[SelectionDAG] Add PromoteIntOp_INSERT_SUBVECTOR.

[sdesmalen]

Summary

This is required to codegen something like:

<vscale x 8 x i16> @llvm.experimental.vector.insert(<vscale x 8 x i16> %vec,
                                                    <vscale x 2 x i16> %subvec,
                                                    i64 %idx)

where the output vector is legal, but the input vector needs promoting.

It implements this by performing the whole operation on the promoted type,
and then truncating the result.

[david-arm]

LGTM!

[craig.topper]

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

5089

I have a bit of distrust for changeVectorElementType. If V0.getValueType() is representable as an MVT, but the promoted type isn't, it will fail. It branches a between an MVT code path and EVT path internally. The MVT path would require an LLVMContext to produce an EVT which it can't get so it will assert.

[sdesmalen]

Use EVT::getVectorVT instead of EVT::changeVectorElementType().

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

5089

That's a good point, we should probably change the interface by requiring a Context and then calling getVectorVT in it's implementation, which handles the case of 'Simple' vs 'Extended' correctly. However, that is a big change which is separate to this patch, so I'll just use getVectorVT for now.

[craig.topper]

LGTM

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

bool DAGTypeLegalizer::PromoteIntegerOperand(SDNode *N, unsigned OpNo) {
   case ISD::MSCATTER: Res = PromoteIntOp_MSCATTER(cast<MaskedScatterSDNode>(N),
                                                   OpNo); break;
   case ISD::TRUNCATE:     Res = PromoteIntOp_TRUNCATE(N); break;
   case ISD::FP16_TO_FP:
   case ISD::UINT_TO_FP:   Res = PromoteIntOp_UINT_TO_FP(N); break;
   case ISD::STRICT_UINT_TO_FP:  Res = PromoteIntOp_STRICT_UINT_TO_FP(N); break;
   case ISD::ZERO_EXTEND:  Res = PromoteIntOp_ZERO_EXTEND(N); break;
   case ISD::EXTRACT_SUBVECTOR: Res = PromoteIntOp_EXTRACT_SUBVECTOR(N); break;
+  case ISD::INSERT_SUBVECTOR: Res = PromoteIntOp_INSERT_SUBVECTOR(N); break;
 
   case ISD::SHL:
   case ISD::SRA:
   case ISD::SRL:
   case ISD::ROTL:
   case ISD::ROTR: Res = PromoteIntOp_Shift(N); break;
 
   case ISD::SADDO_CARRY:
SDValue Ext = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, dl,
     V0->getValueType(0).getScalarType(), V0, V1);
 
   // EXTRACT_VECTOR_ELT can return types which are wider than the incoming
   // element types. If this is the case then we need to expand the outgoing
   // value and not truncate it.
   return DAG.getAnyExtOrTrunc(Ext, dl, N->getValueType(0));
 }
 
+SDValue DAGTypeLegalizer::PromoteIntOp_INSERT_SUBVECTOR(SDNode *N) {
+  SDLoc dl(N);
+  // The result type is equal to the first input operand's type, so the
+  // type that needs promoting must be the second source vector.
+  SDValue V0 = N->getOperand(0);
+  SDValue V1 = GetPromotedInteger(N->getOperand(1));
+  SDValue Idx = N->getOperand(2);
+  EVT PromVT = EVT::getVectorVT(*DAG.getContext(),
+                                V1.getValueType().getVectorElementType(),
+                                V0.getValueType().getVectorElementCount());
+  V0 = DAG.getAnyExtOrTrunc(V0, dl, PromVT);
+  SDValue Ext = DAG.getNode(ISD::INSERT_SUBVECTOR, dl, PromVT, V0, V1, Idx);
+  return DAG.getAnyExtOrTrunc(Ext, dl, N->getValueType(0));
+}
+
 SDValue DAGTypeLegalizer::PromoteIntOp_EXTRACT_SUBVECTOR(SDNode *N) {
   SDLoc dl(N);
   SDValue V0 = GetPromotedInteger(N->getOperand(0));
   MVT InVT = V0.getValueType().getSimpleVT();
   MVT OutVT = MVT::getVectorVT(InVT.getVectorElementType(),
                                N->getValueType(0).getVectorNumElements());
   SDValue Ext = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, OutVT, V0, N->getOperand(1));
   return DAG.getNode(ISD::TRUNCATE, dl, N->getValueType(0), Ext);

llvm/lib/CodeGen/SelectionDAG/LegalizeTypes.h

private:
   SDValue PromoteIntOp_BITCAST(SDNode *N);
   SDValue PromoteIntOp_BUILD_PAIR(SDNode *N);
   SDValue PromoteIntOp_BR_CC(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_BRCOND(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_BUILD_VECTOR(SDNode *N);
   SDValue PromoteIntOp_INSERT_VECTOR_ELT(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_EXTRACT_VECTOR_ELT(SDNode *N);
   SDValue PromoteIntOp_EXTRACT_SUBVECTOR(SDNode *N);
+  SDValue PromoteIntOp_INSERT_SUBVECTOR(SDNode *N);
   SDValue PromoteIntOp_CONCAT_VECTORS(SDNode *N);
   SDValue PromoteIntOp_SCALAR_TO_VECTOR(SDNode *N);
   SDValue PromoteIntOp_SPLAT_VECTOR(SDNode *N);
   SDValue PromoteIntOp_SELECT(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_SELECT_CC(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_SETCC(SDNode *N, unsigned OpNo);
   SDValue PromoteIntOp_Shift(SDNode *N);
   SDValue PromoteIntOp_SIGN_EXTEND(SDNode *N);

llvm/test/CodeGen/AArch64/sve-insert-vector.ll

 ; CHECK-NEXT:    ret
 entry:
   %0 = insertelement <vscale x 1 x i16> undef, i16 1, i32 0
   %subvec = shufflevector <vscale x 1 x i16> %0, <vscale x 1 x i16> undef, <vscale x 1 x i32> zeroinitializer
   %retval = call <vscale x 6 x i16> @llvm.experimental.vector.insert.nxv6i16.nxv1i16(<vscale x 6 x i16> undef, <vscale x 1 x i16> %subvec, i64 0)
   ret <vscale x 6 x i16> %retval
 }
 
+; This tests promotion of the input operand to INSERT_SUBVECTOR.
+define <vscale x 8 x i16> @insert_nxv8i16_nxv2i16(<vscale x 8 x i16> %vec, <vscale x 2 x i16> %in) nounwind {
+; CHECK-LABEL: insert_nxv8i16_nxv2i16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    uunpklo z2.s, z0.h
+; CHECK-NEXT:    uunpklo z2.d, z2.s
+; CHECK-NEXT:    uzp1 z1.s, z2.s, z1.s
+; CHECK-NEXT:    uunpkhi z0.s, z0.h
+; CHECK-NEXT:    uzp1 z0.h, z1.h, z0.h
+; CHECK-NEXT:    ret
+  %r = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.nxv2i16(<vscale x 8 x i16> %vec, <vscale x 2 x i16> %in, i64 2)
+  ret <vscale x 8 x i16> %r
+}
+
+
 ; Fixed length clamping
 
 define <vscale x 2 x i64> @insert_fixed_v2i64_nxv2i64(<vscale x 2 x i64> %vec, <2 x i64> %subvec) nounwind #0 {
 ; CHECK-LABEL: insert_fixed_v2i64_nxv2i64:
 ; CHECK:       // %bb.0:
 ; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
 ; CHECK-NEXT:    addvl sp, sp, #-1
 ; CHECK-NEXT:    cntd x9

 declare <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8>, <16 x i8>, i64)
 
 declare <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v4i64(<vscale x 2 x i64>, <4 x i64>, i64)
 
 declare <vscale x 16 x i64> @llvm.experimental.vector.insert.nxv8i64.nxv16i64(<vscale x 16 x i64>, <vscale x 8 x i64>, i64)
 declare <vscale x 16 x i64> @llvm.experimental.vector.insert.v2i64.nxv16i64(<vscale x 16 x i64>, <2 x i64>, i64)
 declare <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.nxv1i32(<vscale x 4 x i32>, <vscale x 1 x i32>, i64)
 declare <vscale x 6 x i16> @llvm.experimental.vector.insert.nxv6i16.nxv1i16(<vscale x 6 x i16>, <vscale x 1 x i16>, i64)
+
+declare <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.nxv2i16(<vscale x 8 x i16>, <vscale x 2 x i16>, i64)