1, 0b00, op11_8, op7_4, (outs),
           (ins addrmode6:$Rn, DPR:$Vd, DPR:$src2, nohash_imm:$lane),

 class N3VDSL<bits<2> op21_20, bits<4> op11_8,
              InstrItinClass itin, string OpcodeStr, string Dt,
              ValueType Ty, SDNode ShOp>
   : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (Ty DPR:$Vd),
               (Ty (ShOp (Ty DPR:$Vn),
-                        (Ty (NEONvduplane (Ty DPR_VFP2:$Vm),imm:$lane)))))]> {
+                        (Ty (ARMvduplane (Ty DPR_VFP2:$Vm),imm:$lane)))))]> {
   // All of these have a two-operand InstAlias.
   let TwoOperandAliasConstraint = "$Vn = $Vd";
   let isCommutable = 0;
 }
 class N3VDSL16<bits<2> op21_20, bits<4> op11_8,
                string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp>
   : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, IIC_VMULi16D, OpcodeStr, Dt,"$Vd, $Vn, $Vm$lane","",
         [(set (Ty DPR:$Vd),
               (Ty (ShOp (Ty DPR:$Vn),
-                        (Ty (NEONvduplane (Ty DPR_8:$Vm), imm:$lane)))))]> {
+                        (Ty (ARMvduplane (Ty DPR_8:$Vm), imm:$lane)))))]> {
   // All of these have a two-operand InstAlias.
   let TwoOperandAliasConstraint = "$Vn = $Vd";
   let isCommutable = 0;
 }
 
 class N3VQ<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
            InstrItinClass itin, string OpcodeStr, string Dt,
            ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable>

 class N3VQSL<bits<2> op21_20, bits<4> op11_8,
              InstrItinClass itin, string OpcodeStr, string Dt,
              ValueType ResTy, ValueType OpTy, SDNode ShOp>
   : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins QPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (ShOp (ResTy QPR:$Vn),
-                           (ResTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
+                           (ResTy (ARMvduplane (OpTy DPR_VFP2:$Vm),
                                                 imm:$lane)))))]> {
   // All of these have a two-operand InstAlias.
   let TwoOperandAliasConstraint = "$Vn = $Vd";
   let isCommutable = 0;
 }
 class N3VQSL16<bits<2> op21_20, bits<4> op11_8, string OpcodeStr, string Dt,
                ValueType ResTy, ValueType OpTy, SDNode ShOp>
   : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins QPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, IIC_VMULi16Q, OpcodeStr, Dt,"$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (ShOp (ResTy QPR:$Vn),
-                           (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
+                           (ResTy (ARMvduplane (OpTy DPR_8:$Vm),
                                                 imm:$lane)))))]> {
   // All of these have a two-operand InstAlias.
   let TwoOperandAliasConstraint = "$Vn = $Vd";
   let isCommutable = 0;
 }
 
 // Basic 3-register intrinsics, both double- and quad-register.
 class N3VDInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,

 
 class N3VDIntSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                 string OpcodeStr, string Dt, ValueType Ty, SDPatternOperator IntOp>
   : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (Ty DPR:$Vd),
               (Ty (IntOp (Ty DPR:$Vn),
-                         (Ty (NEONvduplane (Ty DPR_VFP2:$Vm),
+                         (Ty (ARMvduplane (Ty DPR_VFP2:$Vm),
                                            imm:$lane)))))]> {
   let isCommutable = 0;
 }
 
 class N3VDIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                   string OpcodeStr, string Dt, ValueType Ty, SDPatternOperator IntOp>
   : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (Ty DPR:$Vd),
               (Ty (IntOp (Ty DPR:$Vn),
-                         (Ty (NEONvduplane (Ty DPR_8:$Vm), imm:$lane)))))]> {
+                         (Ty (ARMvduplane (Ty DPR_8:$Vm), imm:$lane)))))]> {
   let isCommutable = 0;
 }
 class N3VDIntSh<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
               Format f, InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
         (outs DPR:$Vd), (ins DPR:$Vm, DPR:$Vn), f, itin,
         OpcodeStr, Dt, "$Vd, $Vm, $Vn", "",

 class N3VQIntSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                 string OpcodeStr, string Dt,
                 ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins QPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (ResTy QPR:$Vn),
-                            (ResTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
+                            (ResTy (ARMvduplane (OpTy DPR_VFP2:$Vm),
                                                  imm:$lane)))))]> {
   let isCommutable = 0;
 }
 class N3VQIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                   string OpcodeStr, string Dt,
                   ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins QPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (ResTy QPR:$Vn),
-                            (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
+                            (ResTy (ARMvduplane (OpTy DPR_8:$Vm),
                                                  imm:$lane)))))]> {
   let isCommutable = 0;
 }
 class N3VQIntSh<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
               Format f, InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3V<op24, op23, op21_20, op11_8, 1, op4,
         (outs QPR:$Vd), (ins QPR:$Vm, QPR:$Vn), f, itin,
class N3VDMulOpSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
   : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd),
         (ins DPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (Ty DPR:$Vd),
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (MulOp DPR:$Vn,
-                                   (Ty (NEONvduplane (Ty DPR_VFP2:$Vm),
+                                   (Ty (ARMvduplane (Ty DPR_VFP2:$Vm),
                                                      imm:$lane)))))))]>;
 class N3VDMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                     string OpcodeStr, string Dt,
                     ValueType Ty, SDPatternOperator MulOp, SDPatternOperator ShOp>
   : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
         (outs DPR:$Vd),
         (ins DPR:$src1, DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (Ty DPR:$Vd),
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (MulOp DPR:$Vn,
-                                   (Ty (NEONvduplane (Ty DPR_8:$Vm),
+                                   (Ty (ARMvduplane (Ty DPR_8:$Vm),
                                                      imm:$lane)))))))]>;
 
 class N3VQMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
                 InstrItinClass itin, string OpcodeStr, string Dt, ValueType Ty,
                 SDPatternOperator MulOp, SDPatternOperator OpNode>
   : N3V<op24, op23, op21_20, op11_8, 1, op4,
         (outs QPR:$Vd), (ins QPR:$src1, QPR:$Vn, QPR:$Vm), N3RegFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm", "$src1 = $Vd",
         [(set QPR:$Vd, (Ty (OpNode QPR:$src1,
                              (Ty (MulOp QPR:$Vn, QPR:$Vm)))))]>;
 class N3VQMulOpSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                   string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy,
                   SDPatternOperator MulOp, SDPatternOperator ShOp>
   : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd),
         (ins QPR:$src1, QPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (ResTy QPR:$Vd),
               (ResTy (ShOp (ResTy QPR:$src1),
                            (ResTy (MulOp QPR:$Vn,
-                                   (ResTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
+                                   (ResTy (ARMvduplane (OpTy DPR_VFP2:$Vm),
                                                         imm:$lane)))))))]>;
 class N3VQMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                     string OpcodeStr, string Dt,
                     ValueType ResTy, ValueType OpTy,
                     SDPatternOperator MulOp, SDPatternOperator ShOp>
   : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd),
         (ins QPR:$src1, QPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (ResTy QPR:$Vd),
               (ResTy (ShOp (ResTy QPR:$src1),
                            (ResTy (MulOp QPR:$Vn,
-                                   (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
+                                   (ResTy (ARMvduplane (OpTy DPR_8:$Vm),
                                                         imm:$lane)))))))]>;
 
 // Neon Intrinsic-Op instructions (VABA): double- and quad-register.
 class N3VDIntOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
                 InstrItinClass itin, string OpcodeStr, string Dt,
                 ValueType Ty, SDPatternOperator IntOp, SDNode OpNode>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
         (outs DPR:$Vd), (ins DPR:$src1, DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
class N3VLMulOpSL<bit op24, bits<2> op21_20, bits<4> op11_8,
                   ValueType TyQ, ValueType TyD, SDNode MulOp, SDNode OpNode>
   : N3VLane32<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$Vd),
         (ins QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set QPR:$Vd,
           (OpNode (TyQ QPR:$src1),
                   (TyQ (MulOp (TyD DPR:$Vn),
-                              (TyD (NEONvduplane (TyD DPR_VFP2:$Vm),
+                              (TyD (ARMvduplane (TyD DPR_VFP2:$Vm),
                                                  imm:$lane))))))]>;
 class N3VLMulOpSL16<bit op24, bits<2> op21_20, bits<4> op11_8,
                     InstrItinClass itin, string OpcodeStr, string Dt,
                     ValueType TyQ, ValueType TyD, SDNode MulOp, SDNode OpNode>
   : N3VLane16<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$Vd),
         (ins QPR:$src1, DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set QPR:$Vd,
           (OpNode (TyQ QPR:$src1),
                   (TyQ (MulOp (TyD DPR:$Vn),
-                              (TyD (NEONvduplane (TyD DPR_8:$Vm),
+                              (TyD (ARMvduplane (TyD DPR_8:$Vm),
                                                  imm:$lane))))))]>;
 
 // Long Intrinsic-Op vector operations with explicit extend (VABAL).
 class N3VLIntExtOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
                    InstrItinClass itin, string OpcodeStr, string Dt,
                    ValueType TyQ, ValueType TyD, SDPatternOperator IntOp, SDNode ExtOp,
                    SDNode OpNode>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
class N3VLInt3SL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
   : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd),
         (ins QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (ResTy QPR:$src1),
                             (OpTy DPR:$Vn),
-                            (OpTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
+                            (OpTy (ARMvduplane (OpTy DPR_VFP2:$Vm),
                                                 imm:$lane)))))]>;
 class N3VLInt3SL16<bit op24, bits<2> op21_20, bits<4> op11_8,
                    InstrItinClass itin, string OpcodeStr, string Dt,
                    ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd),
         (ins QPR:$src1, DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin,
         OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "$src1 = $Vd",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (ResTy QPR:$src1),
                             (OpTy DPR:$Vn),
-                            (OpTy (NEONvduplane (OpTy DPR_8:$Vm),
+                            (OpTy (ARMvduplane (OpTy DPR_8:$Vm),
                                                 imm:$lane)))))]>;
 
 // Narrowing 3-register intrinsics.
 class N3VNInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
               string OpcodeStr, string Dt, ValueType TyD, ValueType TyQ,
               SDPatternOperator IntOp, bit Commutable>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
         (outs DPR:$Vd), (ins QPR:$Vn, QPR:$Vm), N3RegFrm, IIC_VBINi4D,

 class N3VLSL<bit op24, bits<2> op21_20, bits<4> op11_8,
              InstrItinClass itin, string OpcodeStr, string Dt,
              ValueType TyQ, ValueType TyD, SDNode OpNode>
   : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set QPR:$Vd,
           (TyQ (OpNode (TyD DPR:$Vn),
-                       (TyD (NEONvduplane (TyD DPR_VFP2:$Vm),imm:$lane)))))]>;
+                       (TyD (ARMvduplane (TyD DPR_VFP2:$Vm),imm:$lane)))))]>;
 class N3VLSL16<bit op24, bits<2> op21_20, bits<4> op11_8,
                InstrItinClass itin, string OpcodeStr, string Dt,
                ValueType TyQ, ValueType TyD, SDNode OpNode>
   : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set QPR:$Vd,
           (TyQ (OpNode (TyD DPR:$Vn),
-                       (TyD (NEONvduplane (TyD DPR_8:$Vm), imm:$lane)))))]>;
+                       (TyD (ARMvduplane (TyD DPR_8:$Vm), imm:$lane)))))]>;
 
 // Long 3-register operations with explicitly extended operands.
 class N3VLExt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
               InstrItinClass itin, string OpcodeStr, string Dt,
               ValueType TyQ, ValueType TyD, SDNode OpNode, SDNode ExtOp,
               bit Commutable>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
         (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin,

 class N3VLIntSL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                 string OpcodeStr, string Dt,
                 ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, VectorIndex32:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (OpTy DPR:$Vn),
-                            (OpTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
+                            (OpTy (ARMvduplane (OpTy DPR_VFP2:$Vm),
                                                 imm:$lane)))))]>;
 class N3VLIntSL16<bit op24, bits<2> op21_20, bits<4> op11_8,
                   InstrItinClass itin, string OpcodeStr, string Dt,
                   ValueType ResTy, ValueType OpTy, SDPatternOperator IntOp>
   : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
         (outs QPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, VectorIndex16:$lane),
         NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm$lane", "",
         [(set (ResTy QPR:$Vd),
               (ResTy (IntOp (OpTy DPR:$Vn),
-                            (OpTy (NEONvduplane (OpTy DPR_8:$Vm),
+                            (OpTy (ARMvduplane (OpTy DPR_8:$Vm),
                                                 imm:$lane)))))]>;
 
 // Wide 3-register operations.
 class N3VW<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
            string OpcodeStr, string Dt, ValueType TyQ, ValueType TyD,
            SDNode OpNode, SDNode ExtOp, bit Commutable>
   : N3V<op24, op23, op21_20, op11_8, 0, op4,
         (outs QPR:$Vd), (ins QPR:$Vn, DPR:$Vm), N3RegFrm, IIC_VSUBiD,

 def  VMULslhd : N3VDSL16<0b01, 0b1001, "vmul", "f16", v4f16, fmul>,
                 Requires<[HasNEON,HasFullFP16]>;
 def  VMULslhq : N3VQSL16<0b01, 0b1001, "vmul", "f16", v8f16,
                        v4f16, fmul>,
                 Requires<[HasNEON,HasFullFP16]>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v8i16 (mul (v8i16 QPR:$src1),
-                      (v8i16 (NEONvduplane (v8i16 QPR:$src2), imm:$lane)))),
+                      (v8i16 (ARMvduplane (v8i16 QPR:$src2), imm:$lane)))),
           (v8i16 (VMULslv8i16 (v8i16 QPR:$src1),
                               (v4i16 (EXTRACT_SUBREG QPR:$src2,
                                       (DSubReg_i16_reg imm:$lane))),
                               (SubReg_i16_lane imm:$lane)))>;
 def : Pat<(v4i32 (mul (v4i32 QPR:$src1),
-                      (v4i32 (NEONvduplane (v4i32 QPR:$src2), imm:$lane)))),
+                      (v4i32 (ARMvduplane (v4i32 QPR:$src2), imm:$lane)))),
           (v4i32 (VMULslv4i32 (v4i32 QPR:$src1),
                               (v2i32 (EXTRACT_SUBREG QPR:$src2,
                                       (DSubReg_i32_reg imm:$lane))),
                               (SubReg_i32_lane imm:$lane)))>;
 def : Pat<(v4f32 (fmul (v4f32 QPR:$src1),
-                       (v4f32 (NEONvduplane (v4f32 QPR:$src2), imm:$lane)))),
+                       (v4f32 (ARMvduplane (v4f32 QPR:$src2), imm:$lane)))),
           (v4f32 (VMULslfq (v4f32 QPR:$src1),
                            (v2f32 (EXTRACT_SUBREG QPR:$src2,
                                    (DSubReg_i32_reg imm:$lane))),
                            (SubReg_i32_lane imm:$lane)))>;
 def : Pat<(v8f16 (fmul (v8f16 QPR:$src1),
-                       (v8f16 (NEONvduplane (v8f16 QPR:$src2), imm:$lane)))),
+                       (v8f16 (ARMvduplane (v8f16 QPR:$src2), imm:$lane)))),
           (v8f16 (VMULslhq(v8f16 QPR:$src1),
                            (v4f16 (EXTRACT_SUBREG QPR:$src2,
                                    (DSubReg_i16_reg imm:$lane))),
                            (SubReg_i16_lane imm:$lane)))>;
 
-def : Pat<(v2f32 (fmul DPR:$Rn, (NEONvdup (f32 SPR:$Rm)))),
+def : Pat<(v2f32 (fmul DPR:$Rn, (ARMvdup (f32 SPR:$Rm)))),
           (VMULslfd DPR:$Rn,
             (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$Rm, ssub_0),
             (i32 0))>;
-def : Pat<(v4f16 (fmul DPR:$Rn, (NEONvdup (f16 HPR:$Rm)))),
+def : Pat<(v4f16 (fmul DPR:$Rn, (ARMvdup (f16 HPR:$Rm)))),
           (VMULslhd DPR:$Rn,
             (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)), HPR:$Rm, ssub_0),
             (i32 0))>;
-def : Pat<(v4f32 (fmul QPR:$Rn, (NEONvdup (f32 SPR:$Rm)))),
+def : Pat<(v4f32 (fmul QPR:$Rn, (ARMvdup (f32 SPR:$Rm)))),
           (VMULslfq QPR:$Rn,
             (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), SPR:$Rm, ssub_0),
             (i32 0))>;
-def : Pat<(v8f16 (fmul QPR:$Rn, (NEONvdup (f16 HPR:$Rm)))),
+def : Pat<(v8f16 (fmul QPR:$Rn, (ARMvdup (f16 HPR:$Rm)))),
           (VMULslhq QPR:$Rn,
             (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)), HPR:$Rm, ssub_0),
             (i32 0))>;
 }
 
 //   VQDMULH  : Vector Saturating Doubling Multiply Returning High Half
 defm VQDMULH  : N3VInt_HS<0, 0, 0b1011, 0, N3RegFrm, IIC_VMULi16D, IIC_VMULi32D,
                           IIC_VMULi16Q, IIC_VMULi32Q,
                           "vqdmulh", "s", int_arm_neon_vqdmulh, 1>;
 defm VQDMULHsl: N3VIntSL_HS<0b1100, IIC_VMULi16D, IIC_VMULi32D,
                             IIC_VMULi16Q, IIC_VMULi32Q,
                             "vqdmulh", "s",  int_arm_neon_vqdmulh>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v8i16 (int_arm_neon_vqdmulh (v8i16 QPR:$src1),
-                                       (v8i16 (NEONvduplane (v8i16 QPR:$src2),
+                                       (v8i16 (ARMvduplane (v8i16 QPR:$src2),
                                                             imm:$lane)))),
           (v8i16 (VQDMULHslv8i16 (v8i16 QPR:$src1),
                                  (v4i16 (EXTRACT_SUBREG QPR:$src2,
                                          (DSubReg_i16_reg imm:$lane))),
                                  (SubReg_i16_lane imm:$lane)))>;
 def : Pat<(v4i32 (int_arm_neon_vqdmulh (v4i32 QPR:$src1),
-                                       (v4i32 (NEONvduplane (v4i32 QPR:$src2),
+                                       (v4i32 (ARMvduplane (v4i32 QPR:$src2),
                                                             imm:$lane)))),
           (v4i32 (VQDMULHslv4i32 (v4i32 QPR:$src1),
                                  (v2i32 (EXTRACT_SUBREG QPR:$src2,
                                          (DSubReg_i32_reg imm:$lane))),
                                  (SubReg_i32_lane imm:$lane)))>;
 }
 
 //   VQRDMULH : Vector Rounding Saturating Doubling Multiply Returning High Half
 defm VQRDMULH   : N3VInt_HS<1, 0, 0b1011, 0, N3RegFrm,
                             IIC_VMULi16D,IIC_VMULi32D,IIC_VMULi16Q,IIC_VMULi32Q,
                             "vqrdmulh", "s", int_arm_neon_vqrdmulh, 1>;
 defm VQRDMULHsl : N3VIntSL_HS<0b1101, IIC_VMULi16D, IIC_VMULi32D,
                               IIC_VMULi16Q, IIC_VMULi32Q,
                               "vqrdmulh", "s",  int_arm_neon_vqrdmulh>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v8i16 (int_arm_neon_vqrdmulh (v8i16 QPR:$src1),
-                                        (v8i16 (NEONvduplane (v8i16 QPR:$src2),
+                                        (v8i16 (ARMvduplane (v8i16 QPR:$src2),
                                                              imm:$lane)))),
           (v8i16 (VQRDMULHslv8i16 (v8i16 QPR:$src1),
                                   (v4i16 (EXTRACT_SUBREG QPR:$src2,
                                           (DSubReg_i16_reg imm:$lane))),
                                   (SubReg_i16_lane imm:$lane)))>;
 def : Pat<(v4i32 (int_arm_neon_vqrdmulh (v4i32 QPR:$src1),
-                                        (v4i32 (NEONvduplane (v4i32 QPR:$src2),
+                                        (v4i32 (ARMvduplane (v4i32 QPR:$src2),
                                                              imm:$lane)))),
           (v4i32 (VQRDMULHslv4i32 (v4i32 QPR:$src1),
                                   (v2i32 (EXTRACT_SUBREG QPR:$src2,
                                           (DSubReg_i32_reg imm:$lane))),
                                   (SubReg_i32_lane imm:$lane)))>;
 }
 
 //   VMULL    : Vector Multiply Long (integer and polynomial) (Q = D * D)
def  VMLAslhd : N3VDMulOpSL16<0b01, 0b0001, IIC_VMACD, "vmla", "f16",
                 Requires<[HasNEON, HasFullFP16, UseFPVMLx]>;
 def  VMLAslhq : N3VQMulOpSL16<0b01, 0b0001, IIC_VMACQ, "vmla", "f16",
                             v8f16, v4f16, fmul, fadd>,
                 Requires<[HasNEON, HasFullFP16, UseFPVMLx]>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v8i16 (add (v8i16 QPR:$src1),
                   (mul (v8i16 QPR:$src2),
-                       (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))),
+                       (v8i16 (ARMvduplane (v8i16 QPR:$src3), imm:$lane))))),
           (v8i16 (VMLAslv8i16 (v8i16 QPR:$src1), (v8i16 QPR:$src2),
                               (v4i16 (EXTRACT_SUBREG QPR:$src3,
                                       (DSubReg_i16_reg imm:$lane))),
                               (SubReg_i16_lane imm:$lane)))>;
 
 def : Pat<(v4i32 (add (v4i32 QPR:$src1),
                   (mul (v4i32 QPR:$src2),
-                       (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))),
+                       (v4i32 (ARMvduplane (v4i32 QPR:$src3), imm:$lane))))),
           (v4i32 (VMLAslv4i32 (v4i32 QPR:$src1), (v4i32 QPR:$src2),
                               (v2i32 (EXTRACT_SUBREG QPR:$src3,
                                       (DSubReg_i32_reg imm:$lane))),
                               (SubReg_i32_lane imm:$lane)))>;
 }
 
 def : Pat<(v4f32 (fadd_mlx (v4f32 QPR:$src1),
                   (fmul_su (v4f32 QPR:$src2),
-                        (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))),
+                        (v4f32 (ARMvduplane (v4f32 QPR:$src3), imm:$lane))))),
           (v4f32 (VMLAslfq (v4f32 QPR:$src1),
                            (v4f32 QPR:$src2),
                            (v2f32 (EXTRACT_SUBREG QPR:$src3,
                                    (DSubReg_i32_reg imm:$lane))),
                            (SubReg_i32_lane imm:$lane)))>,
           Requires<[HasNEON, UseFPVMLx]>;
 
 //   VMLAL    : Vector Multiply Accumulate Long (Q += D * D)
let Predicates = [HasNEON, HasV8_1a] in {
 
   defm VQRDMLAHsl : N3VMulOpSL_HS<0b1110, IIC_VMACi16D, IIC_VMACi32D,
                                   IIC_VMACi16Q, IIC_VMACi32Q, "vqrdmlah", "s",
                                   null_frag>;
   def : Pat<(v4i16 (int_arm_neon_vqadds
                      (v4i16 DPR:$src1),
                      (v4i16 (int_arm_neon_vqrdmulh
                               (v4i16 DPR:$Vn),
-                              (v4i16 (NEONvduplane (v4i16 DPR_8:$Vm),
+                              (v4i16 (ARMvduplane (v4i16 DPR_8:$Vm),
                                                    imm:$lane)))))),
             (v4i16 (VQRDMLAHslv4i16 DPR:$src1, DPR:$Vn, DPR_8:$Vm,
                                     imm:$lane))>;
   def : Pat<(v2i32 (int_arm_neon_vqadds
                      (v2i32 DPR:$src1),
                      (v2i32 (int_arm_neon_vqrdmulh
                               (v2i32 DPR:$Vn),
-                              (v2i32 (NEONvduplane (v2i32 DPR_VFP2:$Vm),
+                              (v2i32 (ARMvduplane (v2i32 DPR_VFP2:$Vm),
                                                    imm:$lane)))))),
             (v2i32 (VQRDMLAHslv2i32 DPR:$src1, DPR:$Vn, DPR_VFP2:$Vm,
                                     imm:$lane))>;
   def : Pat<(v8i16 (int_arm_neon_vqadds
                      (v8i16 QPR:$src1),
                      (v8i16 (int_arm_neon_vqrdmulh
                               (v8i16 QPR:$src2),
-                              (v8i16 (NEONvduplane (v8i16 QPR:$src3),
+                              (v8i16 (ARMvduplane (v8i16 QPR:$src3),
                                                    imm:$lane)))))),
             (v8i16 (VQRDMLAHslv8i16 (v8i16 QPR:$src1),
                                     (v8i16 QPR:$src2),
                                     (v4i16 (EXTRACT_SUBREG
                                              QPR:$src3,
                                              (DSubReg_i16_reg imm:$lane))),
                                     (SubReg_i16_lane imm:$lane)))>;
   def : Pat<(v4i32 (int_arm_neon_vqadds
                      (v4i32 QPR:$src1),
                      (v4i32 (int_arm_neon_vqrdmulh 
                               (v4i32 QPR:$src2),
-                              (v4i32 (NEONvduplane (v4i32 QPR:$src3), 
+                              (v4i32 (ARMvduplane (v4i32 QPR:$src3), 
                                                    imm:$lane)))))),
             (v4i32 (VQRDMLAHslv4i32 (v4i32 QPR:$src1),
                                     (v4i32 QPR:$src2),
                                     (v2i32 (EXTRACT_SUBREG
                                              QPR:$src3,
                                              (DSubReg_i32_reg imm:$lane))),
                                     (SubReg_i32_lane imm:$lane)))>;
 
let Predicates = [HasNEON, HasV8_1a] in {
 
   defm VQRDMLSHsl : N3VMulOpSL_HS<0b1111, IIC_VMACi16D, IIC_VMACi32D,
                                   IIC_VMACi16Q, IIC_VMACi32Q, "vqrdmlsh", "s",
                                   null_frag>;
   def : Pat<(v4i16 (int_arm_neon_vqsubs
                      (v4i16 DPR:$src1),
                      (v4i16 (int_arm_neon_vqrdmulh
                               (v4i16 DPR:$Vn),
-                              (v4i16 (NEONvduplane (v4i16 DPR_8:$Vm),
+                              (v4i16 (ARMvduplane (v4i16 DPR_8:$Vm),
                                                    imm:$lane)))))),
             (v4i16 (VQRDMLSHslv4i16 DPR:$src1, DPR:$Vn, DPR_8:$Vm, imm:$lane))>;
   def : Pat<(v2i32 (int_arm_neon_vqsubs
                      (v2i32 DPR:$src1),
                      (v2i32 (int_arm_neon_vqrdmulh
                               (v2i32 DPR:$Vn),
-                              (v2i32 (NEONvduplane (v2i32 DPR_VFP2:$Vm),
+                              (v2i32 (ARMvduplane (v2i32 DPR_VFP2:$Vm),
                                                    imm:$lane)))))),
             (v2i32 (VQRDMLSHslv2i32 DPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, 
                                     imm:$lane))>;
   def : Pat<(v8i16 (int_arm_neon_vqsubs
                      (v8i16 QPR:$src1),
                      (v8i16 (int_arm_neon_vqrdmulh
                               (v8i16 QPR:$src2),
-                              (v8i16 (NEONvduplane (v8i16 QPR:$src3), 
+                              (v8i16 (ARMvduplane (v8i16 QPR:$src3), 
                                                    imm:$lane)))))),
             (v8i16 (VQRDMLSHslv8i16 (v8i16 QPR:$src1),
                                     (v8i16 QPR:$src2),
                                     (v4i16 (EXTRACT_SUBREG 
                                              QPR:$src3,
                                              (DSubReg_i16_reg imm:$lane))),
                                     (SubReg_i16_lane imm:$lane)))>;
   def : Pat<(v4i32 (int_arm_neon_vqsubs
                      (v4i32 QPR:$src1),
                      (v4i32 (int_arm_neon_vqrdmulh
                               (v4i32 QPR:$src2),
-                              (v4i32 (NEONvduplane (v4i32 QPR:$src3),
+                              (v4i32 (ARMvduplane (v4i32 QPR:$src3),
                                                     imm:$lane)))))),
             (v4i32 (VQRDMLSHslv4i32 (v4i32 QPR:$src1),
                                     (v4i32 QPR:$src2),
                                     (v2i32 (EXTRACT_SUBREG 
                                              QPR:$src3,
                                              (DSubReg_i32_reg imm:$lane))),
                                     (SubReg_i32_lane imm:$lane)))>;
 }
 //   VQDMLAL  : Vector Saturating Doubling Multiply Accumulate Long (Q += D * D)
 defm VQDMLAL  : N3VLInt3_HS<0, 1, 0b1001, 0, IIC_VMACi16D, IIC_VMACi32D,
                             "vqdmlal", "s", null_frag>;
 defm VQDMLALsl: N3VLInt3SL_HS<0, 0b0011, "vqdmlal", "s", null_frag>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v4i32 (int_arm_neon_vqadds (v4i32 QPR:$src1),
                      (v4i32 (int_arm_neon_vqdmull (v4i16 DPR:$Vn),
                                                   (v4i16 DPR:$Vm))))),
           (VQDMLALv4i32 QPR:$src1, DPR:$Vn, DPR:$Vm)>;
 def : Pat<(v2i64 (int_arm_neon_vqadds (v2i64 QPR:$src1),
                      (v2i64 (int_arm_neon_vqdmull (v2i32 DPR:$Vn),
                                                   (v2i32 DPR:$Vm))))),
           (VQDMLALv2i64 QPR:$src1, DPR:$Vn, DPR:$Vm)>;
 def : Pat<(v4i32 (int_arm_neon_vqadds (v4i32 QPR:$src1),
                      (v4i32 (int_arm_neon_vqdmull (v4i16 DPR:$Vn),
-                                (v4i16 (NEONvduplane (v4i16 DPR_8:$Vm),
+                                (v4i16 (ARMvduplane (v4i16 DPR_8:$Vm),
                                                      imm:$lane)))))),
           (VQDMLALslv4i16 QPR:$src1, DPR:$Vn, DPR_8:$Vm, imm:$lane)>;
 def : Pat<(v2i64 (int_arm_neon_vqadds (v2i64 QPR:$src1),
                      (v2i64 (int_arm_neon_vqdmull (v2i32 DPR:$Vn),
-                                (v2i32 (NEONvduplane (v2i32 DPR_VFP2:$Vm),
+                                (v2i32 (ARMvduplane (v2i32 DPR_VFP2:$Vm),
                                                      imm:$lane)))))),
           (VQDMLALslv2i32 QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, imm:$lane)>;
 }
 
 //   VMLS     : Vector Multiply Subtract (integer and floating-point)
 defm VMLS     : N3VMulOp_QHS<1, 0, 0b1001, 0, IIC_VMACi16D, IIC_VMACi32D,
                              IIC_VMACi16Q, IIC_VMACi32Q, "vmls", "i", sub>;
 def  VMLSfd   : N3VDMulOp<0, 0, 0b10, 0b1101, 1, IIC_VMACD, "vmls", "f32",
def  VMLSslhd : N3VDMulOpSL16<0b01, 0b0101, IIC_VMACD, "vmls", "f16",
                 Requires<[HasNEON, HasFullFP16, UseFPVMLx]>;
 def  VMLSslhq : N3VQMulOpSL16<0b01, 0b0101, IIC_VMACQ, "vmls", "f16",
                             v8f16, v4f16, fmul, fsub>,
                 Requires<[HasNEON, HasFullFP16, UseFPVMLx]>;
 
 let Predicates = [HasNEON] in {
 def : Pat<(v8i16 (sub (v8i16 QPR:$src1),
                   (mul (v8i16 QPR:$src2),
-                       (v8i16 (NEONvduplane (v8i16 QPR:$src3), imm:$lane))))),
+                       (v8i16 (ARMvduplane (v8i16 QPR:$src3), imm:$lane))))),
           (v8i16 (VMLSslv8i16 (v8i16 QPR:$src1), (v8i16 QPR:$src2),
                               (v4i16 (EXTRACT_SUBREG QPR:$src3,
                                       (DSubReg_i16_reg imm:$lane))),
                               (SubReg_i16_lane imm:$lane)))>;
 
 def : Pat<(v4i32 (sub (v4i32 QPR:$src1),
                   (mul (v4i32 QPR:$src2),
-                     (v4i32 (NEONvduplane (v4i32 QPR:$src3), imm:$lane))))),
+                     (v4i32 (ARMvduplane (v4i32 QPR:$src3), imm:$lane))))),
           (v4i32 (VMLSslv4i32 (v4i32 QPR:$src1), (v4i32 QPR:$src2),
                               (v2i32 (EXTRACT_SUBREG QPR:$src3,
                                       (DSubReg_i32_reg imm:$lane))),
                               (SubReg_i32_lane imm:$lane)))>;
 }
 
 def : Pat<(v4f32 (fsub_mlx (v4f32 QPR:$src1),
                   (fmul_su (v4f32 QPR:$src2),
-                        (v4f32 (NEONvduplane (v4f32 QPR:$src3), imm:$lane))))),
+                        (v4f32 (ARMvduplane (v4f32 QPR:$src3), imm:$lane))))),
           (v4f32 (VMLSslfq (v4f32 QPR:$src1), (v4f32 QPR:$src2),
                            (v2f32 (EXTRACT_SUBREG QPR:$src3,
                                    (DSubReg_i32_reg imm:$lane))),
                            (SubReg_i32_lane imm:$lane)))>,
           Requires<[HasNEON, UseFPVMLx]>;
 
 //   VMLSL    : Vector Multiply Subtract Long (Q -= D * D)
 defm VMLSLs   : N3VLMulOp_QHS<0,1,0b1010,0, IIC_VMACi16D, IIC_VMACi32D,
def : Pat<(v4i32 (int_arm_neon_vqsubs (v4i32 QPR:$src1),
                                                   (v4i16 DPR:$Vm))))),
           (VQDMLSLv4i32 QPR:$src1, DPR:$Vn, DPR:$Vm)>;
 def : Pat<(v2i64 (int_arm_neon_vqsubs (v2i64 QPR:$src1),
                      (v2i64 (int_arm_neon_vqdmull (v2i32 DPR:$Vn),
                                                   (v2i32 DPR:$Vm))))),
           (VQDMLSLv2i64 QPR:$src1, DPR:$Vn, DPR:$Vm)>;
 def : Pat<(v4i32 (int_arm_neon_vqsubs (v4i32 QPR:$src1),
                      (v4i32 (int_arm_neon_vqdmull (v4i16 DPR:$Vn),
-                                (v4i16 (NEONvduplane (v4i16 DPR_8:$Vm),
+                                (v4i16 (ARMvduplane (v4i16 DPR_8:$Vm),
                                                      imm:$lane)))))),
           (VQDMLSLslv4i16 QPR:$src1, DPR:$Vn, DPR_8:$Vm, imm:$lane)>;
 def : Pat<(v2i64 (int_arm_neon_vqsubs (v2i64 QPR:$src1),
                      (v2i64 (int_arm_neon_vqdmull (v2i32 DPR:$Vn),
-                                (v2i32 (NEONvduplane (v2i32 DPR_VFP2:$Vm),
+                                (v2i32 (ARMvduplane (v2i32 DPR_VFP2:$Vm),
                                                      imm:$lane)))))),
           (VQDMLSLslv2i32 QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, imm:$lane)>;
 }
 
 // Fused Vector Multiply-Accumulate and Fused Multiply-Subtract Operations.
 def  VFMAfd   : N3VDMulOp<0, 0, 0b00, 0b1100, 1, IIC_VFMACD, "vfma", "f32",
                           v2f32, fmul_su, fadd_mlx>,
                 Requires<[HasNEON,HasVFP4,UseFusedMAC]>;
def "" : N3Vnp<0b11100, 0b10, 0b1101, Q, U, (outs Ty:$dst),
     let Predicates = [HasDotProd];
     let DecoderNamespace = "VFPV8";
   }
 
   def : Pat<
     (AccumType (OpNode (AccumType Ty:$Vd),
                        (InputType Ty:$Vn),
                        (InputType (bitconvert (AccumType
-                                  (NEONvduplane (AccumType Ty:$Vm),
+                                  (ARMvduplane (AccumType Ty:$Vm),
                                                  VectorIndex32:$lane)))))),
     (!cast<Instruction>(NAME) Ty:$Vd, Ty:$Vn, RHS, VectorIndex32:$lane)>;
 }
 
 defm VUDOTDI : DOTI<"vudot", "u8", 0b0, 0b1, DPR, v2i32, v8i8,
                     int_arm_neon_udot, (v2i32 DPR_VFP2:$Vm)>;
 defm VSDOTDI : DOTI<"vsdot", "s8", 0b0, 0b0, DPR, v2i32, v8i8,
                     int_arm_neon_sdot, (v2i32 DPR_VFP2:$Vm)>;
def VMOVQ0 : ARMPseudoExpand<(outs QPR:$Vd), (ins), 4, IIC_VMOVImm,
                Requires<[HasZCZ]>;
 }
 
 //   VMOV     : Vector Get Lane (move scalar to ARM core register)
 
 def VGETLNs8  : NVGetLane<{1,1,1,0,0,1,?,1}, 0b1011, {?,?},
                           (outs GPR:$R), (ins DPR:$V, VectorIndex8:$lane),
                           IIC_VMOVSI, "vmov", "s8", "$R, $V$lane",
-                          [(set GPR:$R, (NEONvgetlanes (v8i8 DPR:$V),
+                          [(set GPR:$R, (ARMvgetlanes (v8i8 DPR:$V),
                                            imm:$lane))]> {
   let Inst{21}  = lane{2};
   let Inst{6-5} = lane{1-0};
 }
 def VGETLNs16 : NVGetLane<{1,1,1,0,0,0,?,1}, 0b1011, {?,1},
                           (outs GPR:$R), (ins DPR:$V, VectorIndex16:$lane),
                           IIC_VMOVSI, "vmov", "s16", "$R, $V$lane",
-                          [(set GPR:$R, (NEONvgetlanes (v4i16 DPR:$V),
+                          [(set GPR:$R, (ARMvgetlanes (v4i16 DPR:$V),
                                            imm:$lane))]> {
   let Inst{21} = lane{1};
   let Inst{6}  = lane{0};
 }
 def VGETLNu8  : NVGetLane<{1,1,1,0,1,1,?,1}, 0b1011, {?,?},
                           (outs GPR:$R), (ins DPR:$V, VectorIndex8:$lane),
                           IIC_VMOVSI, "vmov", "u8", "$R, $V$lane",
-                          [(set GPR:$R, (NEONvgetlaneu (v8i8 DPR:$V),
+                          [(set GPR:$R, (ARMvgetlaneu (v8i8 DPR:$V),
                                            imm:$lane))]> {
   let Inst{21}  = lane{2};
   let Inst{6-5} = lane{1-0};
 }
 def VGETLNu16 : NVGetLane<{1,1,1,0,1,0,?,1}, 0b1011, {?,1},
                           (outs GPR:$R), (ins DPR:$V, VectorIndex16:$lane),
                           IIC_VMOVSI, "vmov", "u16", "$R, $V$lane",
-                          [(set GPR:$R, (NEONvgetlaneu (v4i16 DPR:$V),
+                          [(set GPR:$R, (ARMvgetlaneu (v4i16 DPR:$V),
                                            imm:$lane))]> {
   let Inst{21} = lane{1};
   let Inst{6}  = lane{0};
 }
 def VGETLNi32 : NVGetLane<{1,1,1,0,0,0,?,1}, 0b1011, 0b00,
                           (outs GPR:$R), (ins DPR:$V, VectorIndex32:$lane),
                           IIC_VMOVSI, "vmov", "32", "$R, $V$lane",
                           [(set GPR:$R, (extractelt (v2i32 DPR:$V),
                                            imm:$lane))]>,
                 Requires<[HasFPRegs, HasFastVGETLNi32]> {
   let Inst{21} = lane{0};
 }
 let Predicates = [HasNEON] in {
 // def VGETLNf32: see FMRDH and FMRDL in ARMInstrVFP.td
-def : Pat<(NEONvgetlanes (v16i8 QPR:$src), imm:$lane),
+def : Pat<(ARMvgetlanes (v16i8 QPR:$src), imm:$lane),
           (VGETLNs8 (v8i8 (EXTRACT_SUBREG QPR:$src,
                            (DSubReg_i8_reg imm:$lane))),
                      (SubReg_i8_lane imm:$lane))>;
-def : Pat<(NEONvgetlanes (v8i16 QPR:$src), imm:$lane),
+def : Pat<(ARMvgetlanes (v8i16 QPR:$src), imm:$lane),
           (VGETLNs16 (v4i16 (EXTRACT_SUBREG QPR:$src,
                              (DSubReg_i16_reg imm:$lane))),
                      (SubReg_i16_lane imm:$lane))>;
-def : Pat<(NEONvgetlaneu (v16i8 QPR:$src), imm:$lane),
+def : Pat<(ARMvgetlaneu (v16i8 QPR:$src), imm:$lane),
           (VGETLNu8 (v8i8 (EXTRACT_SUBREG QPR:$src,
                            (DSubReg_i8_reg imm:$lane))),
                      (SubReg_i8_lane imm:$lane))>;
-def : Pat<(NEONvgetlaneu (v8i16 QPR:$src), imm:$lane),
+def : Pat<(ARMvgetlaneu (v8i16 QPR:$src), imm:$lane),
           (VGETLNu16 (v4i16 (EXTRACT_SUBREG QPR:$src,
                              (DSubReg_i16_reg imm:$lane))),
                      (SubReg_i16_lane imm:$lane))>;
 }
 def : Pat<(extractelt (v4i32 QPR:$src), imm:$lane),
           (VGETLNi32 (v2i32 (EXTRACT_SUBREG QPR:$src,
                              (DSubReg_i32_reg imm:$lane))),
                      (SubReg_i32_lane imm:$lane))>,
def : Pat<(v4i32 (scalar_to_vector GPR:$src)),
                          dsub_0)>;
 }
 
 //   VDUP     : Vector Duplicate (from ARM core register to all elements)
 
 class VDUPD<bits<8> opcod1, bits<2> opcod3, string Dt, ValueType Ty>
   : NVDup<opcod1, 0b1011, opcod3, (outs DPR:$V), (ins GPR:$R),
           IIC_VMOVIS, "vdup", Dt, "$V, $R",
-          [(set DPR:$V, (Ty (NEONvdup (i32 GPR:$R))))]>;
+          [(set DPR:$V, (Ty (ARMvdup (i32 GPR:$R))))]>;
 class VDUPQ<bits<8> opcod1, bits<2> opcod3, string Dt, ValueType Ty>
   : NVDup<opcod1, 0b1011, opcod3, (outs QPR:$V), (ins GPR:$R),
           IIC_VMOVIS, "vdup", Dt, "$V, $R",
-          [(set QPR:$V, (Ty (NEONvdup (i32 GPR:$R))))]>;
+          [(set QPR:$V, (Ty (ARMvdup (i32 GPR:$R))))]>;
 
 def  VDUP8d   : VDUPD<0b11101100, 0b00, "8", v8i8>;
 def  VDUP16d  : VDUPD<0b11101000, 0b01, "16", v4i16>;
 def  VDUP32d  : VDUPD<0b11101000, 0b00, "32", v2i32>,
                 Requires<[HasNEON, HasFastVDUP32]>;
 def  VDUP8q   : VDUPQ<0b11101110, 0b00, "8", v16i8>;
 def  VDUP16q  : VDUPQ<0b11101010, 0b01, "16", v8i16>;
 def  VDUP32q  : VDUPQ<0b11101010, 0b00, "32", v4i32>;
 
-// NEONvdup patterns for uarchs with fast VDUP.32.
-def : Pat<(v2f32 (NEONvdup (f32 (bitconvert GPR:$R)))), (VDUP32d GPR:$R)>,
+// ARMvdup patterns for uarchs with fast VDUP.32.
+def : Pat<(v2f32 (ARMvdup (f32 (bitconvert GPR:$R)))), (VDUP32d GPR:$R)>,
       Requires<[HasNEON,HasFastVDUP32]>;
-def : Pat<(v4f32 (NEONvdup (f32 (bitconvert GPR:$R)))), (VDUP32q GPR:$R)>,
+def : Pat<(v4f32 (ARMvdup (f32 (bitconvert GPR:$R)))), (VDUP32q GPR:$R)>,
       Requires<[HasNEON]>;
 
-// NEONvdup patterns for uarchs with slow VDUP.32 - use VMOVDRR instead.
-def : Pat<(v2i32 (NEONvdup (i32 GPR:$R))), (VMOVDRR GPR:$R, GPR:$R)>,
+// ARMvdup patterns for uarchs with slow VDUP.32 - use VMOVDRR instead.
+def : Pat<(v2i32 (ARMvdup (i32 GPR:$R))), (VMOVDRR GPR:$R, GPR:$R)>,
       Requires<[HasNEON,HasSlowVDUP32]>;
-def : Pat<(v2f32 (NEONvdup (f32 (bitconvert GPR:$R)))), (VMOVDRR GPR:$R, GPR:$R)>,
+def : Pat<(v2f32 (ARMvdup (f32 (bitconvert GPR:$R)))), (VMOVDRR GPR:$R, GPR:$R)>,
       Requires<[HasNEON,HasSlowVDUP32]>;
 
 //   VDUP     : Vector Duplicate Lane (from scalar to all elements)
 
 class VDUPLND<bits<4> op19_16, string OpcodeStr, string Dt,
               ValueType Ty, Operand IdxTy>
   : NVDupLane<op19_16, 0, (outs DPR:$Vd), (ins DPR:$Vm, IdxTy:$lane),
               IIC_VMOVD, OpcodeStr, Dt, "$Vd, $Vm$lane",
-              [(set DPR:$Vd, (Ty (NEONvduplane (Ty DPR:$Vm), imm:$lane)))]>;
+              [(set DPR:$Vd, (Ty (ARMvduplane (Ty DPR:$Vm), imm:$lane)))]>;
 
 class VDUPLNQ<bits<4> op19_16, string OpcodeStr, string Dt,
               ValueType ResTy, ValueType OpTy, Operand IdxTy>
   : NVDupLane<op19_16, 1, (outs QPR:$Vd), (ins DPR:$Vm, IdxTy:$lane),
               IIC_VMOVQ, OpcodeStr, Dt, "$Vd, $Vm$lane",
-              [(set QPR:$Vd, (ResTy (NEONvduplane (OpTy DPR:$Vm),
+              [(set QPR:$Vd, (ResTy (ARMvduplane (OpTy DPR:$Vm),
                                       VectorIndex32:$lane)))]>;
 
 // Inst{19-16} is partially specified depending on the element size.
 
 def VDUPLN8d  : VDUPLND<{?,?,?,1}, "vdup", "8", v8i8, VectorIndex8> {
   bits<3> lane;
   let Inst{19-17} = lane{2-0};
 }
def VDUPLN16q : VDUPLNQ<{?,?,1,0}, "vdup", "16", v8i16, v4i16, VectorIndex16> {
   let Inst{19-18} = lane{1-0};
 }
 def VDUPLN32q : VDUPLNQ<{?,1,0,0}, "vdup", "32", v4i32, v2i32, VectorIndex32> {
   bits<1> lane;
   let Inst{19} = lane{0};
 }
 
 let Predicates = [HasNEON] in {
-def : Pat<(v4f16 (NEONvduplane (v4f16 DPR:$Vm), imm:$lane)),
+def : Pat<(v4f16 (ARMvduplane (v4f16 DPR:$Vm), imm:$lane)),
           (VDUPLN32d DPR:$Vm, imm:$lane)>;
 
-def : Pat<(v2f32 (NEONvduplane (v2f32 DPR:$Vm), imm:$lane)),
+def : Pat<(v2f32 (ARMvduplane (v2f32 DPR:$Vm), imm:$lane)),
           (VDUPLN32d DPR:$Vm, imm:$lane)>;
 
-def : Pat<(v4f32 (NEONvduplane (v2f32 DPR:$Vm), imm:$lane)),
+def : Pat<(v4f32 (ARMvduplane (v2f32 DPR:$Vm), imm:$lane)),
           (VDUPLN32q DPR:$Vm, imm:$lane)>;
 
-def : Pat<(v16i8 (NEONvduplane (v16i8 QPR:$src), imm:$lane)),
+def : Pat<(v16i8 (ARMvduplane (v16i8 QPR:$src), imm:$lane)),
           (v16i8 (VDUPLN8q (v8i8 (EXTRACT_SUBREG QPR:$src,
                                   (DSubReg_i8_reg imm:$lane))),
                            (SubReg_i8_lane imm:$lane)))>;
-def : Pat<(v8i16 (NEONvduplane (v8i16 QPR:$src), imm:$lane)),
+def : Pat<(v8i16 (ARMvduplane (v8i16 QPR:$src), imm:$lane)),
           (v8i16 (VDUPLN16q (v4i16 (EXTRACT_SUBREG QPR:$src,
                                     (DSubReg_i16_reg imm:$lane))),
                             (SubReg_i16_lane imm:$lane)))>;
-def : Pat<(v8f16 (NEONvduplane (v8f16 QPR:$src), imm:$lane)),
+def : Pat<(v8f16 (ARMvduplane (v8f16 QPR:$src), imm:$lane)),
           (v8f16 (VDUPLN16q (v4f16 (EXTRACT_SUBREG QPR:$src,
                                     (DSubReg_i16_reg imm:$lane))),
                             (SubReg_i16_lane imm:$lane)))>;
-def : Pat<(v4i32 (NEONvduplane (v4i32 QPR:$src), imm:$lane)),
+def : Pat<(v4i32 (ARMvduplane (v4i32 QPR:$src), imm:$lane)),
           (v4i32 (VDUPLN32q (v2i32 (EXTRACT_SUBREG QPR:$src,
                                     (DSubReg_i32_reg imm:$lane))),
                             (SubReg_i32_lane imm:$lane)))>;
-def : Pat<(v4f32 (NEONvduplane (v4f32 QPR:$src), imm:$lane)),
+def : Pat<(v4f32 (ARMvduplane (v4f32 QPR:$src), imm:$lane)),
           (v4f32 (VDUPLN32q (v2f32 (EXTRACT_SUBREG QPR:$src,
                                    (DSubReg_i32_reg imm:$lane))),
                            (SubReg_i32_lane imm:$lane)))>;
 
-def : Pat<(v4f16 (NEONvdup HPR:$src)),
+def : Pat<(v4f16 (ARMvdup HPR:$src)),
           (v4f16 (VDUPLN16d (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)),
                              HPR:$src, ssub_0), (i32 0)))>;
-def : Pat<(v2f32 (NEONvdup (f32 SPR:$src))),
+def : Pat<(v2f32 (ARMvdup (f32 SPR:$src))),
           (v2f32 (VDUPLN32d (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
                              SPR:$src, ssub_0), (i32 0)))>;
-def : Pat<(v4f32 (NEONvdup (f32 SPR:$src))),
+def : Pat<(v4f32 (ARMvdup (f32 SPR:$src))),
           (v4f32 (VDUPLN32q (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
                              SPR:$src, ssub_0), (i32 0)))>;
-def : Pat<(v8f16 (NEONvdup HPR:$src)),
+def : Pat<(v8f16 (ARMvdup HPR:$src)),
           (v8f16 (VDUPLN16q (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)),
                              HPR:$src, ssub_0), (i32 0)))>;
 }
 
 //   VMOVN    : Vector Narrowing Move
 defm VMOVN    : N2VN_HSD<0b11,0b11,0b10,0b00100,0,0, IIC_VMOVN,
                          "vmovn", "i", trunc>;
 //   VQMOVN   : Vector Saturating Narrowing Move

 // Vector Reverse.
 
 //   VREV64   : Vector Reverse elements within 64-bit doublewords
 
 class VREV64D<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00000, 0, 0, (outs DPR:$Vd),
         (ins DPR:$Vm), IIC_VMOVD,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set DPR:$Vd, (Ty (NEONvrev64 (Ty DPR:$Vm))))]>;
+        [(set DPR:$Vd, (Ty (ARMvrev64 (Ty DPR:$Vm))))]>;
 class VREV64Q<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00000, 1, 0, (outs QPR:$Vd),
         (ins QPR:$Vm), IIC_VMOVQ,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set QPR:$Vd, (Ty (NEONvrev64 (Ty QPR:$Vm))))]>;
+        [(set QPR:$Vd, (Ty (ARMvrev64 (Ty QPR:$Vm))))]>;
 
 def VREV64d8  : VREV64D<0b00, "vrev64", "8", v8i8>;
 def VREV64d16 : VREV64D<0b01, "vrev64", "16", v4i16>;
 def VREV64d32 : VREV64D<0b10, "vrev64", "32", v2i32>;
 let Predicates = [HasNEON] in {
-def : Pat<(v2f32 (NEONvrev64 (v2f32 DPR:$Vm))), (VREV64d32 DPR:$Vm)>;
+def : Pat<(v2f32 (ARMvrev64 (v2f32 DPR:$Vm))), (VREV64d32 DPR:$Vm)>;
 }
 
 def VREV64q8  : VREV64Q<0b00, "vrev64", "8", v16i8>;
 def VREV64q16 : VREV64Q<0b01, "vrev64", "16", v8i16>;
 def VREV64q32 : VREV64Q<0b10, "vrev64", "32", v4i32>;
 
 let Predicates = [HasNEON] in {
-def : Pat<(v4f32 (NEONvrev64 (v4f32 QPR:$Vm))), (VREV64q32 QPR:$Vm)>;
-def : Pat<(v8f16 (NEONvrev64 (v8f16 QPR:$Vm))), (VREV64q16 QPR:$Vm)>;
-def : Pat<(v4f16 (NEONvrev64 (v4f16 DPR:$Vm))), (VREV64d16 DPR:$Vm)>;
+def : Pat<(v4f32 (ARMvrev64 (v4f32 QPR:$Vm))), (VREV64q32 QPR:$Vm)>;
+def : Pat<(v8f16 (ARMvrev64 (v8f16 QPR:$Vm))), (VREV64q16 QPR:$Vm)>;
+def : Pat<(v4f16 (ARMvrev64 (v4f16 DPR:$Vm))), (VREV64d16 DPR:$Vm)>;
 }
 
 //   VREV32   : Vector Reverse elements within 32-bit words
 
 class VREV32D<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00001, 0, 0, (outs DPR:$Vd),
         (ins DPR:$Vm), IIC_VMOVD,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set DPR:$Vd, (Ty (NEONvrev32 (Ty DPR:$Vm))))]>;
+        [(set DPR:$Vd, (Ty (ARMvrev32 (Ty DPR:$Vm))))]>;
 class VREV32Q<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00001, 1, 0, (outs QPR:$Vd),
         (ins QPR:$Vm), IIC_VMOVQ,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set QPR:$Vd, (Ty (NEONvrev32 (Ty QPR:$Vm))))]>;
+        [(set QPR:$Vd, (Ty (ARMvrev32 (Ty QPR:$Vm))))]>;
 
 def VREV32d8  : VREV32D<0b00, "vrev32", "8", v8i8>;
 def VREV32d16 : VREV32D<0b01, "vrev32", "16", v4i16>;
 
 def VREV32q8  : VREV32Q<0b00, "vrev32", "8", v16i8>;
 def VREV32q16 : VREV32Q<0b01, "vrev32", "16", v8i16>;
 
 //   VREV16   : Vector Reverse elements within 16-bit halfwords
 
 class VREV16D<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00010, 0, 0, (outs DPR:$Vd),
         (ins DPR:$Vm), IIC_VMOVD,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set DPR:$Vd, (Ty (NEONvrev16 (Ty DPR:$Vm))))]>;
+        [(set DPR:$Vd, (Ty (ARMvrev16 (Ty DPR:$Vm))))]>;
 class VREV16Q<bits<2> op19_18, string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, 0b00, 0b00010, 1, 0, (outs QPR:$Vd),
         (ins QPR:$Vm), IIC_VMOVQ,
         OpcodeStr, Dt, "$Vd, $Vm", "",
-        [(set QPR:$Vd, (Ty (NEONvrev16 (Ty QPR:$Vm))))]>;
+        [(set QPR:$Vd, (Ty (ARMvrev16 (Ty QPR:$Vm))))]>;
 
 def VREV16d8  : VREV16D<0b00, "vrev16", "8", v8i8>;
 def VREV16q8  : VREV16Q<0b00, "vrev16", "8", v16i8>;
 
 // Other Vector Shuffles.
 
 //  Aligned extractions: really just dropping registers
 

llvm/test/CodeGen/Thumb2/mve-shuffle.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -O3 -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve.fp %s -o - | FileCheck %s
+
+define arm_aapcs_vfpcc <4 x i32> @shuffle1_i32(<4 x i32> %src) {
+; CHECK-LABEL: shuffle1_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f32 s4, s3
+; CHECK-NEXT:    vmov.f32 s5, s2
+; CHECK-NEXT:    vmov.f32 s6, s1
+; CHECK-NEXT:    vmov.f32 s7, s0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @shuffle2_i32(<4 x i32> %src) {
+; CHECK-LABEL: shuffle2_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @shuffle3_i32(<4 x i32> %src) {
+; CHECK-LABEL: shuffle3_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f32 s4, s3
+; CHECK-NEXT:    vmov.f32 s5, s1
+; CHECK-NEXT:    vmov.f32 s6, s2
+; CHECK-NEXT:    vmov.f32 s7, s0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 3, i32 1, i32 2, i32 0>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @shuffle5_i32(<4 x i32> %src) {
+; CHECK-LABEL: shuffle5_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.32 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <4 x i32> @shuffle6_i32(<4 x i32> %src) {
+; CHECK-LABEL: shuffle6_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 undef, i32 undef, i32 undef, i32 3>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @shuffle1_i16(<8 x i16> %src) {
+; CHECK-LABEL: shuffle1_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov q1, q0
+; CHECK-NEXT:    vmov.u16 r0, q0[7]
+; CHECK-NEXT:    vmov.16 q0[0], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[6]
+; CHECK-NEXT:    vmov.16 q0[1], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[5]
+; CHECK-NEXT:    vmov.16 q0[2], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[4]
+; CHECK-NEXT:    vmov.16 q0[3], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[3]
+; CHECK-NEXT:    vmov.16 q0[4], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[2]
+; CHECK-NEXT:    vmov.16 q0[5], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[1]
+; CHECK-NEXT:    vmov.16 q0[6], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[0]
+; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @shuffle2_i16(<8 x i16> %src) {
+; CHECK-LABEL: shuffle2_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @shuffle3_i16(<8 x i16> %src) {
+; CHECK-LABEL: shuffle3_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov q1, q0
+; CHECK-NEXT:    vmov.u16 r0, q0[4]
+; CHECK-NEXT:    vmov.16 q0[0], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[5]
+; CHECK-NEXT:    vmov.16 q0[1], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[7]
+; CHECK-NEXT:    vmov.16 q0[2], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[6]
+; CHECK-NEXT:    vmov.16 q0[3], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[3]
+; CHECK-NEXT:    vmov.16 q0[4], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[1]
+; CHECK-NEXT:    vmov.16 q0[5], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[2]
+; CHECK-NEXT:    vmov.16 q0[6], r0
+; CHECK-NEXT:    vmov.u16 r0, q1[0]
+; CHECK-NEXT:    vmov.16 q0[7], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 4, i32 5, i32 7, i32 6, i32 3, i32 1, i32 2, i32 0>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @shuffle5_i16(<8 x i16> %src) {
+; CHECK-LABEL: shuffle5_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.16 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @shuffle6_i16(<8 x i16> %src) {
+; CHECK-LABEL: shuffle6_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev32.16 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle1_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle1_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov q1, q0
+; CHECK-NEXT:    vmov.u8 r0, q0[15]
+; CHECK-NEXT:    vmov.8 q0[0], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[14]
+; CHECK-NEXT:    vmov.8 q0[1], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[13]
+; CHECK-NEXT:    vmov.8 q0[2], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[12]
+; CHECK-NEXT:    vmov.8 q0[3], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[11]
+; CHECK-NEXT:    vmov.8 q0[4], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[10]
+; CHECK-NEXT:    vmov.8 q0[5], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[9]
+; CHECK-NEXT:    vmov.8 q0[6], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[8]
+; CHECK-NEXT:    vmov.8 q0[7], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[7]
+; CHECK-NEXT:    vmov.8 q0[8], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[6]
+; CHECK-NEXT:    vmov.8 q0[9], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[5]
+; CHECK-NEXT:    vmov.8 q0[10], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[4]
+; CHECK-NEXT:    vmov.8 q0[11], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[3]
+; CHECK-NEXT:    vmov.8 q0[12], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[2]
+; CHECK-NEXT:    vmov.8 q0[13], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[1]
+; CHECK-NEXT:    vmov.8 q0[14], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[0]
+; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle2_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle2_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle3_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle3_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov q1, q0
+; CHECK-NEXT:    vmov.u8 r0, q0[4]
+; CHECK-NEXT:    vmov.8 q0[0], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[5]
+; CHECK-NEXT:    vmov.8 q0[1], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[15]
+; CHECK-NEXT:    vmov.8 q0[2], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[7]
+; CHECK-NEXT:    vmov.8 q0[3], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[14]
+; CHECK-NEXT:    vmov.8 q0[4], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[9]
+; CHECK-NEXT:    vmov.8 q0[5], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[6]
+; CHECK-NEXT:    vmov.8 q0[6], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[3]
+; CHECK-NEXT:    vmov.8 q0[7], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[10]
+; CHECK-NEXT:    vmov.8 q0[8], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[12]
+; CHECK-NEXT:    vmov.8 q0[9], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[1]
+; CHECK-NEXT:    vmov.8 q0[10], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[13]
+; CHECK-NEXT:    vmov.8 q0[11], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[2]
+; CHECK-NEXT:    vmov.8 q0[12], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[8]
+; CHECK-NEXT:    vmov.8 q0[13], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[0]
+; CHECK-NEXT:    vmov.8 q0[14], r0
+; CHECK-NEXT:    vmov.u8 r0, q1[11]
+; CHECK-NEXT:    vmov.8 q0[15], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 4, i32 5, i32 15, i32 7, i32 14, i32 9, i32 6, i32 3, i32 10, i32 12, i32 1, i32 13, i32 2, i32 8, i32 0, i32 11>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle5_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle5_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.8 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle6_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle6_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev32.8 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 11, i32 10, i32 9, i32 8, i32 15, i32 14, i32 13, i32 12>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @shuffle7_i8(<16 x i8> %src) {
+; CHECK-LABEL: shuffle7_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev16.8 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6, i32 9, i32 8, i32 11, i32 10, i32 13, i32 12, i32 15, i32 14>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @shuffle1_f32(<4 x float> %src) {
+; CHECK-LABEL: shuffle1_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f32 s4, s3
+; CHECK-NEXT:    vmov.f32 s5, s2
+; CHECK-NEXT:    vmov.f32 s6, s1
+; CHECK-NEXT:    vmov.f32 s7, s0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x float> %src, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @shuffle2_f32(<4 x float> %src) {
+; CHECK-LABEL: shuffle2_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x float> %src, <4 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @shuffle3_f32(<4 x float> %src) {
+; CHECK-LABEL: shuffle3_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f32 s4, s3
+; CHECK-NEXT:    vmov.f32 s5, s1
+; CHECK-NEXT:    vmov.f32 s6, s2
+; CHECK-NEXT:    vmov.f32 s7, s0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x float> %src, <4 x float> undef, <4 x i32> <i32 3, i32 1, i32 2, i32 0>
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @shuffle5_f32(<4 x float> %src) {
+; CHECK-LABEL: shuffle5_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.32 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x float> %src, <4 x float> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @shuffle1_f16(<8 x half> %src) {
+; CHECK-LABEL: shuffle1_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[7]
+; CHECK-NEXT:    vmov.u16 r1, q0[6]
+; CHECK-NEXT:    vmov.16 q1[0], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[5]
+; CHECK-NEXT:    vmov.16 q1[1], r1
+; CHECK-NEXT:    vmov.16 q1[2], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[4]
+; CHECK-NEXT:    vmov.16 q1[3], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[3]
+; CHECK-NEXT:    vmov.16 q1[4], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[2]
+; CHECK-NEXT:    vmov.16 q1[5], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[1]
+; CHECK-NEXT:    vmov.16 q1[6], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[0]
+; CHECK-NEXT:    vmov.16 q1[7], r0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+  ret <8 x half> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @shuffle2_f16(<8 x half> %src) {
+; CHECK-LABEL: shuffle2_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x half> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @shuffle3_f16(<8 x half> %src) {
+; CHECK-LABEL: shuffle3_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[4]
+; CHECK-NEXT:    vmov.u16 r1, q0[5]
+; CHECK-NEXT:    vmov.16 q1[0], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[7]
+; CHECK-NEXT:    vmov.16 q1[1], r1
+; CHECK-NEXT:    vmov.16 q1[2], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[6]
+; CHECK-NEXT:    vmov.16 q1[3], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[3]
+; CHECK-NEXT:    vmov.16 q1[4], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[1]
+; CHECK-NEXT:    vmov.16 q1[5], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[2]
+; CHECK-NEXT:    vmov.16 q1[6], r0
+; CHECK-NEXT:    vmov.u16 r0, q0[0]
+; CHECK-NEXT:    vmov.16 q1[7], r0
+; CHECK-NEXT:    vmov q0, q1
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 4, i32 5, i32 7, i32 6, i32 3, i32 1, i32 2, i32 0>
+  ret <8 x half> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @shuffle5_f16(<8 x half> %src) {
+; CHECK-LABEL: shuffle5_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev64.16 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
+  ret <8 x half> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @shuffle6_f16(<8 x half> %src) {
+; CHECK-LABEL: shuffle6_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vrev32.16 q0, q0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>
+  ret <8 x half> %out
+}
+
+
+define arm_aapcs_vfpcc <4 x i32> @insert_i32(i32 %a) {
+; CHECK-LABEL: insert_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.32 q0[0], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = insertelement <4 x i32> undef, i32 %a, i32 0
+  ret <4 x i32> %res
+}
+
+define arm_aapcs_vfpcc <8 x i16> @insert_i16(i16 %a) {
+; CHECK-LABEL: insert_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.16 q0[0], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = insertelement <8 x i16> undef, i16 %a, i32 0
+  ret <8 x i16> %res
+}
+
+define arm_aapcs_vfpcc <16 x i8> @insert_i8(i8 %a) {
+; CHECK-LABEL: insert_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.8 q0[0], r0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = insertelement <16 x i8> undef, i8 %a, i32 0
+  ret <16 x i8> %res
+}
+
+define arm_aapcs_vfpcc <4 x float> @insert_f32(float %a) {
+; CHECK-LABEL: insert_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    @ kill: def $s0 killed $s0 def $q0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = insertelement <4 x float> undef, float %a, i32 0
+  ret <4 x float> %res
+}
+
+; TODO: Calling convention needs fixing to pass half types directly to functions
+define arm_aapcs_vfpcc <8 x half> @insert_f16(half *%aa) {
+; CHECK-LABEL: insert_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vldr.16 s0, [r0]
+; CHECK-NEXT:    bx lr
+entry:
+  %a = load half, half* %aa
+  %res = insertelement <8 x half> undef, half %a, i32 0
+  ret <8 x half> %res
+}
+
+define arm_aapcs_vfpcc i64 @scalar_to_vector_i32(<8 x i16> %v) {
+; CHECK-LABEL: scalar_to_vector_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    .pad #8
+; CHECK-NEXT:    sub sp, #8
+; CHECK-NEXT:    movs r0, #7
+; CHECK-NEXT:    movs r1, #1
+; CHECK-NEXT:    strh.w r0, [sp, #2]
+; CHECK-NEXT:    vmov.u16 r0, q0[0]
+; CHECK-NEXT:    strh.w r0, [sp]
+; CHECK-NEXT:    movt r1, #9
+; CHECK-NEXT:    ldr r0, [sp]
+; CHECK-NEXT:    add sp, #8
+; CHECK-NEXT:    bx lr
+entry:
+  %f = shufflevector <8 x i16> %v, <8 x i16> <i16 undef, i16 7, i16 1, i16 9, i16 undef, i16 undef, i16 undef, i16 undef>, <4 x i32> <i32 0, i32 9, i32 10, i32 11>
+  %0 = bitcast <4 x i16> %f to i64
+  ret i64 %0
+}
+
+
+define arm_aapcs_vfpcc i32 @extract_i32_0(<4 x i32> %a) {
+; CHECK-LABEL: extract_i32_0:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov r0, s0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <4 x i32> %a, i32 0
+  ret i32 %res
+}
+
+define arm_aapcs_vfpcc i32 @extract_i32_3(<4 x i32> %a) {
+; CHECK-LABEL: extract_i32_3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov r0, s3
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <4 x i32> %a, i32 3
+  ret i32 %res
+}
+
+define arm_aapcs_vfpcc i16 @extract_i16_0(<8 x i16> %a) {
+; CHECK-LABEL: extract_i16_0:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[0]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <8 x i16> %a, i32 0
+  ret i16 %res
+}
+
+define arm_aapcs_vfpcc i16 @extract_i16_3(<8 x i16> %a) {
+; CHECK-LABEL: extract_i16_3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[3]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <8 x i16> %a, i32 3
+  ret i16 %res
+}
+
+define arm_aapcs_vfpcc i8 @extract_i8_0(<16 x i8> %a) {
+; CHECK-LABEL: extract_i8_0:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u8 r0, q0[0]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <16 x i8> %a, i32 0
+  ret i8 %res
+}
+
+define arm_aapcs_vfpcc i8 @extract_i8_3(<16 x i8> %a) {
+; CHECK-LABEL: extract_i8_3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u8 r0, q0[3]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <16 x i8> %a, i32 3
+  ret i8 %res
+}
+
+define arm_aapcs_vfpcc float @extract_f32_0(<4 x float> %a) {
+; CHECK-LABEL: extract_f32_0:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    @ kill: def $s0 killed $s0 killed $q0
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <4 x float> %a, i32 0
+  ret float %res
+}
+
+define arm_aapcs_vfpcc float @extract_f32_3(<4 x float> %a) {
+; CHECK-LABEL: extract_f32_3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.f32 s0, s3
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <4 x float> %a, i32 3
+  ret float %res
+}
+
+define arm_aapcs_vfpcc half @extract_f16_0(<8 x half> %a) {
+; CHECK-LABEL: extract_f16_0:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r1, q0[0]
+; CHECK-NEXT:    vmov s0, r1
+; CHECK-NEXT:    vstr.16 s0, [r0]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <8 x half> %a, i32 0
+  ret half %res
+}
+
+; TODO: Calling convention needs fixing to pass half types correctly
+define arm_aapcs_vfpcc half @extract_f16_3(<8 x half> %a) {
+; CHECK-LABEL: extract_f16_3:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r1, q0[3]
+; CHECK-NEXT:    vmov s0, r1
+; CHECK-NEXT:    vstr.16 s0, [r0]
+; CHECK-NEXT:    bx lr
+entry:
+  %res = extractelement <8 x half> %a, i32 3
+  ret half %res
+}

llvm/test/CodeGen/Thumb2/mve-vdup.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve.fp %s -o - | FileCheck %s
+
+define arm_aapcs_vfpcc <4 x i32> @vdup_i32(i32 %src) {
+; CHECK-LABEL: vdup_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vdup.32 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = insertelement <4 x i32> undef, i32 %src, i32 0
+  %out = shufflevector <4 x i32> %0, <4 x i32> undef, <4 x i32> zeroinitializer
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vdup_i16(i16 %src) {
+; CHECK-LABEL: vdup_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vdup.16 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = insertelement <8 x i16> undef, i16 %src, i32 0
+  %out = shufflevector <8 x i16> %0, <8 x i16> undef, <8 x i32> zeroinitializer
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vdup_i8(i8 %src) {
+; CHECK-LABEL: vdup_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vdup.8 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = insertelement <16 x i8> undef, i8 %src, i32 0
+  %out = shufflevector <16 x i8> %0, <16 x i8> undef, <16 x i32> zeroinitializer
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @vdup_f32_1(float %src) {
+; CHECK-LABEL: vdup_f32_1:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov r0, s0
+; CHECK-NEXT:    vdup.32 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = insertelement <4 x float> undef, float %src, i32 0
+  %out = shufflevector <4 x float> %0, <4 x float> undef, <4 x i32> zeroinitializer
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @vdup_f32_2(float %src1, float %src2) {
+; CHECK-LABEL: vdup_f32_2:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vadd.f32 s0, s0, s1
+; CHECK-NEXT:    vmov r0, s0
+; CHECK-NEXT:    vdup.32 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = fadd float %src1, %src2
+  %1 = insertelement <4 x float> undef, float %0, i32 0
+  %out = shufflevector <4 x float> %1, <4 x float> undef, <4 x i32> zeroinitializer
+  ret <4 x float> %out
+}
+
+; TODO: Calling convention needs fixing to pass half types directly to functions
+define arm_aapcs_vfpcc <8 x half> @vdup_f16(half* %src1, half* %src2) {
+; CHECK-LABEL: vdup_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vldr.16 s0, [r1]
+; CHECK-NEXT:    vldr.16 s2, [r0]
+; CHECK-NEXT:    vadd.f16 s0, s2, s0
+; CHECK-NEXT:    vmov r0, s0
+; CHECK-NEXT:    vdup.16 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %0 = load half, half *%src1, align 2
+  %1 = load half, half *%src2, align 2
+  %2 = fadd half %0, %1
+  %3 = insertelement <8 x half> undef, half %2, i32 0
+  %out = shufflevector <8 x half> %3, <8 x half> undef, <8 x i32> zeroinitializer
+  ret <8 x half> %out
+}
+
+
+
+define arm_aapcs_vfpcc <4 x i32> @vduplane_i32(<4 x i32> %src) {
+; CHECK-LABEL: vduplane_i32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.32 r0, q0[3]
+; CHECK-NEXT:    vdup.32 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x i32> %src, <4 x i32> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+  ret <4 x i32> %out
+}
+
+define arm_aapcs_vfpcc <8 x i16> @vduplane_i16(<8 x i16> %src) {
+; CHECK-LABEL: vduplane_i16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[3]
+; CHECK-NEXT:    vdup.16 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x i16> %src, <8 x i16> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+  ret <8 x i16> %out
+}
+
+define arm_aapcs_vfpcc <16 x i8> @vduplane_i8(<16 x i8> %src) {
+; CHECK-LABEL: vduplane_i8:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u8 r0, q0[3]
+; CHECK-NEXT:    vdup.8 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <16 x i8> %src, <16 x i8> undef, <16 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+  ret <16 x i8> %out
+}
+
+define arm_aapcs_vfpcc <4 x float> @vduplane_f32(<4 x float> %src) {
+; CHECK-LABEL: vduplane_f32:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.32 r0, q0[3]
+; CHECK-NEXT:    vdup.32 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <4 x float> %src, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
+  ret <4 x float> %out
+}
+
+define arm_aapcs_vfpcc <8 x half> @vduplane_f16(<8 x half> %src) {
+; CHECK-LABEL: vduplane_f16:
+; CHECK:       @ %bb.0: @ %entry
+; CHECK-NEXT:    vmov.u16 r0, q0[3]
+; CHECK-NEXT:    vdup.16 q0, r0
+; CHECK-NEXT:    bx lr
+entry:
+  %out = shufflevector <8 x half> %src, <8 x half> undef, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
+  ret <8 x half> %out
+}