//
//===----------------------------------------------------------------------===//
+
+//===----------------------------------------------------------------------===//
+// NEON-specific Operands.
+//===----------------------------------------------------------------------===//
+def VectorIndex8Operand : AsmOperandClass { let Name = "VectorIndex8"; }
+def VectorIndex16Operand : AsmOperandClass { let Name = "VectorIndex16"; }
+def VectorIndex32Operand : AsmOperandClass { let Name = "VectorIndex32"; }
+def VectorIndex8 : Operand<i32>, ImmLeaf<i32, [{
+ return ((uint64_t)Imm) < 8;
+}]> {
+ let ParserMatchClass = VectorIndex8Operand;
+ let PrintMethod = "printVectorIndex";
+ let MIOperandInfo = (ops i32imm);
+}
+def VectorIndex16 : Operand<i32>, ImmLeaf<i32, [{
+ return ((uint64_t)Imm) < 4;
+}]> {
+ let ParserMatchClass = VectorIndex16Operand;
+ let PrintMethod = "printVectorIndex";
+ let MIOperandInfo = (ops i32imm);
+}
+def VectorIndex32 : Operand<i32>, ImmLeaf<i32, [{
+ return ((uint64_t)Imm) < 2;
+}]> {
+ let ParserMatchClass = VectorIndex32Operand;
+ let PrintMethod = "printVectorIndex";
+ let MIOperandInfo = (ops i32imm);
+}
+
//===----------------------------------------------------------------------===//
// NEON-specific DAG Nodes.
//===----------------------------------------------------------------------===//
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,
: PseudoVFPLdStM<(outs QPR:$dst), (ins GPR:$Rn),
IIC_fpLoad_m, "",
[(set QPR:$dst, (v2f64 (load GPR:$Rn)))]>;
-def VLDMQDB
- : PseudoVFPLdStM<(outs QPR:$dst), (ins GPR:$Rn),
- IIC_fpLoad_m, "",
- [(set QPR:$dst, (v2f64 (load GPR:$Rn)))]>;
// Use VSTM to store a Q register as a D register pair.
// This is a pseudo instruction that is expanded to VSTMD after reg alloc.
: PseudoVFPLdStM<(outs), (ins QPR:$src, GPR:$Rn),
IIC_fpStore_m, "",
[(store (v2f64 QPR:$src), GPR:$Rn)]>;
-def VSTMQDB
- : PseudoVFPLdStM<(outs), (ins QPR:$src, GPR:$Rn),
- IIC_fpStore_m, "",
- [(store (v2f64 QPR:$src), GPR:$Rn)]>;
// Classes for VLD* pseudo-instructions with multi-register operands.
// These are expanded to real instructions after register allocation.
: PseudoNLdSt<(outs QQPR:$dst, GPR:$wb),
(ins addrmode6:$addr, am6offset:$offset), itin,
"$addr.addr = $wb">;
+class VLDQQQQPseudo<InstrItinClass itin>
+ : PseudoNLdSt<(outs QQQQPR:$dst), (ins addrmode6:$addr, QQQQPR:$src),itin,
+ "$src = $dst">;
class VLDQQQQWBPseudo<InstrItinClass itin>
: PseudoNLdSt<(outs QQQQPR:$dst, GPR:$wb),
(ins addrmode6:$addr, am6offset:$offset, QQQQPR:$src), itin,
"vld1", Dt, "\\{$Vd\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD1Q<bits<4> op7_4, string Dt>
: NLdSt<0,0b10,0b1010,op7_4, (outs DPR:$Vd, DPR:$dst2),
"vld1", Dt, "\\{$Vd, $dst2\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD1d8 : VLD1D<{0,0,0,?}, "8">;
"vld1", Dt, "\\{$Vd\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD1QWB<bits<4> op7_4, string Dt>
: NLdSt<0,0b10,0b1010,op7_4, (outs DPR:$Vd, DPR:$dst2, GPR:$wb),
"vld1", Dt, "\\{$Vd, $dst2\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD1d8_UPD : VLD1DWB<{0,0,0,?}, "8">;
"\\{$Vd, $dst2, $dst3\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD1D3WB<bits<4> op7_4, string Dt>
: NLdSt<0,0b10,0b0110,op7_4, (outs DPR:$Vd, DPR:$dst2, DPR:$dst3, GPR:$wb),
(ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD1x3u, "vld1", Dt,
"\\{$Vd, $dst2, $dst3\\}, $Rn$Rm", "$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD1d8T : VLD1D3<{0,0,0,?}, "8">;
"\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD1D4WB<bits<4> op7_4, string Dt>
: NLdSt<0,0b10,0b0010,op7_4,
(outs DPR:$Vd, DPR:$dst2, DPR:$dst3, DPR:$dst4, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD4, "vld1", Dt,
+ (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD1x4u, "vld1", Dt,
"\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn$Rm", "$Rn.addr = $wb",
[]> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD1d8Q : VLD1D4<{0,0,?,?}, "8">;
"vld2", Dt, "\\{$Vd, $dst2\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD2Q<bits<4> op7_4, string Dt>
: NLdSt<0, 0b10, 0b0011, op7_4,
"vld2", Dt, "\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD2d8 : VLD2D<0b1000, {0,0,?,?}, "8">;
"vld2", Dt, "\\{$Vd, $dst2\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
class VLD2QWB<bits<4> op7_4, string Dt>
: NLdSt<0, 0b10, 0b0011, op7_4,
"vld2", Dt, "\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD2d8_UPD : VLD2DWB<0b1000, {0,0,?,?}, "8">;
"vld3", Dt, "\\{$Vd, $dst2, $dst3\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD3d8 : VLD3D<0b0100, {0,0,0,?}, "8">;
"vld3", Dt, "\\{$Vd, $dst2, $dst3\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD3d8_UPD : VLD3DWB<0b0100, {0,0,0,?}, "8">;
def VLD3d16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3u>;
def VLD3d32Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3u>;
-// ...with double-spaced registers (non-updating versions for disassembly only):
+// ...with double-spaced registers:
def VLD3q8 : VLD3D<0b0101, {0,0,0,?}, "8">;
def VLD3q16 : VLD3D<0b0101, {0,1,0,?}, "16">;
def VLD3q32 : VLD3D<0b0101, {1,0,0,?}, "32">;
def VLD3q32Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD3u>;
// ...alternate versions to be allocated odd register numbers:
+def VLD3q8oddPseudo : VLDQQQQPseudo<IIC_VLD3>;
+def VLD3q16oddPseudo : VLDQQQQPseudo<IIC_VLD3>;
+def VLD3q32oddPseudo : VLDQQQQPseudo<IIC_VLD3>;
+
def VLD3q8oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD3u>;
def VLD3q16oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD3u>;
def VLD3q32oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD3u>;
"vld4", Dt, "\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD4d8 : VLD4D<0b0000, {0,0,?,?}, "8">;
class VLD4DWB<bits<4> op11_8, bits<4> op7_4, string Dt>
: NLdSt<0, 0b10, op11_8, op7_4,
(outs DPR:$Vd, DPR:$dst2, DPR:$dst3, DPR:$dst4, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD4,
+ (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD4u,
"vld4", Dt, "\\{$Vd, $dst2, $dst3, $dst4\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVLDInstruction";
}
def VLD4d8_UPD : VLD4DWB<0b0000, {0,0,?,?}, "8">;
def VLD4d16_UPD : VLD4DWB<0b0000, {0,1,?,?}, "16">;
def VLD4d32_UPD : VLD4DWB<0b0000, {1,0,?,?}, "32">;
-def VLD4d8Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4>;
-def VLD4d16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4>;
-def VLD4d32Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4>;
+def VLD4d8Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4u>;
+def VLD4d16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4u>;
+def VLD4d32Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4u>;
-// ...with double-spaced registers (non-updating versions for disassembly only):
+// ...with double-spaced registers:
def VLD4q8 : VLD4D<0b0001, {0,0,?,?}, "8">;
def VLD4q16 : VLD4D<0b0001, {0,1,?,?}, "16">;
def VLD4q32 : VLD4D<0b0001, {1,0,?,?}, "32">;
def VLD4q16_UPD : VLD4DWB<0b0001, {0,1,?,?}, "16">;
def VLD4q32_UPD : VLD4DWB<0b0001, {1,0,?,?}, "32">;
-def VLD4q8Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
-def VLD4q16Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
-def VLD4q32Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
+def VLD4q8Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
+def VLD4q16Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
+def VLD4q32Pseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
// ...alternate versions to be allocated odd register numbers:
-def VLD4q8oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
-def VLD4q16oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
-def VLD4q32oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4>;
+def VLD4q8oddPseudo : VLDQQQQPseudo<IIC_VLD4>;
+def VLD4q16oddPseudo : VLDQQQQPseudo<IIC_VLD4>;
+def VLD4q32oddPseudo : VLDQQQQPseudo<IIC_VLD4>;
+
+def VLD4q8oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
+def VLD4q16oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
+def VLD4q32oddPseudo_UPD : VLDQQQQWBPseudo<IIC_VLD4u>;
} // mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1
(i32 (LoadOp addrmode6:$Rn)),
imm:$lane))]> {
let Rm = 0b1111;
+ let DecoderMethod = "DecodeVLD1LN";
+}
+class VLD1LN32<bits<4> op11_8, bits<4> op7_4, string Dt, ValueType Ty,
+ PatFrag LoadOp>
+ : NLdStLn<1, 0b10, op11_8, op7_4, (outs DPR:$Vd),
+ (ins addrmode6oneL32:$Rn, DPR:$src, nohash_imm:$lane),
+ IIC_VLD1ln, "vld1", Dt, "\\{$Vd[$lane]\\}, $Rn",
+ "$src = $Vd",
+ [(set DPR:$Vd, (vector_insert (Ty DPR:$src),
+ (i32 (LoadOp addrmode6oneL32:$Rn)),
+ imm:$lane))]> {
+ let Rm = 0b1111;
+ let DecoderMethod = "DecodeVLD1LN";
}
class VLD1QLNPseudo<ValueType Ty, PatFrag LoadOp> : VLDQLNPseudo<IIC_VLD1ln> {
let Pattern = [(set QPR:$dst, (vector_insert (Ty QPR:$src),
let Inst{7-6} = lane{1-0};
let Inst{4} = Rn{4};
}
-def VLD1LNd32 : VLD1LN<0b1000, {?,0,?,?}, "32", v2i32, load> {
+def VLD1LNd32 : VLD1LN32<0b1000, {?,0,?,?}, "32", v2i32, load> {
let Inst{7} = lane{0};
let Inst{5} = Rn{4};
let Inst{4} = Rn{4};
def VLD1LNq16Pseudo : VLD1QLNPseudo<v8i16, extloadi16>;
def VLD1LNq32Pseudo : VLD1QLNPseudo<v4i32, load>;
+def : Pat<(vector_insert (v2f32 DPR:$src),
+ (f32 (load addrmode6:$addr)), imm:$lane),
+ (VLD1LNd32 addrmode6:$addr, DPR:$src, imm:$lane)>;
+def : Pat<(vector_insert (v4f32 QPR:$src),
+ (f32 (load addrmode6:$addr)), imm:$lane),
+ (VLD1LNq32Pseudo addrmode6:$addr, QPR:$src, imm:$lane)>;
+
let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
// ...with address register writeback:
(ins addrmode6:$Rn, am6offset:$Rm,
DPR:$src, nohash_imm:$lane), IIC_VLD1lnu, "vld1", Dt,
"\\{$Vd[$lane]\\}, $Rn$Rm",
- "$src = $Vd, $Rn.addr = $wb", []>;
+ "$src = $Vd, $Rn.addr = $wb", []> {
+ let DecoderMethod = "DecodeVLD1LN";
+}
def VLD1LNd8_UPD : VLD1LNWB<0b0000, {?,?,?,0}, "8"> {
let Inst{7-5} = lane{2-0};
"$src1 = $Vd, $src2 = $dst2", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD2LN";
}
def VLD2LNd8 : VLD2LN<0b0001, {?,?,?,?}, "8"> {
"\\{$Vd[$lane], $dst2[$lane]\\}, $Rn$Rm",
"$src1 = $Vd, $src2 = $dst2, $Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD2LN";
}
def VLD2LNd8_UPD : VLD2LNWB<0b0001, {?,?,?,?}, "8"> {
"\\{$Vd[$lane], $dst2[$lane], $dst3[$lane]\\}, $Rn",
"$src1 = $Vd, $src2 = $dst2, $src3 = $dst3", []> {
let Rm = 0b1111;
+ let DecoderMethod = "DecodeVLD3LN";
}
def VLD3LNd8 : VLD3LN<0b0010, {?,?,?,0}, "8"> {
IIC_VLD3lnu, "vld3", Dt,
"\\{$Vd[$lane], $dst2[$lane], $dst3[$lane]\\}, $Rn$Rm",
"$src1 = $Vd, $src2 = $dst2, $src3 = $dst3, $Rn.addr = $wb",
- []>;
+ []> {
+ let DecoderMethod = "DecodeVLD3LN";
+}
def VLD3LNd8_UPD : VLD3LNWB<0b0010, {?,?,?,0}, "8"> {
let Inst{7-5} = lane{2-0};
"$src1 = $Vd, $src2 = $dst2, $src3 = $dst3, $src4 = $dst4", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD4LN";
}
def VLD4LNd8 : VLD4LN<0b0011, {?,?,?,?}, "8"> {
(outs DPR:$Vd, DPR:$dst2, DPR:$dst3, DPR:$dst4, GPR:$wb),
(ins addrmode6:$Rn, am6offset:$Rm,
DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4, nohash_imm:$lane),
- IIC_VLD4ln, "vld4", Dt,
+ IIC_VLD4lnu, "vld4", Dt,
"\\{$Vd[$lane], $dst2[$lane], $dst3[$lane], $dst4[$lane]\\}, $Rn$Rm",
"$src1 = $Vd, $src2 = $dst2, $src3 = $dst3, $src4 = $dst4, $Rn.addr = $wb",
[]> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD4LN" ;
}
def VLD4LNd8_UPD : VLD4LNWB<0b0011, {?,?,?,?}, "8"> {
// VLD1DUP : Vector Load (single element to all lanes)
class VLD1DUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp>
- : NLdSt<1, 0b10, 0b1100, op7_4, (outs DPR:$Vd), (ins addrmode6:$Rn),
+ : NLdSt<1, 0b10, 0b1100, op7_4, (outs DPR:$Vd), (ins addrmode6dup:$Rn),
IIC_VLD1dup, "vld1", Dt, "\\{$Vd[]\\}, $Rn", "",
- [(set DPR:$Vd, (Ty (NEONvdup (i32 (LoadOp addrmode6:$Rn)))))]> {
+ [(set DPR:$Vd, (Ty (NEONvdup (i32 (LoadOp addrmode6dup:$Rn)))))]> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD1DupInstruction";
}
class VLD1QDUPPseudo<ValueType Ty, PatFrag LoadOp> : VLDQPseudo<IIC_VLD1dup> {
let Pattern = [(set QPR:$dst,
- (Ty (NEONvdup (i32 (LoadOp addrmode6:$addr)))))];
+ (Ty (NEONvdup (i32 (LoadOp addrmode6dup:$addr)))))];
}
def VLD1DUPd8 : VLD1DUP<{0,0,0,?}, "8", v8i8, extloadi8>;
def VLD1DUPq16Pseudo : VLD1QDUPPseudo<v8i16, extloadi16>;
def VLD1DUPq32Pseudo : VLD1QDUPPseudo<v4i32, load>;
+def : Pat<(v2f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
+ (VLD1DUPd32 addrmode6:$addr)>;
+def : Pat<(v4f32 (NEONvdup (f32 (load addrmode6dup:$addr)))),
+ (VLD1DUPq32Pseudo addrmode6:$addr)>;
+
let mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1 in {
-class VLD1QDUP<bits<4> op7_4, string Dt, ValueType Ty, PatFrag LoadOp>
+class VLD1QDUP<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1100, op7_4, (outs DPR:$Vd, DPR:$dst2),
- (ins addrmode6:$Rn), IIC_VLD1dup,
+ (ins addrmode6dup:$Rn), IIC_VLD1dup,
"vld1", Dt, "\\{$Vd[], $dst2[]\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD1DupInstruction";
}
-def VLD1DUPq8 : VLD1QDUP<{0,0,1,0}, "8", v16i8, extloadi8>;
-def VLD1DUPq16 : VLD1QDUP<{0,1,1,?}, "16", v8i16, extloadi16>;
-def VLD1DUPq32 : VLD1QDUP<{1,0,1,?}, "32", v4i32, load>;
+def VLD1DUPq8 : VLD1QDUP<{0,0,1,0}, "8">;
+def VLD1DUPq16 : VLD1QDUP<{0,1,1,?}, "16">;
+def VLD1DUPq32 : VLD1QDUP<{1,0,1,?}, "32">;
// ...with address register writeback:
class VLD1DUPWB<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1100, op7_4, (outs DPR:$Vd, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD1dupu,
+ (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD1dupu,
"vld1", Dt, "\\{$Vd[]\\}, $Rn$Rm", "$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD1DupInstruction";
}
class VLD1QDUPWB<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1100, op7_4, (outs DPR:$Vd, DPR:$dst2, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD1dupu,
+ (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD1dupu,
"vld1", Dt, "\\{$Vd[], $dst2[]\\}, $Rn$Rm", "$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD1DupInstruction";
}
def VLD1DUPd8_UPD : VLD1DUPWB<{0,0,0,0}, "8">;
// VLD2DUP : Vector Load (single 2-element structure to all lanes)
class VLD2DUP<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1101, op7_4, (outs DPR:$Vd, DPR:$dst2),
- (ins addrmode6:$Rn), IIC_VLD2dup,
+ (ins addrmode6dup:$Rn), IIC_VLD2dup,
"vld2", Dt, "\\{$Vd[], $dst2[]\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD2DupInstruction";
}
def VLD2DUPd8 : VLD2DUP<{0,0,0,?}, "8">;
def VLD2DUPd32Pseudo : VLDQPseudo<IIC_VLD2dup>;
// ...with double-spaced registers (not used for codegen):
-def VLD2DUPd8Q : VLD2DUP<{0,0,1,?}, "8">;
-def VLD2DUPd16Q : VLD2DUP<{0,1,1,?}, "16">;
-def VLD2DUPd32Q : VLD2DUP<{1,0,1,?}, "32">;
+def VLD2DUPd8x2 : VLD2DUP<{0,0,1,?}, "8">;
+def VLD2DUPd16x2 : VLD2DUP<{0,1,1,?}, "16">;
+def VLD2DUPd32x2 : VLD2DUP<{1,0,1,?}, "32">;
// ...with address register writeback:
class VLD2DUPWB<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1101, op7_4, (outs DPR:$Vd, DPR:$dst2, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD2dupu,
+ (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD2dupu,
"vld2", Dt, "\\{$Vd[], $dst2[]\\}, $Rn$Rm", "$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD2DupInstruction";
}
def VLD2DUPd8_UPD : VLD2DUPWB<{0,0,0,0}, "8">;
def VLD2DUPd16_UPD : VLD2DUPWB<{0,1,0,?}, "16">;
def VLD2DUPd32_UPD : VLD2DUPWB<{1,0,0,?}, "32">;
-def VLD2DUPd8Q_UPD : VLD2DUPWB<{0,0,1,0}, "8">;
-def VLD2DUPd16Q_UPD : VLD2DUPWB<{0,1,1,?}, "16">;
-def VLD2DUPd32Q_UPD : VLD2DUPWB<{1,0,1,?}, "32">;
+def VLD2DUPd8x2_UPD : VLD2DUPWB<{0,0,1,0}, "8">;
+def VLD2DUPd16x2_UPD : VLD2DUPWB<{0,1,1,?}, "16">;
+def VLD2DUPd32x2_UPD : VLD2DUPWB<{1,0,1,?}, "32">;
def VLD2DUPd8Pseudo_UPD : VLDQWBPseudo<IIC_VLD2dupu>;
def VLD2DUPd16Pseudo_UPD : VLDQWBPseudo<IIC_VLD2dupu>;
// VLD3DUP : Vector Load (single 3-element structure to all lanes)
class VLD3DUP<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1110, op7_4, (outs DPR:$Vd, DPR:$dst2, DPR:$dst3),
- (ins addrmode6:$Rn), IIC_VLD3dup,
+ (ins addrmode6dup:$Rn), IIC_VLD3dup,
"vld3", Dt, "\\{$Vd[], $dst2[], $dst3[]\\}, $Rn", "", []> {
let Rm = 0b1111;
- let Inst{4} = Rn{4};
+ let Inst{4} = 0;
+ let DecoderMethod = "DecodeVLD3DupInstruction";
}
def VLD3DUPd8 : VLD3DUP<{0,0,0,?}, "8">;
def VLD3DUPd32Pseudo : VLDQQPseudo<IIC_VLD3dup>;
// ...with double-spaced registers (not used for codegen):
-def VLD3DUPd8T : VLD3DUP<{0,0,1,?}, "8">;
-def VLD3DUPd16T : VLD3DUP<{0,1,1,?}, "16">;
-def VLD3DUPd32T : VLD3DUP<{1,0,1,?}, "32">;
+def VLD3DUPd8x2 : VLD3DUP<{0,0,1,?}, "8">;
+def VLD3DUPd16x2 : VLD3DUP<{0,1,1,?}, "16">;
+def VLD3DUPd32x2 : VLD3DUP<{1,0,1,?}, "32">;
// ...with address register writeback:
class VLD3DUPWB<bits<4> op7_4, string Dt>
: NLdSt<1, 0b10, 0b1110, op7_4, (outs DPR:$Vd, DPR:$dst2, DPR:$dst3, GPR:$wb),
- (ins addrmode6:$Rn, am6offset:$Rm), IIC_VLD3dupu,
+ (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD3dupu,
"vld3", Dt, "\\{$Vd[], $dst2[], $dst3[]\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
- let Inst{4} = Rn{4};
+ let Inst{4} = 0;
+ let DecoderMethod = "DecodeVLD3DupInstruction";
}
def VLD3DUPd8_UPD : VLD3DUPWB<{0,0,0,0}, "8">;
def VLD3DUPd16_UPD : VLD3DUPWB<{0,1,0,?}, "16">;
def VLD3DUPd32_UPD : VLD3DUPWB<{1,0,0,?}, "32">;
-def VLD3DUPd8T_UPD : VLD3DUPWB<{0,0,1,0}, "8">;
-def VLD3DUPd16T_UPD : VLD3DUPWB<{0,1,1,?}, "16">;
-def VLD3DUPd32T_UPD : VLD3DUPWB<{1,0,1,?}, "32">;
+def VLD3DUPd8x2_UPD : VLD3DUPWB<{0,0,1,0}, "8">;
+def VLD3DUPd16x2_UPD : VLD3DUPWB<{0,1,1,?}, "16">;
+def VLD3DUPd32x2_UPD : VLD3DUPWB<{1,0,1,?}, "32">;
def VLD3DUPd8Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3dupu>;
def VLD3DUPd16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3dupu>;
def VLD3DUPd32Pseudo_UPD : VLDQQWBPseudo<IIC_VLD3dupu>;
// VLD4DUP : Vector Load (single 4-element structure to all lanes)
-// FIXME: Not yet implemented.
+class VLD4DUP<bits<4> op7_4, string Dt>
+ : NLdSt<1, 0b10, 0b1111, op7_4,
+ (outs DPR:$Vd, DPR:$dst2, DPR:$dst3, DPR:$dst4),
+ (ins addrmode6dup:$Rn), IIC_VLD4dup,
+ "vld4", Dt, "\\{$Vd[], $dst2[], $dst3[], $dst4[]\\}, $Rn", "", []> {
+ let Rm = 0b1111;
+ let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD4DupInstruction";
+}
+
+def VLD4DUPd8 : VLD4DUP<{0,0,0,?}, "8">;
+def VLD4DUPd16 : VLD4DUP<{0,1,0,?}, "16">;
+def VLD4DUPd32 : VLD4DUP<{1,?,0,?}, "32"> { let Inst{6} = Rn{5}; }
+
+def VLD4DUPd8Pseudo : VLDQQPseudo<IIC_VLD4dup>;
+def VLD4DUPd16Pseudo : VLDQQPseudo<IIC_VLD4dup>;
+def VLD4DUPd32Pseudo : VLDQQPseudo<IIC_VLD4dup>;
+
+// ...with double-spaced registers (not used for codegen):
+def VLD4DUPd8x2 : VLD4DUP<{0,0,1,?}, "8">;
+def VLD4DUPd16x2 : VLD4DUP<{0,1,1,?}, "16">;
+def VLD4DUPd32x2 : VLD4DUP<{1,?,1,?}, "32"> { let Inst{6} = Rn{5}; }
+
+// ...with address register writeback:
+class VLD4DUPWB<bits<4> op7_4, string Dt>
+ : NLdSt<1, 0b10, 0b1111, op7_4,
+ (outs DPR:$Vd, DPR:$dst2, DPR:$dst3, DPR:$dst4, GPR:$wb),
+ (ins addrmode6dup:$Rn, am6offset:$Rm), IIC_VLD4dupu,
+ "vld4", Dt, "\\{$Vd[], $dst2[], $dst3[], $dst4[]\\}, $Rn$Rm",
+ "$Rn.addr = $wb", []> {
+ let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVLD4DupInstruction";
+}
+
+def VLD4DUPd8_UPD : VLD4DUPWB<{0,0,0,0}, "8">;
+def VLD4DUPd16_UPD : VLD4DUPWB<{0,1,0,?}, "16">;
+def VLD4DUPd32_UPD : VLD4DUPWB<{1,?,0,?}, "32"> { let Inst{6} = Rn{5}; }
+
+def VLD4DUPd8x2_UPD : VLD4DUPWB<{0,0,1,0}, "8">;
+def VLD4DUPd16x2_UPD : VLD4DUPWB<{0,1,1,?}, "16">;
+def VLD4DUPd32x2_UPD : VLD4DUPWB<{1,?,1,?}, "32"> { let Inst{6} = Rn{5}; }
+
+def VLD4DUPd8Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4dupu>;
+def VLD4DUPd16Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4dupu>;
+def VLD4DUPd32Pseudo_UPD : VLDQQWBPseudo<IIC_VLD4dupu>;
+
} // mayLoad = 1, neverHasSideEffects = 1, hasExtraDefRegAllocReq = 1
let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in {
: PseudoNLdSt<(outs GPR:$wb),
(ins addrmode6:$addr, am6offset:$offset, QQPR:$src), itin,
"$addr.addr = $wb">;
+class VSTQQQQPseudo<InstrItinClass itin>
+ : PseudoNLdSt<(outs), (ins addrmode6:$addr, QQQQPR:$src), itin, "">;
class VSTQQQQWBPseudo<InstrItinClass itin>
: PseudoNLdSt<(outs GPR:$wb),
(ins addrmode6:$addr, am6offset:$offset, QQQQPR:$src), itin,
IIC_VST1, "vst1", Dt, "\\{$Vd\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST1Q<bits<4> op7_4, string Dt>
: NLdSt<0,0b00,0b1010,op7_4, (outs),
"vst1", Dt, "\\{$Vd, $src2\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST1d8 : VST1D<{0,0,0,?}, "8">;
(ins addrmode6:$Rn, am6offset:$Rm, DPR:$Vd), IIC_VST1u,
"vst1", Dt, "\\{$Vd\\}, $Rn$Rm", "$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST1QWB<bits<4> op7_4, string Dt>
: NLdSt<0, 0b00, 0b1010, op7_4, (outs GPR:$wb),
IIC_VST1x2u, "vst1", Dt, "\\{$Vd, $src2\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST1d8_UPD : VST1DWB<{0,0,0,?}, "8">;
IIC_VST1x3, "vst1", Dt, "\\{$Vd, $src2, $src3\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST1D3WB<bits<4> op7_4, string Dt>
: NLdSt<0, 0b00, 0b0110, op7_4, (outs GPR:$wb),
IIC_VST1x3u, "vst1", Dt, "\\{$Vd, $src2, $src3\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST1d8T : VST1D3<{0,0,0,?}, "8">;
[]> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST1D4WB<bits<4> op7_4, string Dt>
: NLdSt<0, 0b00, 0b0010, op7_4, (outs GPR:$wb),
"vst1", Dt, "\\{$Vd, $src2, $src3, $src4\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST1d8Q : VST1D4<{0,0,?,?}, "8">;
IIC_VST2, "vst2", Dt, "\\{$Vd, $src2\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST2Q<bits<4> op7_4, string Dt>
: NLdSt<0, 0b00, 0b0011, op7_4, (outs),
"", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST2d8 : VST2D<0b1000, {0,0,?,?}, "8">;
IIC_VST2u, "vst2", Dt, "\\{$Vd, $src2\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
class VST2QWB<bits<4> op7_4, string Dt>
: NLdSt<0, 0b00, 0b0011, op7_4, (outs GPR:$wb),
"vst2", Dt, "\\{$Vd, $src2, $src3, $src4\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST2d8_UPD : VST2DWB<0b1000, {0,0,?,?}, "8">;
"vst3", Dt, "\\{$Vd, $src2, $src3\\}, $Rn", "", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST3d8 : VST3D<0b0100, {0,0,0,?}, "8">;
"vst3", Dt, "\\{$Vd, $src2, $src3\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST3d8_UPD : VST3DWB<0b0100, {0,0,0,?}, "8">;
def VST3d16Pseudo_UPD : VSTQQWBPseudo<IIC_VST3u>;
def VST3d32Pseudo_UPD : VSTQQWBPseudo<IIC_VST3u>;
-// ...with double-spaced registers (non-updating versions for disassembly only):
+// ...with double-spaced registers:
def VST3q8 : VST3D<0b0101, {0,0,0,?}, "8">;
def VST3q16 : VST3D<0b0101, {0,1,0,?}, "16">;
def VST3q32 : VST3D<0b0101, {1,0,0,?}, "32">;
def VST3q32Pseudo_UPD : VSTQQQQWBPseudo<IIC_VST3u>;
// ...alternate versions to be allocated odd register numbers:
+def VST3q8oddPseudo : VSTQQQQPseudo<IIC_VST3>;
+def VST3q16oddPseudo : VSTQQQQPseudo<IIC_VST3>;
+def VST3q32oddPseudo : VSTQQQQPseudo<IIC_VST3>;
+
def VST3q8oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST3u>;
def VST3q16oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST3u>;
def VST3q32oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST3u>;
"", []> {
let Rm = 0b1111;
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST4d8 : VST4D<0b0000, {0,0,?,?}, "8">;
"vst4", Dt, "\\{$Vd, $src2, $src3, $src4\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{5-4} = Rn{5-4};
+ let DecoderMethod = "DecodeVSTInstruction";
}
def VST4d8_UPD : VST4DWB<0b0000, {0,0,?,?}, "8">;
def VST4d16Pseudo_UPD : VSTQQWBPseudo<IIC_VST4u>;
def VST4d32Pseudo_UPD : VSTQQWBPseudo<IIC_VST4u>;
-// ...with double-spaced registers (non-updating versions for disassembly only):
+// ...with double-spaced registers:
def VST4q8 : VST4D<0b0001, {0,0,?,?}, "8">;
def VST4q16 : VST4D<0b0001, {0,1,?,?}, "16">;
def VST4q32 : VST4D<0b0001, {1,0,?,?}, "32">;
def VST4q32Pseudo_UPD : VSTQQQQWBPseudo<IIC_VST4u>;
// ...alternate versions to be allocated odd register numbers:
+def VST4q8oddPseudo : VSTQQQQPseudo<IIC_VST4>;
+def VST4q16oddPseudo : VSTQQQQPseudo<IIC_VST4>;
+def VST4q32oddPseudo : VSTQQQQPseudo<IIC_VST4>;
+
def VST4q8oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST4u>;
def VST4q16oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST4u>;
def VST4q32oddPseudo_UPD : VSTQQQQWBPseudo<IIC_VST4u>;
IIC_VST1ln, "vst1", Dt, "\\{$Vd[$lane]\\}, $Rn", "",
[(StoreOp (ExtractOp (Ty DPR:$Vd), imm:$lane), addrmode6:$Rn)]> {
let Rm = 0b1111;
+ let DecoderMethod = "DecodeVST1LN";
+}
+class VST1LN32<bits<4> op11_8, bits<4> op7_4, string Dt, ValueType Ty,
+ PatFrag StoreOp, SDNode ExtractOp>
+ : NLdStLn<1, 0b00, op11_8, op7_4, (outs),
+ (ins addrmode6oneL32:$Rn, DPR:$Vd, nohash_imm:$lane),
+ IIC_VST1ln, "vst1", Dt, "\\{$Vd[$lane]\\}, $Rn", "",
+ [(StoreOp (ExtractOp (Ty DPR:$Vd), imm:$lane), addrmode6oneL32:$Rn)]>{
+ let Rm = 0b1111;
+ let DecoderMethod = "DecodeVST1LN";
}
class VST1QLNPseudo<ValueType Ty, PatFrag StoreOp, SDNode ExtractOp>
: VSTQLNPseudo<IIC_VST1ln> {
let Inst{7-6} = lane{1-0};
let Inst{4} = Rn{5};
}
-def VST1LNd32 : VST1LN<0b1000, {?,0,?,?}, "32", v2i32, store, extractelt> {
+
+def VST1LNd32 : VST1LN32<0b1000, {?,0,?,?}, "32", v2i32, store, extractelt> {
let Inst{7} = lane{0};
let Inst{5-4} = Rn{5-4};
}
def VST1LNq16Pseudo : VST1QLNPseudo<v8i16, truncstorei16, NEONvgetlaneu>;
def VST1LNq32Pseudo : VST1QLNPseudo<v4i32, store, extractelt>;
-let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 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)>;
// ...with address register writeback:
-class VST1LNWB<bits<4> op11_8, bits<4> op7_4, string Dt>
+class VST1LNWB<bits<4> op11_8, bits<4> op7_4, string Dt, ValueType Ty,
+ PatFrag StoreOp, SDNode ExtractOp>
: NLdStLn<1, 0b00, op11_8, op7_4, (outs GPR:$wb),
(ins addrmode6:$Rn, am6offset:$Rm,
DPR:$Vd, nohash_imm:$lane), IIC_VST1lnu, "vst1", Dt,
"\\{$Vd[$lane]\\}, $Rn$Rm",
- "$Rn.addr = $wb", []>;
+ "$Rn.addr = $wb",
+ [(set GPR:$wb, (StoreOp (ExtractOp (Ty DPR:$Vd), imm:$lane),
+ addrmode6:$Rn, am6offset:$Rm))]> {
+ let DecoderMethod = "DecodeVST1LN";
+}
+class VST1QLNWBPseudo<ValueType Ty, PatFrag StoreOp, SDNode ExtractOp>
+ : VSTQLNWBPseudo<IIC_VST1lnu> {
+ let Pattern = [(set GPR:$wb, (StoreOp (ExtractOp (Ty QPR:$src), imm:$lane),
+ addrmode6:$addr, am6offset:$offset))];
+}
-def VST1LNd8_UPD : VST1LNWB<0b0000, {?,?,?,0}, "8"> {
+def VST1LNd8_UPD : VST1LNWB<0b0000, {?,?,?,0}, "8", v8i8, post_truncsti8,
+ NEONvgetlaneu> {
let Inst{7-5} = lane{2-0};
}
-def VST1LNd16_UPD : VST1LNWB<0b0100, {?,?,0,?}, "16"> {
+def VST1LNd16_UPD : VST1LNWB<0b0100, {?,?,0,?}, "16", v4i16, post_truncsti16,
+ NEONvgetlaneu> {
let Inst{7-6} = lane{1-0};
let Inst{4} = Rn{5};
}
-def VST1LNd32_UPD : VST1LNWB<0b1000, {?,0,?,?}, "32"> {
+def VST1LNd32_UPD : VST1LNWB<0b1000, {?,0,?,?}, "32", v2i32, post_store,
+ extractelt> {
let Inst{7} = lane{0};
let Inst{5-4} = Rn{5-4};
}
-def VST1LNq8Pseudo_UPD : VSTQLNWBPseudo<IIC_VST1lnu>;
-def VST1LNq16Pseudo_UPD : VSTQLNWBPseudo<IIC_VST1lnu>;
-def VST1LNq32Pseudo_UPD : VSTQLNWBPseudo<IIC_VST1lnu>;
+def VST1LNq8Pseudo_UPD : VST1QLNWBPseudo<v16i8, post_truncsti8, NEONvgetlaneu>;
+def VST1LNq16Pseudo_UPD : VST1QLNWBPseudo<v8i16, post_truncsti16,NEONvgetlaneu>;
+def VST1LNq32Pseudo_UPD : VST1QLNWBPseudo<v4i32, post_store, extractelt>;
+
+let mayStore = 1, neverHasSideEffects = 1, hasExtraSrcRegAllocReq = 1 in {
// VST2LN : Vector Store (single 2-element structure from one lane)
class VST2LN<bits<4> op11_8, bits<4> op7_4, string Dt>
"", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVST2LN";
}
def VST2LNd8 : VST2LN<0b0001, {?,?,?,?}, "8"> {
"\\{$src1[$lane], $src2[$lane]\\}, $addr$offset",
"$addr.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVST2LN";
}
def VST2LNd8_UPD : VST2LNWB<0b0001, {?,?,?,?}, "8"> {
nohash_imm:$lane), IIC_VST3ln, "vst3", Dt,
"\\{$Vd[$lane], $src2[$lane], $src3[$lane]\\}, $Rn", "", []> {
let Rm = 0b1111;
+ let DecoderMethod = "DecodeVST3LN";
}
def VST3LNd8 : VST3LN<0b0010, {?,?,?,0}, "8"> {
DPR:$Vd, DPR:$src2, DPR:$src3, nohash_imm:$lane),
IIC_VST3lnu, "vst3", Dt,
"\\{$Vd[$lane], $src2[$lane], $src3[$lane]\\}, $Rn$Rm",
- "$Rn.addr = $wb", []>;
+ "$Rn.addr = $wb", []> {
+ let DecoderMethod = "DecodeVST3LN";
+}
def VST3LNd8_UPD : VST3LNWB<0b0010, {?,?,?,0}, "8"> {
let Inst{7-5} = lane{2-0};
"", []> {
let Rm = 0b1111;
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVST4LN";
}
def VST4LNd8 : VST4LN<0b0011, {?,?,?,?}, "8"> {
"\\{$Vd[$lane], $src2[$lane], $src3[$lane], $src4[$lane]\\}, $Rn$Rm",
"$Rn.addr = $wb", []> {
let Inst{4} = Rn{4};
+ let DecoderMethod = "DecodeVST4LN";
}
def VST4LNd8_UPD : VST4LNWB<0b0011, {?,?,?,?}, "8"> {
// Instruction Classes
//===----------------------------------------------------------------------===//
-// Basic 2-register operations: single-, double- and quad-register.
-class N2VS<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
- bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,
- string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4,
- (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src),
- IIC_VUNAD, OpcodeStr, Dt, "$dst, $src", "", []>;
+// Basic 2-register operations: double- and quad-register.
class N2VD<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,
string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$dst),
- (ins DPR:$src), IIC_VUNAD, OpcodeStr, Dt,"$dst, $src", "",
- [(set DPR:$dst, (ResTy (OpNode (OpTy DPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$Vd),
+ (ins DPR:$Vm), IIC_VUNAD, OpcodeStr, Dt,"$Vd, $Vm", "",
+ [(set DPR:$Vd, (ResTy (OpNode (OpTy DPR:$Vm))))]>;
class N2VQ<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,
string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$dst),
- (ins QPR:$src), IIC_VUNAQ, OpcodeStr, Dt,"$dst, $src", "",
- [(set QPR:$dst, (ResTy (OpNode (OpTy QPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$Vd),
+ (ins QPR:$Vm), IIC_VUNAQ, OpcodeStr, Dt,"$Vd, $Vm", "",
+ [(set QPR:$Vd, (ResTy (OpNode (OpTy QPR:$Vm))))]>;
// Basic 2-register intrinsics, both double- and quad-register.
class N2VDInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$dst),
- (ins DPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "",
- [(set DPR:$dst, (ResTy (IntOp (OpTy DPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$Vd),
+ (ins DPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set DPR:$Vd, (ResTy (IntOp (OpTy DPR:$Vm))))]>;
class N2VQInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$dst),
- (ins QPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "",
- [(set QPR:$dst, (ResTy (IntOp (OpTy QPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$Vd),
+ (ins QPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$Vm))))]>;
// Narrow 2-register operations.
class N2VN<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op6, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyD, ValueType TyQ, SDNode OpNode>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs DPR:$dst),
- (ins QPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "",
- [(set DPR:$dst, (TyD (OpNode (TyQ QPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs DPR:$Vd),
+ (ins QPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set DPR:$Vd, (TyD (OpNode (TyQ QPR:$Vm))))]>;
// Narrow 2-register intrinsics.
class N2VNInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op6, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyD, ValueType TyQ, Intrinsic IntOp>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs DPR:$dst),
- (ins QPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "",
- [(set DPR:$dst, (TyD (IntOp (TyQ QPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs DPR:$Vd),
+ (ins QPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set DPR:$Vd, (TyD (IntOp (TyQ QPR:$Vm))))]>;
// Long 2-register operations (currently only used for VMOVL).
class N2VL<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op6, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, SDNode OpNode>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs QPR:$dst),
- (ins DPR:$src), itin, OpcodeStr, Dt, "$dst, $src", "",
- [(set QPR:$dst, (TyQ (OpNode (TyD DPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs QPR:$Vd),
+ (ins DPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set QPR:$Vd, (TyQ (OpNode (TyD DPR:$Vm))))]>;
+
+// Long 2-register intrinsics.
+class N2VLInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
+ bits<2> op17_16, bits<5> op11_7, bit op6, bit op4,
+ InstrItinClass itin, string OpcodeStr, string Dt,
+ ValueType TyQ, ValueType TyD, Intrinsic IntOp>
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, op6, op4, (outs QPR:$Vd),
+ (ins DPR:$Vm), itin, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set QPR:$Vd, (TyQ (IntOp (TyD DPR:$Vm))))]>;
// 2-register shuffles (VTRN/VZIP/VUZP), both double- and quad-register.
class N2VDShuffle<bits<2> op19_18, bits<5> op11_7, string OpcodeStr, string Dt>
- : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 0, 0, (outs DPR:$dst1, DPR:$dst2),
+ : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 0, 0, (outs DPR:$Vd, DPR:$Vm),
(ins DPR:$src1, DPR:$src2), IIC_VPERMD,
- OpcodeStr, Dt, "$dst1, $dst2",
- "$src1 = $dst1, $src2 = $dst2", []>;
+ OpcodeStr, Dt, "$Vd, $Vm",
+ "$src1 = $Vd, $src2 = $Vm", []>;
class N2VQShuffle<bits<2> op19_18, bits<5> op11_7,
InstrItinClass itin, string OpcodeStr, string Dt>
- : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 1, 0, (outs QPR:$dst1, QPR:$dst2),
- (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$dst1, $dst2",
- "$src1 = $dst1, $src2 = $dst2", []>;
-
-// Basic 3-register operations: single-, double- and quad-register.
-class N3VS<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
- string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy,
- SDNode OpNode, bit Commutable>
- : N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src1, DPR_VFP2:$src2), N3RegFrm,
- IIC_VBIND, OpcodeStr, Dt, "$dst, $src1, $src2", "", []> {
- let isCommutable = Commutable;
-}
+ : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 1, 0, (outs QPR:$Vd, QPR:$Vm),
+ (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$Vd, $Vm",
+ "$src1 = $Vd, $src2 = $Vm", []>;
+// Basic 3-register operations: double- and quad-register.
class N3VD<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable>
class N3VDSL<bits<2> op21_20, bits<4> op11_8,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType Ty, SDNode ShOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
- (outs DPR:$dst), (ins DPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (Ty DPR:$dst),
- (Ty (ShOp (Ty DPR:$src1),
- (Ty (NEONvduplane (Ty DPR_VFP2:$src2),imm:$lane)))))]> {
+ : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
+ (outs DPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$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)))))]> {
let isCommutable = 0;
}
class N3VDSL16<bits<2> op21_20, bits<4> op11_8,
string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
- (outs DPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, IIC_VMULi16D, OpcodeStr, Dt,"$dst, $src1, $src2[$lane]","",
- [(set (Ty DPR:$dst),
- (Ty (ShOp (Ty DPR:$src1),
- (Ty (NEONvduplane (Ty DPR_8:$src2), imm:$lane)))))]> {
+ : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
+ (outs DPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, nohash_imm:$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)))))]> {
let isCommutable = 0;
}
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 1, op4,
- (outs QPR:$Qd), (ins QPR:$Qn, QPR:$Qm), N3RegFrm, itin,
- OpcodeStr, Dt, "$Qd, $Qn, $Qm", "",
- [(set QPR:$Qd, (ResTy (OpNode (OpTy QPR:$Qn), (OpTy QPR:$Qm))))]> {
+ (outs QPR:$Vd), (ins QPR:$Vn, QPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (ResTy (OpNode (OpTy QPR:$Vn), (OpTy QPR:$Vm))))]> {
let isCommutable = Commutable;
}
class N3VQX<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
InstrItinClass itin, string OpcodeStr,
ValueType ResTy, ValueType OpTy, SDNode OpNode, bit Commutable>
: N3VX<op24, op23, op21_20, op11_8, 1, op4,
- (outs QPR:$dst), (ins QPR:$src1, QPR:$src2), N3RegFrm, itin,
- OpcodeStr, "$dst, $src1, $src2", "",
- [(set QPR:$dst, (ResTy (OpNode (OpTy QPR:$src1), (OpTy QPR:$src2))))]>{
+ (outs QPR:$Vd), (ins QPR:$Vn, QPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (ResTy (OpNode (OpTy QPR:$Vn), (OpTy QPR:$Vm))))]>{
let isCommutable = Commutable;
}
class N3VQSL<bits<2> op21_20, bits<4> op11_8,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, SDNode ShOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins QPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (ResTy QPR:$dst),
- (ResTy (ShOp (ResTy QPR:$src1),
- (ResTy (NEONvduplane (OpTy DPR_VFP2:$src2),
+ : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins QPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (ShOp (ResTy QPR:$Vn),
+ (ResTy (NEONvduplane (OpTy DPR_VFP2:$Vm),
imm:$lane)))))]> {
let isCommutable = 0;
}
class N3VQSL16<bits<2> op21_20, bits<4> op11_8, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, SDNode ShOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins QPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, IIC_VMULi16Q, OpcodeStr, Dt,"$dst, $src1, $src2[$lane]","",
- [(set (ResTy QPR:$dst),
- (ResTy (ShOp (ResTy QPR:$src1),
- (ResTy (NEONvduplane (OpTy DPR_8:$src2),
+ : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins QPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, IIC_VMULi16Q, OpcodeStr, Dt,"$Vd, $Vn, $Vm[$lane]","",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (ShOp (ResTy QPR:$Vn),
+ (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
imm:$lane)))))]> {
let isCommutable = 0;
}
}
class N3VDIntSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt, ValueType Ty, Intrinsic IntOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
- (outs DPR:$dst), (ins DPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (Ty DPR:$dst),
- (Ty (IntOp (Ty DPR:$src1),
- (Ty (NEONvduplane (Ty DPR_VFP2:$src2),
+ : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
+ (outs DPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (Ty DPR:$Vd),
+ (Ty (IntOp (Ty DPR:$Vn),
+ (Ty (NEONvduplane (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, Intrinsic IntOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
- (outs DPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (Ty DPR:$dst),
- (Ty (IntOp (Ty DPR:$src1),
- (Ty (NEONvduplane (Ty DPR_8:$src2), imm:$lane)))))]> {
+ : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
+ (outs DPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, nohash_imm:$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)))))]> {
let isCommutable = 0;
}
class N3VDIntSh<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
class N3VQIntSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins QPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (ResTy QPR:$dst),
- (ResTy (IntOp (ResTy QPR:$src1),
- (ResTy (NEONvduplane (OpTy DPR_VFP2:$src2),
+ : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins QPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (IntOp (ResTy QPR:$Vn),
+ (ResTy (NEONvduplane (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, Intrinsic IntOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins QPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (ResTy QPR:$dst),
- (ResTy (IntOp (ResTy QPR:$src1),
- (ResTy (NEONvduplane (OpTy DPR_8:$src2),
+ : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins QPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (IntOp (ResTy QPR:$Vn),
+ (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
imm:$lane)))))]> {
let isCommutable = 0;
}
let isCommutable = 0;
}
-// Multiply-Add/Sub operations: single-, double- and quad-register.
-class N3VSMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
- InstrItinClass itin, string OpcodeStr, string Dt,
- ValueType Ty, SDNode MulOp, SDNode OpNode>
- : N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs DPR_VFP2:$dst),
- (ins DPR_VFP2:$src1, DPR_VFP2:$src2, DPR_VFP2:$src3), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3", "$src1 = $dst", []>;
-
+// Multiply-Add/Sub operations: double- and quad-register.
class N3VDMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
- ValueType Ty, SDNode MulOp, SDNode OpNode>
+ ValueType Ty, SDPatternOperator MulOp, SDPatternOperator OpNode>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
(outs DPR:$Vd), (ins DPR:$src1, DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
OpcodeStr, Dt, "$Vd, $Vn, $Vm", "$src1 = $Vd",
class N3VDMulOpSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt,
- ValueType Ty, SDNode MulOp, SDNode ShOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
- (outs DPR:$dst),
- (ins DPR:$src1, DPR:$src2, DPR_VFP2:$src3, nohash_imm:$lane),
+ ValueType Ty, SDPatternOperator MulOp, SDPatternOperator ShOp>
+ : N3VLane32<0, 1, op21_20, op11_8, 1, 0,
+ (outs DPR:$Vd),
+ (ins DPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set (Ty DPR:$dst),
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set (Ty DPR:$Vd),
(Ty (ShOp (Ty DPR:$src1),
- (Ty (MulOp DPR:$src2,
- (Ty (NEONvduplane (Ty DPR_VFP2:$src3),
+ (Ty (MulOp DPR:$Vn,
+ (Ty (NEONvduplane (Ty DPR_VFP2:$Vm),
imm:$lane)))))))]>;
class N3VDMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt,
ValueType Ty, SDNode MulOp, SDNode ShOp>
- : N3V<0, 1, op21_20, op11_8, 1, 0,
+ : N3VLane16<0, 1, op21_20, op11_8, 1, 0,
(outs DPR:$Vd),
(ins DPR:$src1, DPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
class N3VQMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt, ValueType Ty,
- SDNode MulOp, SDNode OpNode>
+ SDPatternOperator MulOp, SDPatternOperator OpNode>
: N3V<op24, op23, op21_20, op11_8, 1, op4,
(outs QPR:$Vd), (ins QPR:$src1, QPR:$Vn, QPR:$Vm), N3RegFrm, itin,
OpcodeStr, Dt, "$Vd, $Vn, $Vm", "$src1 = $Vd",
(Ty (MulOp QPR:$Vn, QPR:$Vm)))))]>;
class N3VQMulOpSL<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy,
- SDNode MulOp, SDNode ShOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst),
- (ins QPR:$src1, QPR:$src2, DPR_VFP2:$src3, nohash_imm:$lane),
+ SDPatternOperator MulOp, SDPatternOperator ShOp>
+ : N3VLane32<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd),
+ (ins QPR:$src1, QPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set (ResTy QPR:$dst),
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set (ResTy QPR:$Vd),
(ResTy (ShOp (ResTy QPR:$src1),
- (ResTy (MulOp QPR:$src2,
- (ResTy (NEONvduplane (OpTy DPR_VFP2:$src3),
+ (ResTy (MulOp QPR:$Vn,
+ (ResTy (NEONvduplane (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,
SDNode MulOp, SDNode ShOp>
- : N3V<1, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst),
- (ins QPR:$src1, QPR:$src2, DPR_8:$src3, nohash_imm:$lane),
+ : N3VLane16<1, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd),
+ (ins QPR:$src1, QPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set (ResTy QPR:$dst),
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set (ResTy QPR:$Vd),
(ResTy (ShOp (ResTy QPR:$src1),
- (ResTy (MulOp QPR:$src2,
- (ResTy (NEONvduplane (OpTy DPR_8:$src3),
+ (ResTy (MulOp QPR:$Vn,
+ (ResTy (NEONvduplane (OpTy DPR_8:$Vm),
imm:$lane)))))))]>;
// Neon Intrinsic-Op instructions (VABA): double- and quad-register.
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs DPR:$dst), (ins DPR:$src1, DPR:$src2, DPR:$src3), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3", "$src1 = $dst",
- [(set DPR:$dst, (ResTy (IntOp (OpTy DPR:$src1),
- (OpTy DPR:$src2), (OpTy DPR:$src3))))]>;
+ (outs DPR:$Vd), (ins DPR:$src1, DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "$src1 = $Vd",
+ [(set DPR:$Vd, (ResTy (IntOp (OpTy DPR:$src1),
+ (OpTy DPR:$Vn), (OpTy DPR:$Vm))))]>;
class N3VQInt3<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
: N3V<op24, op23, op21_20, op11_8, 1, op4,
- (outs QPR:$dst), (ins QPR:$src1, QPR:$src2, QPR:$src3), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3", "$src1 = $dst",
- [(set QPR:$dst, (ResTy (IntOp (OpTy QPR:$src1),
- (OpTy QPR:$src2), (OpTy QPR:$src3))))]>;
+ (outs QPR:$Vd), (ins QPR:$src1, QPR:$Vn, QPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "$src1 = $Vd",
+ [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$src1),
+ (OpTy QPR:$Vn), (OpTy QPR:$Vm))))]>;
// Long Multiply-Add/Sub operations.
class N3VLMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
class N3VLMulOpSL<bit op24, bits<2> op21_20, bits<4> op11_8,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, SDNode MulOp, SDNode OpNode>
- : N3V<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$dst),
- (ins QPR:$src1, DPR:$src2, DPR_VFP2:$src3, nohash_imm:$lane),
+ : N3VLane32<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$Vd),
+ (ins QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set QPR:$dst,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set QPR:$Vd,
(OpNode (TyQ QPR:$src1),
- (TyQ (MulOp (TyD DPR:$src2),
- (TyD (NEONvduplane (TyD DPR_VFP2:$src3),
+ (TyQ (MulOp (TyD DPR:$Vn),
+ (TyD (NEONvduplane (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>
- : N3V<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$dst),
- (ins QPR:$src1, DPR:$src2, DPR_8:$src3, nohash_imm:$lane),
+ : N3VLane16<op24, 1, op21_20, op11_8, 1, 0, (outs QPR:$Vd),
+ (ins QPR:$src1, DPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set QPR:$dst,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set QPR:$Vd,
(OpNode (TyQ QPR:$src1),
- (TyQ (MulOp (TyD DPR:$src2),
- (TyD (NEONvduplane (TyD DPR_8:$src3),
+ (TyQ (MulOp (TyD DPR:$Vn),
+ (TyD (NEONvduplane (TyD DPR_8:$Vm),
imm:$lane))))))]>;
// Long Intrinsic-Op vector operations with explicit extend (VABAL).
class N3VLInt3SL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst),
- (ins QPR:$src1, DPR:$src2, DPR_VFP2:$src3, nohash_imm:$lane),
+ : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd),
+ (ins QPR:$src1, DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set (ResTy QPR:$dst),
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set (ResTy QPR:$Vd),
(ResTy (IntOp (ResTy QPR:$src1),
- (OpTy DPR:$src2),
- (OpTy (NEONvduplane (OpTy DPR_VFP2:$src3),
+ (OpTy DPR:$Vn),
+ (OpTy (NEONvduplane (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, Intrinsic IntOp>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst),
- (ins QPR:$src1, DPR:$src2, DPR_8:$src3, nohash_imm:$lane),
+ : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd),
+ (ins QPR:$src1, DPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
NVMulSLFrm, itin,
- OpcodeStr, Dt, "$dst, $src2, $src3[$lane]", "$src1 = $dst",
- [(set (ResTy QPR:$dst),
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "$src1 = $Vd",
+ [(set (ResTy QPR:$Vd),
(ResTy (IntOp (ResTy QPR:$src1),
- (OpTy DPR:$src2),
- (OpTy (NEONvduplane (OpTy DPR_8:$src3),
+ (OpTy DPR:$Vn),
+ (OpTy (NEONvduplane (OpTy DPR_8:$Vm),
imm:$lane)))))]>;
// Narrowing 3-register intrinsics.
string OpcodeStr, string Dt, ValueType TyD, ValueType TyQ,
Intrinsic IntOp, bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs DPR:$dst), (ins QPR:$src1, QPR:$src2), N3RegFrm, IIC_VBINi4D,
- OpcodeStr, Dt, "$dst, $src1, $src2", "",
- [(set DPR:$dst, (TyD (IntOp (TyQ QPR:$src1), (TyQ QPR:$src2))))]> {
+ (outs DPR:$Vd), (ins QPR:$Vn, QPR:$Vm), N3RegFrm, IIC_VBINi4D,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set DPR:$Vd, (TyD (IntOp (TyQ QPR:$Vn), (TyQ QPR:$Vm))))]> {
let isCommutable = Commutable;
}
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, SDNode OpNode, bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs QPR:$dst), (ins DPR:$src1, DPR:$src2), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src1, $src2", "",
- [(set QPR:$dst, (TyQ (OpNode (TyD DPR:$src1), (TyD DPR:$src2))))]> {
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (TyQ (OpNode (TyD DPR:$Vn), (TyD DPR:$Vm))))]> {
let isCommutable = Commutable;
}
class N3VLSL<bit op24, bits<2> op21_20, bits<4> op11_8,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, SDNode OpNode>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins DPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set QPR:$dst,
- (TyQ (OpNode (TyD DPR:$src1),
- (TyD (NEONvduplane (TyD DPR_VFP2:$src2),imm:$lane)))))]>;
+ : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$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)))))]>;
class N3VLSL16<bit op24, bits<2> op21_20, bits<4> op11_8,
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, SDNode OpNode>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set QPR:$dst,
- (TyQ (OpNode (TyD DPR:$src1),
- (TyD (NEONvduplane (TyD DPR_8:$src2), imm:$lane)))))]>;
+ : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, nohash_imm:$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)))))]>;
// Long 3-register operations with explicitly extended operands.
class N3VLExt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
ValueType TyQ, ValueType TyD, SDNode OpNode, SDNode ExtOp,
bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs QPR:$Qd), (ins DPR:$Dn, DPR:$Dm), N3RegFrm, itin,
- OpcodeStr, Dt, "$Qd, $Dn, $Dm", "",
- [(set QPR:$Qd, (OpNode (TyQ (ExtOp (TyD DPR:$Dn))),
- (TyQ (ExtOp (TyD DPR:$Dm)))))]> {
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (OpNode (TyQ (ExtOp (TyD DPR:$Vn))),
+ (TyQ (ExtOp (TyD DPR:$Vm)))))]> {
let isCommutable = Commutable;
}
ValueType TyQ, ValueType TyD, Intrinsic IntOp, SDNode ExtOp,
bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs QPR:$dst), (ins DPR:$src1, DPR:$src2), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src1, $src2", "",
- [(set QPR:$dst, (TyQ (ExtOp (TyD (IntOp (TyD DPR:$src1),
- (TyD DPR:$src2))))))]> {
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (TyQ (ExtOp (TyD (IntOp (TyD DPR:$Vn),
+ (TyD DPR:$Vm))))))]> {
let isCommutable = Commutable;
}
InstrItinClass itin, string OpcodeStr, string Dt,
ValueType TyQ, ValueType TyD, Intrinsic IntOp, bit Commutable>
: N3V<op24, op23, op21_20, op11_8, 0, op4,
- (outs QPR:$dst), (ins DPR:$src1, DPR:$src2), N3RegFrm, itin,
- OpcodeStr, Dt, "$dst, $src1, $src2", "",
- [(set QPR:$dst, (TyQ (IntOp (TyD DPR:$src1), (TyD DPR:$src2))))]> {
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR:$Vm), N3RegFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (TyQ (IntOp (TyD DPR:$Vn), (TyD DPR:$Vm))))]> {
let isCommutable = Commutable;
}
class N3VLIntSL<bit op24, bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins DPR:$src1, DPR_VFP2:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (ResTy QPR:$dst),
- (ResTy (IntOp (OpTy DPR:$src1),
- (OpTy (NEONvduplane (OpTy DPR_VFP2:$src2),
+ : N3VLane32<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR_VFP2:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (IntOp (OpTy DPR:$Vn),
+ (OpTy (NEONvduplane (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, Intrinsic IntOp>
- : N3V<op24, 1, op21_20, op11_8, 1, 0,
- (outs QPR:$dst), (ins DPR:$src1, DPR_8:$src2, nohash_imm:$lane),
- NVMulSLFrm, itin, OpcodeStr, Dt, "$dst, $src1, $src2[$lane]", "",
- [(set (ResTy QPR:$dst),
- (ResTy (IntOp (OpTy DPR:$src1),
- (OpTy (NEONvduplane (OpTy DPR_8:$src2),
+ : N3VLane16<op24, 1, op21_20, op11_8, 1, 0,
+ (outs QPR:$Vd), (ins DPR:$Vn, DPR_8:$Vm, nohash_imm:$lane),
+ NVMulSLFrm, itin, OpcodeStr, Dt, "$Vd, $Vn, $Vm[$lane]", "",
+ [(set (ResTy QPR:$Vd),
+ (ResTy (IntOp (OpTy DPR:$Vn),
+ (OpTy (NEONvduplane (OpTy DPR_8:$Vm),
imm:$lane)))))]>;
// Wide 3-register operations.
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:$Qd), (ins QPR:$Qn, DPR:$Dm), N3RegFrm, IIC_VSUBiD,
- OpcodeStr, Dt, "$Qd, $Qn, $Dm", "",
- [(set QPR:$Qd, (OpNode (TyQ QPR:$Qn),
- (TyQ (ExtOp (TyD DPR:$Dm)))))]> {
+ (outs QPR:$Vd), (ins QPR:$Vn, DPR:$Vm), N3RegFrm, IIC_VSUBiD,
+ OpcodeStr, Dt, "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (OpNode (TyQ QPR:$Vn),
+ (TyQ (ExtOp (TyD DPR:$Vm)))))]> {
let isCommutable = Commutable;
}
bits<2> op17_16, bits<5> op11_7, bit op4,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$dst),
- (ins DPR:$src), IIC_VSHLiD, OpcodeStr, Dt, "$dst, $src", "",
- [(set DPR:$dst, (ResTy (IntOp (OpTy DPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 0, op4, (outs DPR:$Vd),
+ (ins DPR:$Vm), IIC_VSHLiD, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set DPR:$Vd, (ResTy (IntOp (OpTy DPR:$Vm))))]>;
class N2VQPLInt<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
bits<2> op17_16, bits<5> op11_7, bit op4,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
- : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$dst),
- (ins QPR:$src), IIC_VSHLiD, OpcodeStr, Dt, "$dst, $src", "",
- [(set QPR:$dst, (ResTy (IntOp (OpTy QPR:$src))))]>;
+ : N2V<op24_23, op21_20, op19_18, op17_16, op11_7, 1, op4, (outs QPR:$Vd),
+ (ins QPR:$Vm), IIC_VSHLiD, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set QPR:$Vd, (ResTy (IntOp (OpTy QPR:$Vm))))]>;
// Pairwise long 2-register accumulate intrinsics,
// both double- and quad-register.
// Shift by immediate,
// both double- and quad-register.
class N2VDSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- Format f, InstrItinClass itin, string OpcodeStr, string Dt,
- ValueType Ty, SDNode OpNode>
+ Format f, InstrItinClass itin, Operand ImmTy,
+ string OpcodeStr, string Dt, ValueType Ty, SDNode OpNode>
: N2VImm<op24, op23, op11_8, op7, 0, op4,
- (outs DPR:$dst), (ins DPR:$src, i32imm:$SIMM), f, itin,
- OpcodeStr, Dt, "$dst, $src, $SIMM", "",
- [(set DPR:$dst, (Ty (OpNode (Ty DPR:$src), (i32 imm:$SIMM))))]>;
+ (outs DPR:$Vd), (ins DPR:$Vm, ImmTy:$SIMM), f, itin,
+ OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
+ [(set DPR:$Vd, (Ty (OpNode (Ty DPR:$Vm), (i32 imm:$SIMM))))]>;
class N2VQSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- Format f, InstrItinClass itin, string OpcodeStr, string Dt,
- ValueType Ty, SDNode OpNode>
+ Format f, InstrItinClass itin, Operand ImmTy,
+ string OpcodeStr, string Dt, ValueType Ty, SDNode OpNode>
: N2VImm<op24, op23, op11_8, op7, 1, op4,
- (outs QPR:$dst), (ins QPR:$src, i32imm:$SIMM), f, itin,
- OpcodeStr, Dt, "$dst, $src, $SIMM", "",
- [(set QPR:$dst, (Ty (OpNode (Ty QPR:$src), (i32 imm:$SIMM))))]>;
+ (outs QPR:$Vd), (ins QPR:$Vm, ImmTy:$SIMM), f, itin,
+ OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
+ [(set QPR:$Vd, (Ty (OpNode (Ty QPR:$Vm), (i32 imm:$SIMM))))]>;
// Long shift by immediate.
class N2VLSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
string OpcodeStr, string Dt,
ValueType ResTy, ValueType OpTy, SDNode OpNode>
: N2VImm<op24, op23, op11_8, op7, op6, op4,
- (outs QPR:$dst), (ins DPR:$src, i32imm:$SIMM), N2RegVShLFrm,
- IIC_VSHLiD, OpcodeStr, Dt, "$dst, $src, $SIMM", "",
- [(set QPR:$dst, (ResTy (OpNode (OpTy DPR:$src),
+ (outs QPR:$Vd), (ins DPR:$Vm, i32imm:$SIMM), N2RegVShLFrm,
+ IIC_VSHLiD, OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
+ [(set QPR:$Vd, (ResTy (OpNode (OpTy DPR:$Vm),
(i32 imm:$SIMM))))]>;
// Narrow shift by immediate.
class N2VNSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
InstrItinClass itin, string OpcodeStr, string Dt,
- ValueType ResTy, ValueType OpTy, SDNode OpNode>
+ ValueType ResTy, ValueType OpTy, Operand ImmTy, SDNode OpNode>
: N2VImm<op24, op23, op11_8, op7, op6, op4,
- (outs DPR:$dst), (ins QPR:$src, i32imm:$SIMM), N2RegVShRFrm, itin,
- OpcodeStr, Dt, "$dst, $src, $SIMM", "",
- [(set DPR:$dst, (ResTy (OpNode (OpTy QPR:$src),
+ (outs DPR:$Vd), (ins QPR:$Vm, ImmTy:$SIMM), N2RegVShRFrm, itin,
+ OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "",
+ [(set DPR:$Vd, (ResTy (OpNode (OpTy QPR:$Vm),
(i32 imm:$SIMM))))]>;
// Shift right by immediate and accumulate,
// both double- and quad-register.
class N2VDShAdd<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp>
+ Operand ImmTy, string OpcodeStr, string Dt,
+ ValueType Ty, SDNode ShOp>
: N2VImm<op24, op23, op11_8, op7, 0, op4, (outs DPR:$Vd),
- (ins DPR:$src1, DPR:$Vm, i32imm:$SIMM), N2RegVShRFrm, IIC_VPALiD,
+ (ins DPR:$src1, DPR:$Vm, ImmTy:$SIMM), N2RegVShRFrm, IIC_VPALiD,
OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "$src1 = $Vd",
[(set DPR:$Vd, (Ty (add DPR:$src1,
(Ty (ShOp DPR:$Vm, (i32 imm:$SIMM))))))]>;
class N2VQShAdd<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp>
+ Operand ImmTy, string OpcodeStr, string Dt,
+ ValueType Ty, SDNode ShOp>
: N2VImm<op24, op23, op11_8, op7, 1, op4, (outs QPR:$Vd),
- (ins QPR:$src1, QPR:$Vm, i32imm:$SIMM), N2RegVShRFrm, IIC_VPALiD,
+ (ins QPR:$src1, QPR:$Vm, ImmTy:$SIMM), N2RegVShRFrm, IIC_VPALiD,
OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "$src1 = $Vd",
[(set QPR:$Vd, (Ty (add QPR:$src1,
(Ty (ShOp QPR:$Vm, (i32 imm:$SIMM))))))]>;
// Shift by immediate and insert,
// both double- and quad-register.
class N2VDShIns<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- Format f, string OpcodeStr, string Dt, ValueType Ty,SDNode ShOp>
+ Operand ImmTy, Format f, string OpcodeStr, string Dt,
+ ValueType Ty,SDNode ShOp>
: N2VImm<op24, op23, op11_8, op7, 0, op4, (outs DPR:$Vd),
- (ins DPR:$src1, DPR:$Vm, i32imm:$SIMM), f, IIC_VSHLiD,
+ (ins DPR:$src1, DPR:$Vm, ImmTy:$SIMM), f, IIC_VSHLiD,
OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "$src1 = $Vd",
[(set DPR:$Vd, (Ty (ShOp DPR:$src1, DPR:$Vm, (i32 imm:$SIMM))))]>;
class N2VQShIns<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
- Format f, string OpcodeStr, string Dt, ValueType Ty,SDNode ShOp>
+ Operand ImmTy, Format f, string OpcodeStr, string Dt,
+ ValueType Ty,SDNode ShOp>
: N2VImm<op24, op23, op11_8, op7, 1, op4, (outs QPR:$Vd),
- (ins QPR:$src1, QPR:$Vm, i32imm:$SIMM), f, IIC_VSHLiQ,
+ (ins QPR:$src1, QPR:$Vm, ImmTy:$SIMM), f, IIC_VSHLiQ,
OpcodeStr, Dt, "$Vd, $Vm, $SIMM", "$src1 = $Vd",
[(set QPR:$Vd, (Ty (ShOp QPR:$src1, QPR:$Vm, (i32 imm:$SIMM))))]>;
// S = single int (32 bit) elements
// D = double int (64 bit) elements
-// Neon 2-register vector operations -- for disassembly only.
+// Neon 2-register vector operations and intrinsics.
-// First with only element sizes of 8, 16 and 32 bits:
+// Neon 2-register comparisons.
+// source operand element sizes of 8, 16 and 32 bits:
multiclass N2V_QHS_cmp<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
bits<5> op11_7, bit op4, string opc, string Dt,
string asm, SDNode OpNode> {
// 64-bit vector types.
def v8i8 : N2V<op24_23, op21_20, 0b00, op17_16, op11_7, 0, op4,
- (outs DPR:$dst), (ins DPR:$src), NoItinerary,
+ (outs DPR:$Vd), (ins DPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "8"), asm, "",
- [(set DPR:$dst, (v8i8 (OpNode (v8i8 DPR:$src))))]>;
+ [(set DPR:$Vd, (v8i8 (OpNode (v8i8 DPR:$Vm))))]>;
def v4i16 : N2V<op24_23, op21_20, 0b01, op17_16, op11_7, 0, op4,
- (outs DPR:$dst), (ins DPR:$src), NoItinerary,
+ (outs DPR:$Vd), (ins DPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "16"), asm, "",
- [(set DPR:$dst, (v4i16 (OpNode (v4i16 DPR:$src))))]>;
+ [(set DPR:$Vd, (v4i16 (OpNode (v4i16 DPR:$Vm))))]>;
def v2i32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 0, op4,
- (outs DPR:$dst), (ins DPR:$src), NoItinerary,
+ (outs DPR:$Vd), (ins DPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "32"), asm, "",
- [(set DPR:$dst, (v2i32 (OpNode (v2i32 DPR:$src))))]>;
+ [(set DPR:$Vd, (v2i32 (OpNode (v2i32 DPR:$Vm))))]>;
def v2f32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 0, op4,
- (outs DPR:$dst), (ins DPR:$src), NoItinerary,
+ (outs DPR:$Vd), (ins DPR:$Vm), NoItinerary,
opc, "f32", asm, "",
- [(set DPR:$dst, (v2f32 (OpNode (v2f32 DPR:$src))))]> {
+ [(set DPR:$Vd, (v2i32 (OpNode (v2f32 DPR:$Vm))))]> {
let Inst{10} = 1; // overwrite F = 1
}
// 128-bit vector types.
def v16i8 : N2V<op24_23, op21_20, 0b00, op17_16, op11_7, 1, op4,
- (outs QPR:$dst), (ins QPR:$src), NoItinerary,
+ (outs QPR:$Vd), (ins QPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "8"), asm, "",
- [(set QPR:$dst, (v16i8 (OpNode (v16i8 QPR:$src))))]>;
+ [(set QPR:$Vd, (v16i8 (OpNode (v16i8 QPR:$Vm))))]>;
def v8i16 : N2V<op24_23, op21_20, 0b01, op17_16, op11_7, 1, op4,
- (outs QPR:$dst), (ins QPR:$src), NoItinerary,
+ (outs QPR:$Vd), (ins QPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "16"), asm, "",
- [(set QPR:$dst, (v8i16 (OpNode (v8i16 QPR:$src))))]>;
+ [(set QPR:$Vd, (v8i16 (OpNode (v8i16 QPR:$Vm))))]>;
def v4i32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 1, op4,
- (outs QPR:$dst), (ins QPR:$src), NoItinerary,
+ (outs QPR:$Vd), (ins QPR:$Vm), NoItinerary,
opc, !strconcat(Dt, "32"), asm, "",
- [(set QPR:$dst, (v4i32 (OpNode (v4i32 QPR:$src))))]>;
+ [(set QPR:$Vd, (v4i32 (OpNode (v4i32 QPR:$Vm))))]>;
def v4f32 : N2V<op24_23, op21_20, 0b10, op17_16, op11_7, 1, op4,
- (outs QPR:$dst), (ins QPR:$src), NoItinerary,
+ (outs QPR:$Vd), (ins QPR:$Vm), NoItinerary,
opc, "f32", asm, "",
- [(set QPR:$dst, (v4f32 (OpNode (v4f32 QPR:$src))))]> {
+ [(set QPR:$Vd, (v4i32 (OpNode (v4f32 QPR:$Vm))))]> {
let Inst{10} = 1; // overwrite F = 1
}
}
+
+// Neon 2-register vector intrinsics,
+// element sizes of 8, 16 and 32 bits:
+multiclass N2VInt_QHS<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
+ bits<5> op11_7, bit op4,
+ InstrItinClass itinD, InstrItinClass itinQ,
+ string OpcodeStr, string Dt, Intrinsic IntOp> {
+ // 64-bit vector types.
+ def v8i8 : N2VDInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4,
+ itinD, OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>;
+ def v4i16 : N2VDInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4,
+ itinD, OpcodeStr, !strconcat(Dt, "16"),v4i16,v4i16,IntOp>;
+ def v2i32 : N2VDInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4,
+ itinD, OpcodeStr, !strconcat(Dt, "32"),v2i32,v2i32,IntOp>;
+
+ // 128-bit vector types.
+ def v16i8 : N2VQInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4,
+ itinQ, OpcodeStr, !strconcat(Dt, "8"), v16i8,v16i8,IntOp>;
+ def v8i16 : N2VQInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4,
+ itinQ, OpcodeStr, !strconcat(Dt, "16"),v8i16,v8i16,IntOp>;
+ def v4i32 : N2VQInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4,
+ itinQ, OpcodeStr, !strconcat(Dt, "32"),v4i32,v4i32,IntOp>;
+}
+
+
+// Neon Narrowing 2-register vector operations,
+// source operand element sizes of 16, 32 and 64 bits:
+multiclass N2VN_HSD<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
+ bits<5> op11_7, bit op6, bit op4,
+ InstrItinClass itin, string OpcodeStr, string Dt,
+ SDNode OpNode> {
+ def v8i8 : N2VN<op24_23, op21_20, 0b00, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "16"),
+ v8i8, v8i16, OpNode>;
+ def v4i16 : N2VN<op24_23, op21_20, 0b01, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "32"),
+ v4i16, v4i32, OpNode>;
+ def v2i32 : N2VN<op24_23, op21_20, 0b10, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "64"),
+ v2i32, v2i64, OpNode>;
+}
+
+// Neon Narrowing 2-register vector intrinsics,
+// source operand element sizes of 16, 32 and 64 bits:
+multiclass N2VNInt_HSD<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
+ bits<5> op11_7, bit op6, bit op4,
+ InstrItinClass itin, string OpcodeStr, string Dt,
+ Intrinsic IntOp> {
+ def v8i8 : N2VNInt<op24_23, op21_20, 0b00, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "16"),
+ v8i8, v8i16, IntOp>;
+ def v4i16 : N2VNInt<op24_23, op21_20, 0b01, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "32"),
+ v4i16, v4i32, IntOp>;
+ def v2i32 : N2VNInt<op24_23, op21_20, 0b10, op17_16, op11_7, op6, op4,
+ itin, OpcodeStr, !strconcat(Dt, "64"),
+ v2i32, v2i64, IntOp>;
+}
+
+
+// Neon Lengthening 2-register vector intrinsic (currently specific to VMOVL).
+// source operand element sizes of 16, 32 and 64 bits:
+multiclass N2VL_QHS<bits<2> op24_23, bits<5> op11_7, bit op6, bit op4,
+ string OpcodeStr, string Dt, SDNode OpNode> {
+ def v8i16 : N2VL<op24_23, 0b00, 0b10, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
+ OpcodeStr, !strconcat(Dt, "8"), v8i16, v8i8, OpNode>;
+ def v4i32 : N2VL<op24_23, 0b01, 0b00, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
+ OpcodeStr, !strconcat(Dt, "16"), v4i32, v4i16, OpNode>;
+ def v2i64 : N2VL<op24_23, 0b10, 0b00, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
+ OpcodeStr, !strconcat(Dt, "32"), v2i64, v2i32, OpNode>;
+}
+
+
// Neon 3-register vector operations.
// First with only element sizes of 8, 16 and 32 bits:
}
-// Neon Narrowing 2-register vector operations,
-// source operand element sizes of 16, 32 and 64 bits:
-multiclass N2VN_HSD<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
- bits<5> op11_7, bit op6, bit op4,
- InstrItinClass itin, string OpcodeStr, string Dt,
- SDNode OpNode> {
- def v8i8 : N2VN<op24_23, op21_20, 0b00, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "16"),
- v8i8, v8i16, OpNode>;
- def v4i16 : N2VN<op24_23, op21_20, 0b01, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "32"),
- v4i16, v4i32, OpNode>;
- def v2i32 : N2VN<op24_23, op21_20, 0b10, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "64"),
- v2i32, v2i64, OpNode>;
-}
-
-// Neon Narrowing 2-register vector intrinsics,
-// source operand element sizes of 16, 32 and 64 bits:
-multiclass N2VNInt_HSD<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
- bits<5> op11_7, bit op6, bit op4,
- InstrItinClass itin, string OpcodeStr, string Dt,
- Intrinsic IntOp> {
- def v8i8 : N2VNInt<op24_23, op21_20, 0b00, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "16"),
- v8i8, v8i16, IntOp>;
- def v4i16 : N2VNInt<op24_23, op21_20, 0b01, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "32"),
- v4i16, v4i32, IntOp>;
- def v2i32 : N2VNInt<op24_23, op21_20, 0b10, op17_16, op11_7, op6, op4,
- itin, OpcodeStr, !strconcat(Dt, "64"),
- v2i32, v2i64, IntOp>;
-}
-
-
-// Neon Lengthening 2-register vector intrinsic (currently specific to VMOVL).
-// source operand element sizes of 16, 32 and 64 bits:
-multiclass N2VL_QHS<bits<2> op24_23, bits<5> op11_7, bit op6, bit op4,
- string OpcodeStr, string Dt, SDNode OpNode> {
- def v8i16 : N2VL<op24_23, 0b00, 0b10, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
- OpcodeStr, !strconcat(Dt, "8"), v8i16, v8i8, OpNode>;
- def v4i32 : N2VL<op24_23, 0b01, 0b00, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
- OpcodeStr, !strconcat(Dt, "16"), v4i32, v4i16, OpNode>;
- def v2i64 : N2VL<op24_23, 0b10, 0b00, 0b00, op11_7, op6, op4, IIC_VQUNAiD,
- OpcodeStr, !strconcat(Dt, "32"), v2i64, v2i32, OpNode>;
-}
-
-
// Neon 3-register vector intrinsics.
// First with only element sizes of 16 and 32 bits:
}
-// Neon 2-register vector intrinsics,
-// element sizes of 8, 16 and 32 bits:
-multiclass N2VInt_QHS<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
- bits<5> op11_7, bit op4,
- InstrItinClass itinD, InstrItinClass itinQ,
- string OpcodeStr, string Dt, Intrinsic IntOp> {
- // 64-bit vector types.
- def v8i8 : N2VDInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4,
- itinD, OpcodeStr, !strconcat(Dt, "8"), v8i8, v8i8, IntOp>;
- def v4i16 : N2VDInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4,
- itinD, OpcodeStr, !strconcat(Dt, "16"),v4i16,v4i16,IntOp>;
- def v2i32 : N2VDInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4,
- itinD, OpcodeStr, !strconcat(Dt, "32"),v2i32,v2i32,IntOp>;
-
- // 128-bit vector types.
- def v16i8 : N2VQInt<op24_23, op21_20, 0b00, op17_16, op11_7, op4,
- itinQ, OpcodeStr, !strconcat(Dt, "8"), v16i8,v16i8,IntOp>;
- def v8i16 : N2VQInt<op24_23, op21_20, 0b01, op17_16, op11_7, op4,
- itinQ, OpcodeStr, !strconcat(Dt, "16"),v8i16,v8i16,IntOp>;
- def v4i32 : N2VQInt<op24_23, op21_20, 0b10, op17_16, op11_7, op4,
- itinQ, OpcodeStr, !strconcat(Dt, "32"),v4i32,v4i32,IntOp>;
-}
-
-
// Neon Pairwise long 2-register intrinsics,
// element sizes of 8, 16 and 32 bits:
multiclass N2VPLInt_QHS<bits<2> op24_23, bits<2> op21_20, bits<2> op17_16,
// Neon 2-register vector shift by immediate,
// with f of either N2RegVShLFrm or N2RegVShRFrm
// element sizes of 8, 16, 32 and 64 bits:
-multiclass N2VSh_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
- InstrItinClass itin, string OpcodeStr, string Dt,
- SDNode OpNode, Format f> {
+multiclass N2VShL_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
+ InstrItinClass itin, string OpcodeStr, string Dt,
+ SDNode OpNode> {
// 64-bit vector types.
- def v8i8 : N2VDSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v8i8 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "8"), v8i8, OpNode> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v4i16 : N2VDSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v4i16 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "16"), v4i16, OpNode> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v2i32 : N2VDSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v2i32 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "32"), v2i32, OpNode> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v1i64 : N2VDSh<op24, op23, op11_8, 1, op4, f, itin,
+ def v1i64 : N2VDSh<op24, op23, op11_8, 1, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "64"), v1i64, OpNode>;
// imm6 = xxxxxx
// 128-bit vector types.
- def v16i8 : N2VQSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v16i8 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "8"), v16i8, OpNode> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v8i16 : N2VQSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v8i16 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "16"), v8i16, OpNode> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v4i32 : N2VQSh<op24, op23, op11_8, 0, op4, f, itin,
+ def v4i32 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShLFrm, itin, i32imm,
OpcodeStr, !strconcat(Dt, "32"), v4i32, OpNode> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v2i64 : N2VQSh<op24, op23, op11_8, 1, op4, f, itin,
+ def v2i64 : N2VQSh<op24, op23, op11_8, 1, op4, N2RegVShLFrm, itin, i32imm,
+ OpcodeStr, !strconcat(Dt, "64"), v2i64, OpNode>;
+ // imm6 = xxxxxx
+}
+multiclass N2VShR_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
+ InstrItinClass itin, string OpcodeStr, string Dt,
+ SDNode OpNode> {
+ // 64-bit vector types.
+ def v8i8 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm8,
+ OpcodeStr, !strconcat(Dt, "8"), v8i8, OpNode> {
+ let Inst{21-19} = 0b001; // imm6 = 001xxx
+ }
+ def v4i16 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm16,
+ OpcodeStr, !strconcat(Dt, "16"), v4i16, OpNode> {
+ let Inst{21-20} = 0b01; // imm6 = 01xxxx
+ }
+ def v2i32 : N2VDSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm32,
+ OpcodeStr, !strconcat(Dt, "32"), v2i32, OpNode> {
+ let Inst{21} = 0b1; // imm6 = 1xxxxx
+ }
+ def v1i64 : N2VDSh<op24, op23, op11_8, 1, op4, N2RegVShRFrm, itin, shr_imm64,
+ OpcodeStr, !strconcat(Dt, "64"), v1i64, OpNode>;
+ // imm6 = xxxxxx
+
+ // 128-bit vector types.
+ def v16i8 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm8,
+ OpcodeStr, !strconcat(Dt, "8"), v16i8, OpNode> {
+ let Inst{21-19} = 0b001; // imm6 = 001xxx
+ }
+ def v8i16 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm16,
+ OpcodeStr, !strconcat(Dt, "16"), v8i16, OpNode> {
+ let Inst{21-20} = 0b01; // imm6 = 01xxxx
+ }
+ def v4i32 : N2VQSh<op24, op23, op11_8, 0, op4, N2RegVShRFrm, itin, shr_imm32,
+ OpcodeStr, !strconcat(Dt, "32"), v4i32, OpNode> {
+ let Inst{21} = 0b1; // imm6 = 1xxxxx
+ }
+ def v2i64 : N2VQSh<op24, op23, op11_8, 1, op4, N2RegVShRFrm, itin, shr_imm64,
OpcodeStr, !strconcat(Dt, "64"), v2i64, OpNode>;
// imm6 = xxxxxx
}
multiclass N2VShAdd_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
string OpcodeStr, string Dt, SDNode ShOp> {
// 64-bit vector types.
- def v8i8 : N2VDShAdd<op24, op23, op11_8, 0, op4,
+ def v8i8 : N2VDShAdd<op24, op23, op11_8, 0, op4, shr_imm8,
OpcodeStr, !strconcat(Dt, "8"), v8i8, ShOp> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v4i16 : N2VDShAdd<op24, op23, op11_8, 0, op4,
+ def v4i16 : N2VDShAdd<op24, op23, op11_8, 0, op4, shr_imm16,
OpcodeStr, !strconcat(Dt, "16"), v4i16, ShOp> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v2i32 : N2VDShAdd<op24, op23, op11_8, 0, op4,
+ def v2i32 : N2VDShAdd<op24, op23, op11_8, 0, op4, shr_imm32,
OpcodeStr, !strconcat(Dt, "32"), v2i32, ShOp> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v1i64 : N2VDShAdd<op24, op23, op11_8, 1, op4,
+ def v1i64 : N2VDShAdd<op24, op23, op11_8, 1, op4, shr_imm64,
OpcodeStr, !strconcat(Dt, "64"), v1i64, ShOp>;
// imm6 = xxxxxx
// 128-bit vector types.
- def v16i8 : N2VQShAdd<op24, op23, op11_8, 0, op4,
+ def v16i8 : N2VQShAdd<op24, op23, op11_8, 0, op4, shr_imm8,
OpcodeStr, !strconcat(Dt, "8"), v16i8, ShOp> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v8i16 : N2VQShAdd<op24, op23, op11_8, 0, op4,
+ def v8i16 : N2VQShAdd<op24, op23, op11_8, 0, op4, shr_imm16,
OpcodeStr, !strconcat(Dt, "16"), v8i16, ShOp> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v4i32 : N2VQShAdd<op24, op23, op11_8, 0, op4,
+ def v4i32 : N2VQShAdd<op24, op23, op11_8, 0, op4, shr_imm32,
OpcodeStr, !strconcat(Dt, "32"), v4i32, ShOp> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v2i64 : N2VQShAdd<op24, op23, op11_8, 1, op4,
+ def v2i64 : N2VQShAdd<op24, op23, op11_8, 1, op4, shr_imm64,
OpcodeStr, !strconcat(Dt, "64"), v2i64, ShOp>;
// imm6 = xxxxxx
}
-
// Neon Shift-Insert vector operations,
// with f of either N2RegVShLFrm or N2RegVShRFrm
// element sizes of 8, 16, 32 and 64 bits:
-multiclass N2VShIns_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
- string OpcodeStr, SDNode ShOp,
- Format f> {
+multiclass N2VShInsL_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
+ string OpcodeStr> {
// 64-bit vector types.
- def v8i8 : N2VDShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "8", v8i8, ShOp> {
+ def v8i8 : N2VDShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "8", v8i8, NEONvsli> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v4i16 : N2VDShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "16", v4i16, ShOp> {
+ def v4i16 : N2VDShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "16", v4i16, NEONvsli> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v2i32 : N2VDShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "32", v2i32, ShOp> {
+ def v2i32 : N2VDShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "32", v2i32, NEONvsli> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v1i64 : N2VDShIns<op24, op23, op11_8, 1, op4,
- f, OpcodeStr, "64", v1i64, ShOp>;
+ def v1i64 : N2VDShIns<op24, op23, op11_8, 1, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "64", v1i64, NEONvsli>;
// imm6 = xxxxxx
// 128-bit vector types.
- def v16i8 : N2VQShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "8", v16i8, ShOp> {
+ def v16i8 : N2VQShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "8", v16i8, NEONvsli> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
- def v8i16 : N2VQShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "16", v8i16, ShOp> {
+ def v8i16 : N2VQShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "16", v8i16, NEONvsli> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
- def v4i32 : N2VQShIns<op24, op23, op11_8, 0, op4,
- f, OpcodeStr, "32", v4i32, ShOp> {
+ def v4i32 : N2VQShIns<op24, op23, op11_8, 0, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "32", v4i32, NEONvsli> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
- def v2i64 : N2VQShIns<op24, op23, op11_8, 1, op4,
- f, OpcodeStr, "64", v2i64, ShOp>;
+ def v2i64 : N2VQShIns<op24, op23, op11_8, 1, op4, i32imm,
+ N2RegVShLFrm, OpcodeStr, "64", v2i64, NEONvsli>;
+ // imm6 = xxxxxx
+}
+multiclass N2VShInsR_QHSD<bit op24, bit op23, bits<4> op11_8, bit op4,
+ string OpcodeStr> {
+ // 64-bit vector types.
+ def v8i8 : N2VDShIns<op24, op23, op11_8, 0, op4, shr_imm8,
+ N2RegVShRFrm, OpcodeStr, "8", v8i8, NEONvsri> {
+ let Inst{21-19} = 0b001; // imm6 = 001xxx
+ }
+ def v4i16 : N2VDShIns<op24, op23, op11_8, 0, op4, shr_imm16,
+ N2RegVShRFrm, OpcodeStr, "16", v4i16, NEONvsri> {
+ let Inst{21-20} = 0b01; // imm6 = 01xxxx
+ }
+ def v2i32 : N2VDShIns<op24, op23, op11_8, 0, op4, shr_imm32,
+ N2RegVShRFrm, OpcodeStr, "32", v2i32, NEONvsri> {
+ let Inst{21} = 0b1; // imm6 = 1xxxxx
+ }
+ def v1i64 : N2VDShIns<op24, op23, op11_8, 1, op4, shr_imm64,
+ N2RegVShRFrm, OpcodeStr, "64", v1i64, NEONvsri>;
+ // imm6 = xxxxxx
+
+ // 128-bit vector types.
+ def v16i8 : N2VQShIns<op24, op23, op11_8, 0, op4, shr_imm8,
+ N2RegVShRFrm, OpcodeStr, "8", v16i8, NEONvsri> {
+ let Inst{21-19} = 0b001; // imm6 = 001xxx
+ }
+ def v8i16 : N2VQShIns<op24, op23, op11_8, 0, op4, shr_imm16,
+ N2RegVShRFrm, OpcodeStr, "16", v8i16, NEONvsri> {
+ let Inst{21-20} = 0b01; // imm6 = 01xxxx
+ }
+ def v4i32 : N2VQShIns<op24, op23, op11_8, 0, op4, shr_imm32,
+ N2RegVShRFrm, OpcodeStr, "32", v4i32, NEONvsri> {
+ let Inst{21} = 0b1; // imm6 = 1xxxxx
+ }
+ def v2i64 : N2VQShIns<op24, op23, op11_8, 1, op4, shr_imm64,
+ N2RegVShRFrm, OpcodeStr, "64", v2i64, NEONvsri>;
// imm6 = xxxxxx
}
bit op4, InstrItinClass itin, string OpcodeStr, string Dt,
SDNode OpNode> {
def v8i8 : N2VNSh<op24, op23, op11_8, op7, op6, op4, itin,
- OpcodeStr, !strconcat(Dt, "16"), v8i8, v8i16, OpNode> {
+ OpcodeStr, !strconcat(Dt, "16"),
+ v8i8, v8i16, shr_imm8, OpNode> {
let Inst{21-19} = 0b001; // imm6 = 001xxx
}
def v4i16 : N2VNSh<op24, op23, op11_8, op7, op6, op4, itin,
- OpcodeStr, !strconcat(Dt, "32"), v4i16, v4i32, OpNode> {
+ OpcodeStr, !strconcat(Dt, "32"),
+ v4i16, v4i32, shr_imm16, OpNode> {
let Inst{21-20} = 0b01; // imm6 = 01xxxx
}
def v2i32 : N2VNSh<op24, op23, op11_8, op7, op6, op4, itin,
- OpcodeStr, !strconcat(Dt, "64"), v2i32, v2i64, OpNode> {
+ OpcodeStr, !strconcat(Dt, "64"),
+ v2i32, v2i64, shr_imm32, OpNode> {
let Inst{21} = 0b1; // imm6 = 1xxxxx
}
}
defm VMLA : N3VMulOp_QHS<0, 0, 0b1001, 0, IIC_VMACi16D, IIC_VMACi32D,
IIC_VMACi16Q, IIC_VMACi32Q, "vmla", "i", add>;
def VMLAfd : N3VDMulOp<0, 0, 0b00, 0b1101, 1, IIC_VMACD, "vmla", "f32",
- v2f32, fmul, fadd>;
+ v2f32, fmul_su, fadd_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def VMLAfq : N3VQMulOp<0, 0, 0b00, 0b1101, 1, IIC_VMACQ, "vmla", "f32",
- v4f32, fmul, fadd>;
+ v4f32, fmul_su, fadd_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
defm VMLAsl : N3VMulOpSL_HS<0b0000, IIC_VMACi16D, IIC_VMACi32D,
IIC_VMACi16Q, IIC_VMACi32Q, "vmla", "i", add>;
def VMLAslfd : N3VDMulOpSL<0b10, 0b0001, IIC_VMACD, "vmla", "f32",
- v2f32, fmul, fadd>;
+ v2f32, fmul_su, fadd_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def VMLAslfq : N3VQMulOpSL<0b10, 0b0001, IIC_VMACQ, "vmla", "f32",
- v4f32, v2f32, fmul, fadd>;
+ v4f32, v2f32, fmul_su, fadd_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def : Pat<(v8i16 (add (v8i16 QPR:$src1),
(mul (v8i16 QPR:$src2),
(DSubReg_i32_reg imm:$lane))),
(SubReg_i32_lane imm:$lane)))>;
-def : Pat<(v4f32 (fadd (v4f32 QPR:$src1),
- (fmul (v4f32 QPR:$src2),
+def : Pat<(v4f32 (fadd_mlx (v4f32 QPR:$src1),
+ (fmul_su (v4f32 QPR:$src2),
(v4f32 (NEONvduplane (v4f32 QPR:$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)))>;
+ (SubReg_i32_lane imm:$lane)))>,
+ Requires<[HasNEON, UseFPVMLx]>;
// VMLAL : Vector Multiply Accumulate Long (Q += D * D)
defm VMLALs : N3VLMulOp_QHS<0,1,0b1000,0, IIC_VMACi16D, IIC_VMACi32D,
defm VMLS : N3VMulOp_QHS<1, 0, 0b1001, 0, IIC_VMACi16D, IIC_VMACi32D,
IIC_VMACi16Q, IIC_VMACi32Q, "vmls", "i", sub>;
def VMLSfd : N3VDMulOp<0, 0, 0b10, 0b1101, 1, IIC_VMACD, "vmls", "f32",
- v2f32, fmul, fsub>;
+ v2f32, fmul_su, fsub_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def VMLSfq : N3VQMulOp<0, 0, 0b10, 0b1101, 1, IIC_VMACQ, "vmls", "f32",
- v4f32, fmul, fsub>;
+ v4f32, fmul_su, fsub_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
defm VMLSsl : N3VMulOpSL_HS<0b0100, IIC_VMACi16D, IIC_VMACi32D,
IIC_VMACi16Q, IIC_VMACi32Q, "vmls", "i", sub>;
def VMLSslfd : N3VDMulOpSL<0b10, 0b0101, IIC_VMACD, "vmls", "f32",
- v2f32, fmul, fsub>;
+ v2f32, fmul_su, fsub_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def VMLSslfq : N3VQMulOpSL<0b10, 0b0101, IIC_VMACQ, "vmls", "f32",
- v4f32, v2f32, fmul, fsub>;
+ v4f32, v2f32, fmul_su, fsub_mlx>,
+ Requires<[HasNEON, UseFPVMLx]>;
def : Pat<(v8i16 (sub (v8i16 QPR:$src1),
(mul (v8i16 QPR:$src2),
(DSubReg_i32_reg imm:$lane))),
(SubReg_i32_lane imm:$lane)))>;
-def : Pat<(v4f32 (fsub (v4f32 QPR:$src1),
- (fmul (v4f32 QPR:$src2),
+def : Pat<(v4f32 (fsub_mlx (v4f32 QPR:$src1),
+ (fmul_su (v4f32 QPR:$src2),
(v4f32 (NEONvduplane (v4f32 QPR:$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)))>;
+ (SubReg_i32_lane imm:$lane)))>,
+ Requires<[HasNEON, UseFPVMLx]>;
// VMLSL : Vector Multiply Subtract Long (Q -= D * D)
defm VMLSLs : N3VLMulOp_QHS<0,1,0b1010,0, IIC_VMACi16D, IIC_VMACi32D,
NEONvceq, 1>;
defm VCEQz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00010, 0, "vceq", "i",
- "$dst, $src, #0", NEONvceqz>;
+ "$Vd, $Vm, #0", NEONvceqz>;
// VCGE : Vector Compare Greater Than or Equal
defm VCGEs : N3V_QHS<0, 0, 0b0011, 1, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q,
NEONvcge, 0>;
defm VCGEz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00001, 0, "vcge", "s",
- "$dst, $src, #0", NEONvcgez>;
+ "$Vd, $Vm, #0", NEONvcgez>;
defm VCLEz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00011, 0, "vcle", "s",
- "$dst, $src, #0", NEONvclez>;
+ "$Vd, $Vm, #0", NEONvclez>;
// VCGT : Vector Compare Greater Than
defm VCGTs : N3V_QHS<0, 0, 0b0011, 0, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q,
NEONvcgt, 0>;
defm VCGTz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00000, 0, "vcgt", "s",
- "$dst, $src, #0", NEONvcgtz>;
+ "$Vd, $Vm, #0", NEONvcgtz>;
defm VCLTz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00100, 0, "vclt", "s",
- "$dst, $src, #0", NEONvcltz>;
+ "$Vd, $Vm, #0", NEONvcltz>;
// VACGE : Vector Absolute Compare Greater Than or Equal (aka VCAGE)
def VACGEd : N3VDInt<1, 0, 0b00, 0b1110, 1, N3RegFrm, IIC_VBIND, "vacge",
// VBIC : Vector Bitwise Bit Clear (AND NOT)
-def VBICd : N3VX<0, 0, 0b01, 0b0001, 0, 1, (outs DPR:$dst),
- (ins DPR:$src1, DPR:$src2), N3RegFrm, IIC_VBINiD,
- "vbic", "$dst, $src1, $src2", "",
- [(set DPR:$dst, (v2i32 (and DPR:$src1,
- (vnotd DPR:$src2))))]>;
-def VBICq : N3VX<0, 0, 0b01, 0b0001, 1, 1, (outs QPR:$dst),
- (ins QPR:$src1, QPR:$src2), N3RegFrm, IIC_VBINiQ,
- "vbic", "$dst, $src1, $src2", "",
- [(set QPR:$dst, (v4i32 (and QPR:$src1,
- (vnotq QPR:$src2))))]>;
+def VBICd : N3VX<0, 0, 0b01, 0b0001, 0, 1, (outs DPR:$Vd),
+ (ins DPR:$Vn, DPR:$Vm), N3RegFrm, IIC_VBINiD,
+ "vbic", "$Vd, $Vn, $Vm", "",
+ [(set DPR:$Vd, (v2i32 (and DPR:$Vn,
+ (vnotd DPR:$Vm))))]>;
+def VBICq : N3VX<0, 0, 0b01, 0b0001, 1, 1, (outs QPR:$Vd),
+ (ins QPR:$Vn, QPR:$Vm), N3RegFrm, IIC_VBINiQ,
+ "vbic", "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (v4i32 (and QPR:$Vn,
+ (vnotq QPR:$Vm))))]>;
def VBICiv4i16 : N1ModImm<1, 0b000, {1,0,?,1}, 0, 0, 1, 1,
(outs DPR:$Vd), (ins nModImm:$SIMM, DPR:$src),
}
// VORN : Vector Bitwise OR NOT
-def VORNd : N3VX<0, 0, 0b11, 0b0001, 0, 1, (outs DPR:$dst),
- (ins DPR:$src1, DPR:$src2), N3RegFrm, IIC_VBINiD,
- "vorn", "$dst, $src1, $src2", "",
- [(set DPR:$dst, (v2i32 (or DPR:$src1,
- (vnotd DPR:$src2))))]>;
-def VORNq : N3VX<0, 0, 0b11, 0b0001, 1, 1, (outs QPR:$dst),
- (ins QPR:$src1, QPR:$src2), N3RegFrm, IIC_VBINiQ,
- "vorn", "$dst, $src1, $src2", "",
- [(set QPR:$dst, (v4i32 (or QPR:$src1,
- (vnotq QPR:$src2))))]>;
+def VORNd : N3VX<0, 0, 0b11, 0b0001, 0, 1, (outs DPR:$Vd),
+ (ins DPR:$Vn, DPR:$Vm), N3RegFrm, IIC_VBINiD,
+ "vorn", "$Vd, $Vn, $Vm", "",
+ [(set DPR:$Vd, (v2i32 (or DPR:$Vn,
+ (vnotd DPR:$Vm))))]>;
+def VORNq : N3VX<0, 0, 0b11, 0b0001, 1, 1, (outs QPR:$Vd),
+ (ins QPR:$Vn, QPR:$Vm), N3RegFrm, IIC_VBINiQ,
+ "vorn", "$Vd, $Vn, $Vm", "",
+ [(set QPR:$Vd, (v4i32 (or QPR:$Vn,
+ (vnotq QPR:$Vm))))]>;
// VMVN : Vector Bitwise NOT (Immediate)
let isReMaterializable = 1 in {
-def VMVNv4i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 0, 1, 1, (outs DPR:$dst),
+def VMVNv4i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 0, 1, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmvn", "i16", "$dst, $SIMM", "",
- [(set DPR:$dst, (v4i16 (NEONvmvnImm timm:$SIMM)))]> {
+ "vmvn", "i16", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v4i16 (NEONvmvnImm timm:$SIMM)))]> {
let Inst{9} = SIMM{9};
}
-def VMVNv8i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 1, 1, 1, (outs QPR:$dst),
+def VMVNv8i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 1, 1, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmvn", "i16", "$dst, $SIMM", "",
- [(set QPR:$dst, (v8i16 (NEONvmvnImm timm:$SIMM)))]> {
+ "vmvn", "i16", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v8i16 (NEONvmvnImm timm:$SIMM)))]> {
let Inst{9} = SIMM{9};
}
-def VMVNv2i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 0, 1, 1, (outs DPR:$dst),
+def VMVNv2i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 0, 1, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmvn", "i32", "$dst, $SIMM", "",
- [(set DPR:$dst, (v2i32 (NEONvmvnImm timm:$SIMM)))]> {
+ "vmvn", "i32", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v2i32 (NEONvmvnImm timm:$SIMM)))]> {
let Inst{11-8} = SIMM{11-8};
}
-def VMVNv4i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 1, 1, 1, (outs QPR:$dst),
+def VMVNv4i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 1, 1, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmvn", "i32", "$dst, $SIMM", "",
- [(set QPR:$dst, (v4i32 (NEONvmvnImm timm:$SIMM)))]> {
+ "vmvn", "i32", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v4i32 (NEONvmvnImm timm:$SIMM)))]> {
let Inst{11-8} = SIMM{11-8};
}
}
// VMVN : Vector Bitwise NOT
def VMVNd : N2VX<0b11, 0b11, 0b00, 0b00, 0b01011, 0, 0,
- (outs DPR:$dst), (ins DPR:$src), IIC_VSUBiD,
- "vmvn", "$dst, $src", "",
- [(set DPR:$dst, (v2i32 (vnotd DPR:$src)))]>;
+ (outs DPR:$Vd), (ins DPR:$Vm), IIC_VSUBiD,
+ "vmvn", "$Vd, $Vm", "",
+ [(set DPR:$Vd, (v2i32 (vnotd DPR:$Vm)))]>;
def VMVNq : N2VX<0b11, 0b11, 0b00, 0b00, 0b01011, 1, 0,
- (outs QPR:$dst), (ins QPR:$src), IIC_VSUBiD,
- "vmvn", "$dst, $src", "",
- [(set QPR:$dst, (v4i32 (vnotq QPR:$src)))]>;
+ (outs QPR:$Vd), (ins QPR:$Vm), IIC_VSUBiD,
+ "vmvn", "$Vd, $Vm", "",
+ [(set QPR:$Vd, (v4i32 (vnotq QPR:$Vm)))]>;
def : Pat<(v2i32 (vnotd DPR:$src)), (VMVNd DPR:$src)>;
def : Pat<(v4i32 (vnotq QPR:$src)), (VMVNq QPR:$src)>;
N3RegFrm, IIC_VCNTiD,
"vbsl", "$Vd, $Vn, $Vm", "$src1 = $Vd",
[(set DPR:$Vd,
- (v2i32 (or (and DPR:$Vn, DPR:$src1),
- (and DPR:$Vm, (vnotd DPR:$src1)))))]>;
+ (v2i32 (NEONvbsl DPR:$src1, DPR:$Vn, DPR:$Vm)))]>;
+
+def : Pat<(v2i32 (or (and DPR:$Vn, DPR:$Vd),
+ (and DPR:$Vm, (vnotd DPR:$Vd)))),
+ (VBSLd DPR:$Vd, DPR:$Vn, DPR:$Vm)>;
+
def VBSLq : N3VX<1, 0, 0b01, 0b0001, 1, 1, (outs QPR:$Vd),
(ins QPR:$src1, QPR:$Vn, QPR:$Vm),
N3RegFrm, IIC_VCNTiQ,
"vbsl", "$Vd, $Vn, $Vm", "$src1 = $Vd",
[(set QPR:$Vd,
- (v4i32 (or (and QPR:$Vn, QPR:$src1),
- (and QPR:$Vm, (vnotq QPR:$src1)))))]>;
+ (v4i32 (NEONvbsl QPR:$src1, QPR:$Vn, QPR:$Vm)))]>;
+
+def : Pat<(v4i32 (or (and QPR:$Vn, QPR:$Vd),
+ (and QPR:$Vm, (vnotq QPR:$Vd)))),
+ (VBSLq QPR:$Vd, QPR:$Vn, QPR:$Vm)>;
// VBIF : Vector Bitwise Insert if False
// like VBSL but with: "vbif $dst, $src3, $src1", "$src2 = $dst",
defm VSHLu : N3VInt_QHSDSh<1, 0, 0b0100, 0, N3RegVShFrm,
IIC_VSHLiD, IIC_VSHLiD, IIC_VSHLiQ, IIC_VSHLiQ,
"vshl", "u", int_arm_neon_vshiftu>;
+
// VSHL : Vector Shift Left (Immediate)
-defm VSHLi : N2VSh_QHSD<0, 1, 0b0101, 1, IIC_VSHLiD, "vshl", "i", NEONvshl,
- N2RegVShLFrm>;
+defm VSHLi : N2VShL_QHSD<0, 1, 0b0101, 1, IIC_VSHLiD, "vshl", "i", NEONvshl>;
+
// VSHR : Vector Shift Right (Immediate)
-defm VSHRs : N2VSh_QHSD<0, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "s", NEONvshrs,
- N2RegVShRFrm>;
-defm VSHRu : N2VSh_QHSD<1, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "u", NEONvshru,
- N2RegVShRFrm>;
+defm VSHRs : N2VShR_QHSD<0, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "s",NEONvshrs>;
+defm VSHRu : N2VShR_QHSD<1, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "u",NEONvshru>;
// VSHLL : Vector Shift Left Long
defm VSHLLs : N2VLSh_QHS<0, 1, 0b1010, 0, 0, 1, "vshll", "s", NEONvshlls>;
: N2VLSh<op24, op23, op11_8, op7, op6, op4, OpcodeStr, Dt,
ResTy, OpTy, OpNode> {
let Inst{21-16} = op21_16;
+ let DecoderMethod = "DecodeVSHLMaxInstruction";
}
def VSHLLi8 : N2VLShMax<1, 1, 0b110010, 0b0011, 0, 0, 0, "vshll", "i8",
v8i16, v8i8, NEONvshlli>;
IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, IIC_VSHLi4Q,
"vrshl", "u", int_arm_neon_vrshiftu>;
// VRSHR : Vector Rounding Shift Right
-defm VRSHRs : N2VSh_QHSD<0,1,0b0010,1, IIC_VSHLi4D, "vrshr", "s", NEONvrshrs,
- N2RegVShRFrm>;
-defm VRSHRu : N2VSh_QHSD<1,1,0b0010,1, IIC_VSHLi4D, "vrshr", "u", NEONvrshru,
- N2RegVShRFrm>;
+defm VRSHRs : N2VShR_QHSD<0,1,0b0010,1, IIC_VSHLi4D, "vrshr", "s",NEONvrshrs>;
+defm VRSHRu : N2VShR_QHSD<1,1,0b0010,1, IIC_VSHLi4D, "vrshr", "u",NEONvrshru>;
// VRSHRN : Vector Rounding Shift Right and Narrow
defm VRSHRN : N2VNSh_HSD<0, 1, 0b1000, 0, 1, 1, IIC_VSHLi4D, "vrshrn", "i",
IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q, IIC_VSHLi4Q,
"vqshl", "u", int_arm_neon_vqshiftu>;
// VQSHL : Vector Saturating Shift Left (Immediate)
-defm VQSHLsi : N2VSh_QHSD<0,1,0b0111,1, IIC_VSHLi4D, "vqshl", "s",NEONvqshls,
- N2RegVShLFrm>;
-defm VQSHLui : N2VSh_QHSD<1,1,0b0111,1, IIC_VSHLi4D, "vqshl", "u",NEONvqshlu,
- N2RegVShLFrm>;
+defm VQSHLsi : N2VShL_QHSD<0,1,0b0111,1, IIC_VSHLi4D, "vqshl", "s",NEONvqshls>;
+defm VQSHLui : N2VShL_QHSD<1,1,0b0111,1, IIC_VSHLi4D, "vqshl", "u",NEONvqshlu>;
+
// VQSHLU : Vector Saturating Shift Left (Immediate, Unsigned)
-defm VQSHLsu : N2VSh_QHSD<1,1,0b0110,1, IIC_VSHLi4D,"vqshlu","s",NEONvqshlsu,
- N2RegVShLFrm>;
+defm VQSHLsu : N2VShL_QHSD<1,1,0b0110,1, IIC_VSHLi4D,"vqshlu","s",NEONvqshlsu>;
// VQSHRN : Vector Saturating Shift Right and Narrow
defm VQSHRNs : N2VNSh_HSD<0, 1, 0b1001, 0, 0, 1, IIC_VSHLi4D, "vqshrn", "s",
defm VRSRAu : N2VShAdd_QHSD<1, 1, 0b0011, 1, "vrsra", "u", NEONvrshru>;
// VSLI : Vector Shift Left and Insert
-defm VSLI : N2VShIns_QHSD<1, 1, 0b0101, 1, "vsli", NEONvsli, N2RegVShLFrm>;
+defm VSLI : N2VShInsL_QHSD<1, 1, 0b0101, 1, "vsli">;
+
// VSRI : Vector Shift Right and Insert
-defm VSRI : N2VShIns_QHSD<1, 1, 0b0100, 1, "vsri", NEONvsri, N2RegVShRFrm>;
+defm VSRI : N2VShInsR_QHSD<1, 1, 0b0100, 1, "vsri">;
// Vector Absolute and Saturating Absolute.
(sub (bitconvert (v4i32 NEONimmAllZerosV)), node:$in)>;
class VNEGD<bits<2> size, string OpcodeStr, string Dt, ValueType Ty>
- : N2V<0b11, 0b11, size, 0b01, 0b00111, 0, 0, (outs DPR:$dst), (ins DPR:$src),
- IIC_VSHLiD, OpcodeStr, Dt, "$dst, $src", "",
- [(set DPR:$dst, (Ty (vnegd DPR:$src)))]>;
+ : N2V<0b11, 0b11, size, 0b01, 0b00111, 0, 0, (outs DPR:$Vd), (ins DPR:$Vm),
+ IIC_VSHLiD, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set DPR:$Vd, (Ty (vnegd DPR:$Vm)))]>;
class VNEGQ<bits<2> size, string OpcodeStr, string Dt, ValueType Ty>
- : N2V<0b11, 0b11, size, 0b01, 0b00111, 1, 0, (outs QPR:$dst), (ins QPR:$src),
- IIC_VSHLiQ, OpcodeStr, Dt, "$dst, $src", "",
- [(set QPR:$dst, (Ty (vnegq QPR:$src)))]>;
+ : N2V<0b11, 0b11, size, 0b01, 0b00111, 1, 0, (outs QPR:$Vd), (ins QPR:$Vm),
+ IIC_VSHLiQ, OpcodeStr, Dt, "$Vd, $Vm", "",
+ [(set QPR:$Vd, (Ty (vnegq QPR:$Vm)))]>;
// VNEG : Vector Negate (integer)
def VNEGs8d : VNEGD<0b00, "vneg", "s8", v8i8>;
// VNEG : Vector Negate (floating-point)
def VNEGfd : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0,
- (outs DPR:$dst), (ins DPR:$src), IIC_VUNAD,
- "vneg", "f32", "$dst, $src", "",
- [(set DPR:$dst, (v2f32 (fneg DPR:$src)))]>;
+ (outs DPR:$Vd), (ins DPR:$Vm), IIC_VUNAD,
+ "vneg", "f32", "$Vd, $Vm", "",
+ [(set DPR:$Vd, (v2f32 (fneg DPR:$Vm)))]>;
def VNEGf32q : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 1, 0,
- (outs QPR:$dst), (ins QPR:$src), IIC_VUNAQ,
- "vneg", "f32", "$dst, $src", "",
- [(set QPR:$dst, (v4f32 (fneg QPR:$src)))]>;
+ (outs QPR:$Vd), (ins QPR:$Vm), IIC_VUNAQ,
+ "vneg", "f32", "$Vd, $Vm", "",
+ [(set QPR:$Vd, (v4f32 (fneg QPR:$Vm)))]>;
def : Pat<(v8i8 (vnegd DPR:$src)), (VNEGs8d DPR:$src)>;
def : Pat<(v4i16 (vnegd DPR:$src)), (VNEGs16d DPR:$src)>;
// Vector Swap -- for disassembly only.
def VSWPd : N2VX<0b11, 0b11, 0b00, 0b10, 0b00000, 0, 0,
- (outs DPR:$dst), (ins DPR:$src), NoItinerary,
- "vswp", "$dst, $src", "", []>;
+ (outs DPR:$Vd), (ins DPR:$Vm), NoItinerary,
+ "vswp", "$Vd, $Vm", "", []>;
def VSWPq : N2VX<0b11, 0b11, 0b00, 0b10, 0b00000, 1, 0,
- (outs QPR:$dst), (ins QPR:$src), NoItinerary,
- "vswp", "$dst, $src", "", []>;
+ (outs QPR:$Vd), (ins QPR:$Vm), NoItinerary,
+ "vswp", "$Vd, $Vm", "", []>;
// Vector Move Operations.
// VMOV : Vector Move (Register)
-
-let neverHasSideEffects = 1 in {
-def VMOVDneon: N3VX<0, 0, 0b10, 0b0001, 0, 1, (outs DPR:$Vd), (ins DPR:$Vm),
- N3RegFrm, IIC_VMOV, "vmov", "$Vd, $Vm", "", []> {
- let Vn{4-0} = Vm{4-0};
-}
-def VMOVQ : N3VX<0, 0, 0b10, 0b0001, 1, 1, (outs QPR:$Vd), (ins QPR:$Vm),
- N3RegFrm, IIC_VMOV, "vmov", "$Vd, $Vm", "", []> {
- let Vn{4-0} = Vm{4-0};
-}
-
-// Pseudo vector move instructions for QQ and QQQQ registers. This should
-// be expanded after register allocation is completed.
-def VMOVQQ : PseudoInst<(outs QQPR:$dst), (ins QQPR:$src),
- NoItinerary, []>;
-
-def VMOVQQQQ : PseudoInst<(outs QQQQPR:$dst), (ins QQQQPR:$src),
- NoItinerary, []>;
-} // neverHasSideEffects
+def : InstAlias<"vmov${p} $Vd, $Vm",
+ (VORRd DPR:$Vd, DPR:$Vm, DPR:$Vm, pred:$p)>;
+def : InstAlias<"vmov${p} $Vd, $Vm",
+ (VORRq QPR:$Vd, QPR:$Vm, QPR:$Vm, pred:$p)>;
// VMOV : Vector Move (Immediate)
let isReMaterializable = 1 in {
-def VMOVv8i8 : N1ModImm<1, 0b000, 0b1110, 0, 0, 0, 1, (outs DPR:$dst),
+def VMOVv8i8 : N1ModImm<1, 0b000, 0b1110, 0, 0, 0, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i8", "$dst, $SIMM", "",
- [(set DPR:$dst, (v8i8 (NEONvmovImm timm:$SIMM)))]>;
-def VMOVv16i8 : N1ModImm<1, 0b000, 0b1110, 0, 1, 0, 1, (outs QPR:$dst),
+ "vmov", "i8", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v8i8 (NEONvmovImm timm:$SIMM)))]>;
+def VMOVv16i8 : N1ModImm<1, 0b000, 0b1110, 0, 1, 0, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i8", "$dst, $SIMM", "",
- [(set QPR:$dst, (v16i8 (NEONvmovImm timm:$SIMM)))]>;
+ "vmov", "i8", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v16i8 (NEONvmovImm timm:$SIMM)))]>;
-def VMOVv4i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 0, 0, 1, (outs DPR:$dst),
+def VMOVv4i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 0, 0, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i16", "$dst, $SIMM", "",
- [(set DPR:$dst, (v4i16 (NEONvmovImm timm:$SIMM)))]> {
+ "vmov", "i16", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v4i16 (NEONvmovImm timm:$SIMM)))]> {
let Inst{9} = SIMM{9};
}
-def VMOVv8i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 1, 0, 1, (outs QPR:$dst),
+def VMOVv8i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 1, 0, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i16", "$dst, $SIMM", "",
- [(set QPR:$dst, (v8i16 (NEONvmovImm timm:$SIMM)))]> {
+ "vmov", "i16", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v8i16 (NEONvmovImm timm:$SIMM)))]> {
let Inst{9} = SIMM{9};
}
-def VMOVv2i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 0, 0, 1, (outs DPR:$dst),
+def VMOVv2i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 0, 0, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i32", "$dst, $SIMM", "",
- [(set DPR:$dst, (v2i32 (NEONvmovImm timm:$SIMM)))]> {
+ "vmov", "i32", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v2i32 (NEONvmovImm timm:$SIMM)))]> {
let Inst{11-8} = SIMM{11-8};
}
-def VMOVv4i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 1, 0, 1, (outs QPR:$dst),
+def VMOVv4i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 1, 0, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i32", "$dst, $SIMM", "",
- [(set QPR:$dst, (v4i32 (NEONvmovImm timm:$SIMM)))]> {
+ "vmov", "i32", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v4i32 (NEONvmovImm timm:$SIMM)))]> {
let Inst{11-8} = SIMM{11-8};
}
-def VMOVv1i64 : N1ModImm<1, 0b000, 0b1110, 0, 0, 1, 1, (outs DPR:$dst),
+def VMOVv1i64 : N1ModImm<1, 0b000, 0b1110, 0, 0, 1, 1, (outs DPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i64", "$dst, $SIMM", "",
- [(set DPR:$dst, (v1i64 (NEONvmovImm timm:$SIMM)))]>;
-def VMOVv2i64 : N1ModImm<1, 0b000, 0b1110, 0, 1, 1, 1, (outs QPR:$dst),
+ "vmov", "i64", "$Vd, $SIMM", "",
+ [(set DPR:$Vd, (v1i64 (NEONvmovImm timm:$SIMM)))]>;
+def VMOVv2i64 : N1ModImm<1, 0b000, 0b1110, 0, 1, 1, 1, (outs QPR:$Vd),
(ins nModImm:$SIMM), IIC_VMOVImm,
- "vmov", "i64", "$dst, $SIMM", "",
- [(set QPR:$dst, (v2i64 (NEONvmovImm timm:$SIMM)))]>;
+ "vmov", "i64", "$Vd, $SIMM", "",
+ [(set QPR:$Vd, (v2i64 (NEONvmovImm timm:$SIMM)))]>;
} // isReMaterializable
// VMOV : Vector Get Lane (move scalar to ARM core register)
// VDUP : Vector Duplicate (from ARM core register to all elements)
class VDUPD<bits<8> opcod1, bits<2> opcod3, string Dt, ValueType Ty>
- : NVDup<opcod1, 0b1011, opcod3, (outs DPR:$dst), (ins GPR:$src),
- IIC_VMOVIS, "vdup", Dt, "$dst, $src",
- [(set DPR:$dst, (Ty (NEONvdup (i32 GPR:$src))))]>;
+ : NVDup<opcod1, 0b1011, opcod3, (outs DPR:$V), (ins GPR:$R),
+ IIC_VMOVIS, "vdup", Dt, "$V, $R",
+ [(set DPR:$V, (Ty (NEONvdup (i32 GPR:$R))))]>;
class VDUPQ<bits<8> opcod1, bits<2> opcod3, string Dt, ValueType Ty>
- : NVDup<opcod1, 0b1011, opcod3, (outs QPR:$dst), (ins GPR:$src),
- IIC_VMOVIS, "vdup", Dt, "$dst, $src",
- [(set QPR:$dst, (Ty (NEONvdup (i32 GPR:$src))))]>;
+ : NVDup<opcod1, 0b1011, opcod3, (outs QPR:$V), (ins GPR:$R),
+ IIC_VMOVIS, "vdup", Dt, "$V, $R",
+ [(set QPR:$V, (Ty (NEONvdup (i32 GPR:$R))))]>;
def VDUP8d : VDUPD<0b11101100, 0b00, "8", v8i8>;
def VDUP16d : VDUPD<0b11101000, 0b01, "16", v4i16>;
def VDUP16q : VDUPQ<0b11101010, 0b01, "16", v8i16>;
def VDUP32q : VDUPQ<0b11101010, 0b00, "32", v4i32>;
-def VDUPfd : NVDup<0b11101000, 0b1011, 0b00, (outs DPR:$dst), (ins GPR:$src),
- IIC_VMOVIS, "vdup", "32", "$dst, $src",
- [(set DPR:$dst, (v2f32 (NEONvdup
- (f32 (bitconvert GPR:$src)))))]>;
-def VDUPfq : NVDup<0b11101010, 0b1011, 0b00, (outs QPR:$dst), (ins GPR:$src),
- IIC_VMOVIS, "vdup", "32", "$dst, $src",
- [(set QPR:$dst, (v4f32 (NEONvdup
- (f32 (bitconvert GPR:$src)))))]>;
+def : Pat<(v2f32 (NEONvdup (f32 (bitconvert GPR:$R)))), (VDUP32d GPR:$R)>;
+def : Pat<(v4f32 (NEONvdup (f32 (bitconvert GPR:$R)))), (VDUP32q GPR:$R)>;
// VDUP : Vector Duplicate Lane (from scalar to all elements)
class VDUPLND<bits<4> op19_16, string OpcodeStr, string Dt,
- ValueType Ty>
- : NVDupLane<op19_16, 0, (outs DPR:$dst), (ins DPR:$src, nohash_imm:$lane),
- IIC_VMOVD, OpcodeStr, Dt, "$dst, $src[$lane]",
- [(set DPR:$dst, (Ty (NEONvduplane (Ty DPR:$src), imm:$lane)))]>;
+ ValueType Ty, Operand IdxTy>
+ : NVDupLane<op19_16, 0, (outs DPR:$Vd), (ins DPR:$Vm, IdxTy:$lane),
+ IIC_VMOVD, OpcodeStr, Dt, "$Vd, $Vm$lane",
+ [(set DPR:$Vd, (Ty (NEONvduplane (Ty DPR:$Vm), imm:$lane)))]>;
class VDUPLNQ<bits<4> op19_16, string OpcodeStr, string Dt,
- ValueType ResTy, ValueType OpTy>
- : NVDupLane<op19_16, 1, (outs QPR:$dst), (ins DPR:$src, nohash_imm:$lane),
- IIC_VMOVQ, OpcodeStr, Dt, "$dst, $src[$lane]",
- [(set QPR:$dst, (ResTy (NEONvduplane (OpTy DPR:$src),
- imm:$lane)))]>;
+ ValueType ResTy, ValueType OpTy, Operand IdxTy>
+ : NVDupLane<op19_16, 1, (outs QPR:$Vd), (ins DPR:$Vm, IdxTy:$lane),
+ IIC_VMOVQ, OpcodeStr, Dt, "$Vd, $Vm$lane",
+ [(set QPR:$Vd, (ResTy (NEONvduplane (OpTy DPR:$Vm),
+ VectorIndex32:$lane)))]>;
// Inst{19-16} is partially specified depending on the element size.
-def VDUPLN8d : VDUPLND<{?,?,?,1}, "vdup", "8", v8i8> {
+def VDUPLN8d : VDUPLND<{?,?,?,1}, "vdup", "8", v8i8, VectorIndex8> {
+ bits<3> lane;
let Inst{19-17} = lane{2-0};
}
-def VDUPLN16d : VDUPLND<{?,?,1,0}, "vdup", "16", v4i16> {
+def VDUPLN16d : VDUPLND<{?,?,1,0}, "vdup", "16", v4i16, VectorIndex16> {
+ bits<2> lane;
let Inst{19-18} = lane{1-0};
}
-def VDUPLN32d : VDUPLND<{?,1,0,0}, "vdup", "32", v2i32> {
- let Inst{19} = lane{0};
-}
-def VDUPLNfd : VDUPLND<{?,1,0,0}, "vdup", "32", v2f32> {
+def VDUPLN32d : VDUPLND<{?,1,0,0}, "vdup", "32", v2i32, VectorIndex32> {
+ bits<1> lane;
let Inst{19} = lane{0};
}
-def VDUPLN8q : VDUPLNQ<{?,?,?,1}, "vdup", "8", v16i8, v8i8> {
+def VDUPLN8q : VDUPLNQ<{?,?,?,1}, "vdup", "8", v16i8, v8i8, VectorIndex8> {
+ bits<3> lane;
let Inst{19-17} = lane{2-0};
}
-def VDUPLN16q : VDUPLNQ<{?,?,1,0}, "vdup", "16", v8i16, v4i16> {
+def VDUPLN16q : VDUPLNQ<{?,?,1,0}, "vdup", "16", v8i16, v4i16, VectorIndex16> {
+ bits<2> lane;
let Inst{19-18} = lane{1-0};
}
-def VDUPLN32q : VDUPLNQ<{?,1,0,0}, "vdup", "32", v4i32, v2i32> {
- let Inst{19} = lane{0};
-}
-def VDUPLNfq : VDUPLNQ<{?,1,0,0}, "vdup", "32", v4f32, v2f32> {
+def VDUPLN32q : VDUPLNQ<{?,1,0,0}, "vdup", "32", v4i32, v2i32, VectorIndex32> {
+ bits<1> lane;
let Inst{19} = lane{0};
}
+def : Pat<(v2f32 (NEONvduplane (v2f32 DPR:$Vm), imm:$lane)),
+ (VDUPLN32d DPR:$Vm, imm:$lane)>;
+
+def : Pat<(v4f32 (NEONvduplane (v2f32 DPR:$Vm), imm:$lane)),
+ (VDUPLN32q DPR:$Vm, imm:$lane)>;
+
def : Pat<(v16i8 (NEONvduplane (v16i8 QPR:$src), imm:$lane)),
(v16i8 (VDUPLN8q (v8i8 (EXTRACT_SUBREG QPR:$src,
(DSubReg_i8_reg imm:$lane))),
(DSubReg_i32_reg imm:$lane))),
(SubReg_i32_lane imm:$lane)))>;
def : Pat<(v4f32 (NEONvduplane (v4f32 QPR:$src), imm:$lane)),
- (v4f32 (VDUPLNfq (v2f32 (EXTRACT_SUBREG QPR:$src,
+ (v4f32 (VDUPLN32q (v2f32 (EXTRACT_SUBREG QPR:$src,
(DSubReg_i32_reg imm:$lane))),
(SubReg_i32_lane imm:$lane)))>;
def VCVTxu2fq : N2VCvtQ<1, 1, 0b1110, 0, 1, "vcvt", "f32.u32",
v4f32, v4i32, int_arm_neon_vcvtfxu2fp>;
+// VCVT : Vector Convert Between Half-Precision and Single-Precision.
+def VCVTf2h : N2VNInt<0b11, 0b11, 0b01, 0b10, 0b01100, 0, 0,
+ IIC_VUNAQ, "vcvt", "f16.f32",
+ v4i16, v4f32, int_arm_neon_vcvtfp2hf>,
+ Requires<[HasNEON, HasFP16]>;
+def VCVTh2f : N2VLInt<0b11, 0b11, 0b01, 0b10, 0b01110, 0, 0,
+ IIC_VUNAQ, "vcvt", "f32.f16",
+ v4f32, v4i16, int_arm_neon_vcvthf2fp>,
+ Requires<[HasNEON, HasFP16]>;
+
// Vector Reverse.
// VREV64 : Vector Reverse elements within 64-bit doublewords
def VREV64d8 : VREV64D<0b00, "vrev64", "8", v8i8>;
def VREV64d16 : VREV64D<0b01, "vrev64", "16", v4i16>;
def VREV64d32 : VREV64D<0b10, "vrev64", "32", v2i32>;
-def VREV64df : VREV64D<0b10, "vrev64", "32", v2f32>;
+def : Pat<(v2f32 (NEONvrev64 (v2f32 DPR:$Vm))), (VREV64d32 DPR:$Vm)>;
def VREV64q8 : VREV64Q<0b00, "vrev64", "8", v16i8>;
def VREV64q16 : VREV64Q<0b01, "vrev64", "16", v8i16>;
def VREV64q32 : VREV64Q<0b10, "vrev64", "32", v4i32>;
-def VREV64qf : VREV64Q<0b10, "vrev64", "32", v4f32>;
+def : Pat<(v4f32 (NEONvrev64 (v4f32 QPR:$Vm))), (VREV64q32 QPR:$Vm)>;
// VREV32 : Vector Reverse elements within 32-bit words
// Other Vector Shuffles.
+// Aligned extractions: really just dropping registers
+
+class AlignedVEXTq<ValueType DestTy, ValueType SrcTy, SDNodeXForm LaneCVT>
+ : Pat<(DestTy (vector_extract_subvec (SrcTy QPR:$src), (i32 imm:$start))),
+ (EXTRACT_SUBREG (SrcTy QPR:$src), (LaneCVT imm:$start))>;
+
+def : AlignedVEXTq<v8i8, v16i8, DSubReg_i8_reg>;
+
+def : AlignedVEXTq<v4i16, v8i16, DSubReg_i16_reg>;
+
+def : AlignedVEXTq<v2i32, v4i32, DSubReg_i32_reg>;
+
+def : AlignedVEXTq<v1i64, v2i64, DSubReg_f64_reg>;
+
+def : AlignedVEXTq<v2f32, v4f32, DSubReg_i32_reg>;
+
+
// VEXT : Vector Extract
class VEXTd<string OpcodeStr, string Dt, ValueType Ty>
let Inst{11-10} = index{1-0};
let Inst{9-8} = 0b00;
}
-def VEXTdf : VEXTd<"vext", "32", v2f32> {
- let Inst{11} = index{0};
- let Inst{10-8} = 0b000;
-}
+def : Pat<(v2f32 (NEONvext (v2f32 DPR:$Vn),
+ (v2f32 DPR:$Vm),
+ (i32 imm:$index))),
+ (VEXTd32 DPR:$Vn, DPR:$Vm, imm:$index)>;
def VEXTq8 : VEXTq<"vext", "8", v16i8> {
let Inst{11-8} = index{3-0};
let Inst{11-10} = index{1-0};
let Inst{9-8} = 0b00;
}
-def VEXTqf : VEXTq<"vext", "32", v4f32> {
- let Inst{11} = index{0};
- let Inst{10-8} = 0b000;
-}
+def : Pat<(v4f32 (NEONvext (v4f32 QPR:$Vn),
+ (v4f32 QPR:$Vm),
+ (i32 imm:$index))),
+ (VEXTq32 QPR:$Vn, QPR:$Vm, imm:$index)>;
// VTRN : Vector Transpose
// Vector Table Lookup and Table Extension.
// VTBL : Vector Table Lookup
+let DecoderMethod = "DecodeTBLInstruction" in {
def VTBL1
: N3V<1,1,0b11,0b1000,0,0, (outs DPR:$Vd),
(ins DPR:$Vn, DPR:$Vm), NVTBLFrm, IIC_VTB1,
def VTBX4Pseudo
: PseudoNeonI<(outs DPR:$dst), (ins DPR:$orig, QQPR:$tbl, DPR:$src),
IIC_VTBX4, "$orig = $dst", []>;
+} // DecoderMethod = "DecodeTBLInstruction"
//===----------------------------------------------------------------------===//
// NEON instructions for single-precision FP math
//===----------------------------------------------------------------------===//
-class N2VSPat<SDNode OpNode, ValueType ResTy, ValueType OpTy, NeonI Inst>
- : NEONFPPat<(ResTy (OpNode SPR:$a)),
- (EXTRACT_SUBREG (OpTy (Inst (INSERT_SUBREG (OpTy (IMPLICIT_DEF)),
- SPR:$a, ssub_0))),
- ssub_0)>;
+class N2VSPat<SDNode OpNode, NeonI Inst>
+ : NEONFPPat<(f32 (OpNode SPR:$a)),
+ (EXTRACT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (Inst
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$a, ssub_0)), DPR_VFP2)), ssub_0)>;
class N3VSPat<SDNode OpNode, NeonI Inst>
: NEONFPPat<(f32 (OpNode SPR:$a, SPR:$b)),
- (EXTRACT_SUBREG (v2f32
- (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
- SPR:$a, ssub_0),
- (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
- SPR:$b, ssub_0))),
- ssub_0)>;
+ (EXTRACT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (Inst
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$a, ssub_0),
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$b, ssub_0)), DPR_VFP2)), ssub_0)>;
class N3VSMulOpPat<SDNode MulNode, SDNode OpNode, NeonI Inst>
: NEONFPPat<(f32 (OpNode SPR:$acc, (f32 (MulNode SPR:$a, SPR:$b)))),
- (EXTRACT_SUBREG (Inst (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
- SPR:$acc, ssub_0),
- (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
- SPR:$a, ssub_0),
- (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)),
- SPR:$b, ssub_0)),
- ssub_0)>;
-
-// These need separate instructions because they must use DPR_VFP2 register
-// class which have SPR sub-registers.
-
-// Vector Add Operations used for single-precision FP
-let neverHasSideEffects = 1 in
-def VADDfd_sfp : N3VS<0,0,0b00,0b1101,0, "vadd", "f32", v2f32, v2f32, fadd, 1>;
-def : N3VSPat<fadd, VADDfd_sfp>;
-
-// Vector Sub Operations used for single-precision FP
-let neverHasSideEffects = 1 in
-def VSUBfd_sfp : N3VS<0,0,0b10,0b1101,0, "vsub", "f32", v2f32, v2f32, fsub, 0>;
-def : N3VSPat<fsub, VSUBfd_sfp>;
-
-// Vector Multiply Operations used for single-precision FP
-let neverHasSideEffects = 1 in
-def VMULfd_sfp : N3VS<1,0,0b00,0b1101,1, "vmul", "f32", v2f32, v2f32, fmul, 1>;
-def : N3VSPat<fmul, VMULfd_sfp>;
-
-// Vector Multiply-Accumulate/Subtract used for single-precision FP
-// vml[as].f32 can cause 4-8 cycle stalls in following ASIMD instructions, so
-// we want to avoid them for now. e.g., alternating vmla/vadd instructions.
-
-//let neverHasSideEffects = 1 in
-//def VMLAfd_sfp : N3VSMulOp<0,0,0b00,0b1101,1, IIC_VMACD, "vmla", "f32",
-// v2f32, fmul, fadd>;
-//def : N3VSMulOpPat<fmul, fadd, VMLAfd_sfp>;
-
-//let neverHasSideEffects = 1 in
-//def VMLSfd_sfp : N3VSMulOp<0,0,0b10,0b1101,1, IIC_VMACD, "vmls", "f32",
-// v2f32, fmul, fsub>;
-//def : N3VSMulOpPat<fmul, fsub, VMLSfd_sfp>;
-
-// Vector Absolute used for single-precision FP
-let neverHasSideEffects = 1 in
-def VABSfd_sfp : N2V<0b11, 0b11, 0b10, 0b01, 0b01110, 0, 0,
- (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), IIC_VUNAD,
- "vabs", "f32", "$dst, $src", "", []>;
-def : N2VSPat<fabs, f32, v2f32, VABSfd_sfp>;
-
-// Vector Negate used for single-precision FP
-let neverHasSideEffects = 1 in
-def VNEGfd_sfp : N2V<0b11, 0b11, 0b10, 0b01, 0b01111, 0, 0,
- (outs DPR_VFP2:$dst), (ins DPR_VFP2:$src), IIC_VUNAD,
- "vneg", "f32", "$dst, $src", "", []>;
-def : N2VSPat<fneg, f32, v2f32, VNEGfd_sfp>;
-
-// Vector Maximum used for single-precision FP
-let neverHasSideEffects = 1 in
-def VMAXfd_sfp : N3V<0, 0, 0b00, 0b1111, 0, 0, (outs DPR_VFP2:$dst),
- (ins DPR_VFP2:$src1, DPR_VFP2:$src2), N3RegFrm, IIC_VBIND,
- "vmax", "f32", "$dst, $src1, $src2", "", []>;
-def : N3VSPat<NEONfmax, VMAXfd_sfp>;
-
-// Vector Minimum used for single-precision FP
-let neverHasSideEffects = 1 in
-def VMINfd_sfp : N3V<0, 0, 0b00, 0b1111, 0, 0, (outs DPR_VFP2:$dst),
- (ins DPR_VFP2:$src1, DPR_VFP2:$src2), N3RegFrm, IIC_VBIND,
- "vmin", "f32", "$dst, $src1, $src2", "", []>;
-def : N3VSPat<NEONfmin, VMINfd_sfp>;
-
-// Vector Convert between single-precision FP and integer
-let neverHasSideEffects = 1 in
-def VCVTf2sd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32",
- v2i32, v2f32, fp_to_sint>;
-def : N2VSPat<arm_ftosi, f32, v2f32, VCVTf2sd_sfp>;
-
-let neverHasSideEffects = 1 in
-def VCVTf2ud_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32",
- v2i32, v2f32, fp_to_uint>;
-def : N2VSPat<arm_ftoui, f32, v2f32, VCVTf2ud_sfp>;
-
-let neverHasSideEffects = 1 in
-def VCVTs2fd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32",
- v2f32, v2i32, sint_to_fp>;
-def : N2VSPat<arm_sitof, f32, v2i32, VCVTs2fd_sfp>;
-
-let neverHasSideEffects = 1 in
-def VCVTu2fd_sfp : N2VS<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32",
- v2f32, v2i32, uint_to_fp>;
-def : N2VSPat<arm_uitof, f32, v2i32, VCVTu2fd_sfp>;
+ (EXTRACT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (Inst
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$acc, ssub_0),
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$a, ssub_0),
+ (INSERT_SUBREG
+ (v2f32 (COPY_TO_REGCLASS (v2f32 (IMPLICIT_DEF)), DPR_VFP2)),
+ SPR:$b, ssub_0)), DPR_VFP2)), ssub_0)>;
+
+def : N3VSPat<fadd, VADDfd>;
+def : N3VSPat<fsub, VSUBfd>;
+def : N3VSPat<fmul, VMULfd>;
+def : N3VSMulOpPat<fmul, fadd, VMLAfd>,
+ Requires<[HasNEON, UseNEONForFP, UseFPVMLx]>;
+def : N3VSMulOpPat<fmul, fsub, VMLSfd>,
+ Requires<[HasNEON, UseNEONForFP, UseFPVMLx]>;
+def : N2VSPat<fabs, VABSfd>;
+def : N2VSPat<fneg, VNEGfd>;
+def : N3VSPat<NEONfmax, VMAXfd>;
+def : N3VSPat<NEONfmin, VMINfd>;
+def : N2VSPat<arm_ftosi, VCVTf2sd>;
+def : N2VSPat<arm_ftoui, VCVTf2ud>;
+def : N2VSPat<arm_sitof, VCVTs2fd>;
+def : N2VSPat<arm_uitof, VCVTu2fd>;
//===----------------------------------------------------------------------===//
// Non-Instruction Patterns