aboutsummaryrefslogtreecommitdiff
path: root/llvm/lib/Target/ARM
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/lib/Target/ARM')
-rw-r--r--llvm/lib/Target/ARM/ARMISelLowering.cpp111
-rw-r--r--llvm/lib/Target/ARM/ARMInstrInfo.td6
-rw-r--r--llvm/lib/Target/ARM/ARMInstrVFP.td175
3 files changed, 165 insertions, 127 deletions
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
index 313ae3d..a4d3d62 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -601,10 +601,20 @@ ARMTargetLowering::ARMTargetLowering(const TargetMachine &TM_,
setOperationAction(ISD::FP_TO_SINT_SAT, MVT::i64, Custom);
setOperationAction(ISD::FP_TO_UINT_SAT, MVT::i64, Custom);
- if (!Subtarget->hasVFP2Base())
+ if (!Subtarget->hasVFP2Base()) {
setAllExpand(MVT::f32);
- if (!Subtarget->hasFP64())
+ } else {
+ for (auto Op : {ISD::STRICT_FADD, ISD::STRICT_FSUB, ISD::STRICT_FMUL,
+ ISD::STRICT_FDIV, ISD::STRICT_FMA, ISD::STRICT_FSQRT})
+ setOperationAction(Op, MVT::f32, Legal);
+ }
+ if (!Subtarget->hasFP64()) {
setAllExpand(MVT::f64);
+ } else {
+ for (auto Op : {ISD::STRICT_FADD, ISD::STRICT_FSUB, ISD::STRICT_FMUL,
+ ISD::STRICT_FDIV, ISD::STRICT_FMA, ISD::STRICT_FSQRT})
+ setOperationAction(Op, MVT::f64, Legal);
+ }
}
if (Subtarget->hasFullFP16()) {
@@ -1281,12 +1291,16 @@ ARMTargetLowering::ARMTargetLowering(const TargetMachine &TM_,
if (!Subtarget->hasFPARMv8Base() || !Subtarget->hasFP64()) {
setOperationAction(ISD::FP16_TO_FP, MVT::f64, Expand);
setOperationAction(ISD::FP_TO_FP16, MVT::f64, Expand);
+ setOperationAction(ISD::STRICT_FP16_TO_FP, MVT::f64, LibCall);
+ setOperationAction(ISD::STRICT_FP_TO_FP16, MVT::f64, LibCall);
}
// fp16 is a special v7 extension that adds f16 <-> f32 conversions.
if (!Subtarget->hasFP16()) {
setOperationAction(ISD::FP16_TO_FP, MVT::f32, Expand);
setOperationAction(ISD::FP_TO_FP16, MVT::f32, Expand);
+ setOperationAction(ISD::STRICT_FP16_TO_FP, MVT::f32, LibCall);
+ setOperationAction(ISD::STRICT_FP_TO_FP16, MVT::f32, LibCall);
}
// Strict floating-point comparisons need custom lowering.
@@ -1298,12 +1312,8 @@ ARMTargetLowering::ARMTargetLowering(const TargetMachine &TM_,
setOperationAction(ISD::STRICT_FSETCCS, MVT::f64, Custom);
}
- // Use __sincos_stret if available.
- if (getLibcallName(RTLIB::SINCOS_STRET_F32) != nullptr &&
- getLibcallName(RTLIB::SINCOS_STRET_F64) != nullptr) {
- setOperationAction(ISD::FSINCOS, MVT::f64, Custom);
- setOperationAction(ISD::FSINCOS, MVT::f32, Custom);
- }
+ setOperationAction(ISD::FSINCOS, MVT::f64, Custom);
+ setOperationAction(ISD::FSINCOS, MVT::f32, Custom);
// FP-ARMv8 implements a lot of rounding-like FP operations.
if (Subtarget->hasFPARMv8Base()) {
@@ -1337,31 +1347,42 @@ ARMTargetLowering::ARMTargetLowering(const TargetMachine &TM_,
}
// FP16 often need to be promoted to call lib functions
+ // clang-format off
if (Subtarget->hasFullFP16()) {
- setOperationAction(ISD::FREM, MVT::f16, Promote);
- setOperationAction(ISD::FCOPYSIGN, MVT::f16, Expand);
- setOperationAction(ISD::FSIN, MVT::f16, Promote);
- setOperationAction(ISD::FCOS, MVT::f16, Promote);
- setOperationAction(ISD::FTAN, MVT::f16, Promote);
- setOperationAction(ISD::FSINCOS, MVT::f16, Promote);
- setOperationAction(ISD::FPOWI, MVT::f16, Promote);
- setOperationAction(ISD::FPOW, MVT::f16, Promote);
- setOperationAction(ISD::FEXP, MVT::f16, Promote);
- setOperationAction(ISD::FEXP2, MVT::f16, Promote);
- setOperationAction(ISD::FEXP10, MVT::f16, Promote);
- setOperationAction(ISD::FLOG, MVT::f16, Promote);
- setOperationAction(ISD::FLOG10, MVT::f16, Promote);
- setOperationAction(ISD::FLOG2, MVT::f16, Promote);
setOperationAction(ISD::LRINT, MVT::f16, Expand);
setOperationAction(ISD::LROUND, MVT::f16, Expand);
-
- setOperationAction(ISD::FROUND, MVT::f16, Legal);
- setOperationAction(ISD::FROUNDEVEN, MVT::f16, Legal);
- setOperationAction(ISD::FTRUNC, MVT::f16, Legal);
- setOperationAction(ISD::FNEARBYINT, MVT::f16, Legal);
- setOperationAction(ISD::FRINT, MVT::f16, Legal);
- setOperationAction(ISD::FFLOOR, MVT::f16, Legal);
- setOperationAction(ISD::FCEIL, MVT::f16, Legal);
+ setOperationAction(ISD::FCOPYSIGN, MVT::f16, Expand);
+
+ for (auto Op : {ISD::FREM, ISD::FPOW, ISD::FPOWI,
+ ISD::FCOS, ISD::FSIN, ISD::FSINCOS,
+ ISD::FSINCOSPI, ISD::FMODF, ISD::FACOS,
+ ISD::FASIN, ISD::FATAN, ISD::FATAN2,
+ ISD::FCOSH, ISD::FSINH, ISD::FTANH,
+ ISD::FTAN, ISD::FEXP, ISD::FEXP2,
+ ISD::FEXP10, ISD::FLOG, ISD::FLOG2,
+ ISD::FLOG10, ISD::STRICT_FREM, ISD::STRICT_FPOW,
+ ISD::STRICT_FPOWI, ISD::STRICT_FCOS, ISD::STRICT_FSIN,
+ ISD::STRICT_FACOS, ISD::STRICT_FASIN, ISD::STRICT_FATAN,
+ ISD::STRICT_FATAN2, ISD::STRICT_FCOSH, ISD::STRICT_FSINH,
+ ISD::STRICT_FTANH, ISD::STRICT_FEXP, ISD::STRICT_FEXP2,
+ ISD::STRICT_FLOG, ISD::STRICT_FLOG2, ISD::STRICT_FLOG10,
+ ISD::STRICT_FTAN}) {
+ setOperationAction(Op, MVT::f16, Promote);
+ }
+
+ // Round-to-integer need custom lowering for fp16, as Promote doesn't work
+ // because the result type is integer.
+ for (auto Op : {ISD::STRICT_LROUND, ISD::STRICT_LLROUND, ISD::STRICT_LRINT, ISD::STRICT_LLRINT})
+ setOperationAction(Op, MVT::f16, Custom);
+
+ for (auto Op : {ISD::FROUND, ISD::FROUNDEVEN, ISD::FTRUNC,
+ ISD::FNEARBYINT, ISD::FRINT, ISD::FFLOOR,
+ ISD::FCEIL, ISD::STRICT_FROUND, ISD::STRICT_FROUNDEVEN,
+ ISD::STRICT_FTRUNC, ISD::STRICT_FNEARBYINT, ISD::STRICT_FRINT,
+ ISD::STRICT_FFLOOR, ISD::STRICT_FCEIL}) {
+ setOperationAction(Op, MVT::f16, Legal);
+ }
+ // clang-format on
}
if (Subtarget->hasNEON()) {
@@ -9835,13 +9856,18 @@ static SDValue LowerUADDSUBO_CARRY(SDValue Op, SelectionDAG &DAG) {
}
SDValue ARMTargetLowering::LowerFSINCOS(SDValue Op, SelectionDAG &DAG) const {
- assert(Subtarget->isTargetDarwin());
-
// For iOS, we want to call an alternative entry point: __sincos_stret,
// return values are passed via sret.
SDLoc dl(Op);
SDValue Arg = Op.getOperand(0);
EVT ArgVT = Arg.getValueType();
+ RTLIB::Libcall LC = RTLIB::getSINCOS_STRET(ArgVT);
+ RTLIB::LibcallImpl SincosStret = getLibcallImpl(LC);
+ if (SincosStret == RTLIB::Unsupported)
+ return SDValue();
+
+ assert(Subtarget->isTargetDarwin());
+
Type *ArgTy = ArgVT.getTypeForEVT(*DAG.getContext());
auto PtrVT = getPointerTy(DAG.getDataLayout());
@@ -9871,11 +9897,9 @@ SDValue ARMTargetLowering::LowerFSINCOS(SDValue Op, SelectionDAG &DAG) const {
Args.emplace_back(Arg, ArgTy);
- RTLIB::Libcall LC =
- (ArgVT == MVT::f64) ? RTLIB::SINCOS_STRET_F64 : RTLIB::SINCOS_STRET_F32;
- const char *LibcallName = getLibcallName(LC);
- CallingConv::ID CC = getLibcallCallingConv(LC);
- SDValue Callee = DAG.getExternalSymbol(LibcallName, getPointerTy(DL));
+ StringRef LibcallName = getLibcallImplName(SincosStret);
+ CallingConv::ID CC = getLibcallImplCallingConv(SincosStret);
+ SDValue Callee = DAG.getExternalSymbol(LibcallName.data(), getPointerTy(DL));
TargetLowering::CallLoweringInfo CLI(DAG);
CLI.setDebugLoc(dl)
@@ -10726,6 +10750,19 @@ SDValue ARMTargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
return LowerCMP(Op, DAG);
case ISD::ABS:
return LowerABS(Op, DAG);
+ case ISD::STRICT_LROUND:
+ case ISD::STRICT_LLROUND:
+ case ISD::STRICT_LRINT:
+ case ISD::STRICT_LLRINT: {
+ assert((Op.getOperand(1).getValueType() == MVT::f16 ||
+ Op.getOperand(1).getValueType() == MVT::bf16) &&
+ "Expected custom lowering of rounding operations only for f16");
+ SDLoc DL(Op);
+ SDValue Ext = DAG.getNode(ISD::STRICT_FP_EXTEND, DL, {MVT::f32, MVT::Other},
+ {Op.getOperand(0), Op.getOperand(1)});
+ return DAG.getNode(Op.getOpcode(), DL, {Op.getValueType(), MVT::Other},
+ {Ext.getValue(1), Ext.getValue(0)});
+ }
}
}
diff --git a/llvm/lib/Target/ARM/ARMInstrInfo.td b/llvm/lib/Target/ARM/ARMInstrInfo.td
index 10d4cd5..f7176a6 100644
--- a/llvm/lib/Target/ARM/ARMInstrInfo.td
+++ b/llvm/lib/Target/ARM/ARMInstrInfo.td
@@ -473,15 +473,15 @@ def xor_su : PatFrag<(ops node:$lhs, node:$rhs), (xor node:$lhs, node:$rhs)>;
// An 'fmul' node with a single use.
let HasOneUse = 1 in
-def fmul_su : PatFrag<(ops node:$lhs, node:$rhs), (fmul node:$lhs, node:$rhs)>;
+def fmul_su : PatFrag<(ops node:$lhs, node:$rhs), (any_fmul node:$lhs, node:$rhs)>;
// An 'fadd' node which checks for single non-hazardous use.
-def fadd_mlx : PatFrag<(ops node:$lhs, node:$rhs),(fadd node:$lhs, node:$rhs),[{
+def fadd_mlx : PatFrag<(ops node:$lhs, node:$rhs),(any_fadd node:$lhs, node:$rhs),[{
return hasNoVMLxHazardUse(N);
}]>;
// An 'fsub' node which checks for single non-hazardous use.
-def fsub_mlx : PatFrag<(ops node:$lhs, node:$rhs),(fsub node:$lhs, node:$rhs),[{
+def fsub_mlx : PatFrag<(ops node:$lhs, node:$rhs),(any_fsub node:$lhs, node:$rhs),[{
return hasNoVMLxHazardUse(N);
}]>;
diff --git a/llvm/lib/Target/ARM/ARMInstrVFP.td b/llvm/lib/Target/ARM/ARMInstrVFP.td
index 6771106..e2cc97b 100644
--- a/llvm/lib/Target/ARM/ARMInstrVFP.td
+++ b/llvm/lib/Target/ARM/ARMInstrVFP.td
@@ -439,14 +439,14 @@ let TwoOperandAliasConstraint = "$Dn = $Dd", mayRaiseFPException = 1, Uses = [FP
def VADDD : ADbI<0b11100, 0b11, 0, 0,
(outs DPR:$Dd), (ins DPR:$Dn, DPR:$Dm),
IIC_fpALU64, "vadd", ".f64\t$Dd, $Dn, $Dm",
- [(set DPR:$Dd, (fadd DPR:$Dn, (f64 DPR:$Dm)))]>,
+ [(set DPR:$Dd, (any_fadd DPR:$Dn, (f64 DPR:$Dm)))]>,
Sched<[WriteFPALU64]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VADDS : ASbIn<0b11100, 0b11, 0, 0,
(outs SPR:$Sd), (ins SPR:$Sn, SPR:$Sm),
IIC_fpALU32, "vadd", ".f32\t$Sd, $Sn, $Sm",
- [(set SPR:$Sd, (fadd SPR:$Sn, SPR:$Sm))]>,
+ [(set SPR:$Sd, (any_fadd SPR:$Sn, SPR:$Sm))]>,
Sched<[WriteFPALU32]> {
// Some single precision VFP instructions may be executed on both NEON and
// VFP pipelines on A8.
@@ -457,21 +457,21 @@ let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FP
def VADDH : AHbI<0b11100, 0b11, 0, 0,
(outs HPR:$Sd), (ins HPR:$Sn, HPR:$Sm),
IIC_fpALU16, "vadd", ".f16\t$Sd, $Sn, $Sm",
- [(set (f16 HPR:$Sd), (fadd (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
+ [(set (f16 HPR:$Sd), (any_fadd (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
Sched<[WriteFPALU32]>;
let TwoOperandAliasConstraint = "$Dn = $Dd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VSUBD : ADbI<0b11100, 0b11, 1, 0,
(outs DPR:$Dd), (ins DPR:$Dn, DPR:$Dm),
IIC_fpALU64, "vsub", ".f64\t$Dd, $Dn, $Dm",
- [(set DPR:$Dd, (fsub DPR:$Dn, (f64 DPR:$Dm)))]>,
+ [(set DPR:$Dd, (any_fsub DPR:$Dn, (f64 DPR:$Dm)))]>,
Sched<[WriteFPALU64]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VSUBS : ASbIn<0b11100, 0b11, 1, 0,
(outs SPR:$Sd), (ins SPR:$Sn, SPR:$Sm),
IIC_fpALU32, "vsub", ".f32\t$Sd, $Sn, $Sm",
- [(set SPR:$Sd, (fsub SPR:$Sn, SPR:$Sm))]>,
+ [(set SPR:$Sd, (any_fsub SPR:$Sn, SPR:$Sm))]>,
Sched<[WriteFPALU32]>{
// Some single precision VFP instructions may be executed on both NEON and
// VFP pipelines on A8.
@@ -482,42 +482,42 @@ let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FP
def VSUBH : AHbI<0b11100, 0b11, 1, 0,
(outs HPR:$Sd), (ins HPR:$Sn, HPR:$Sm),
IIC_fpALU16, "vsub", ".f16\t$Sd, $Sn, $Sm",
- [(set (f16 HPR:$Sd), (fsub (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
+ [(set (f16 HPR:$Sd), (any_fsub (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
Sched<[WriteFPALU32]>;
let TwoOperandAliasConstraint = "$Dn = $Dd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VDIVD : ADbI<0b11101, 0b00, 0, 0,
(outs DPR:$Dd), (ins DPR:$Dn, DPR:$Dm),
IIC_fpDIV64, "vdiv", ".f64\t$Dd, $Dn, $Dm",
- [(set DPR:$Dd, (fdiv DPR:$Dn, (f64 DPR:$Dm)))]>,
+ [(set DPR:$Dd, (any_fdiv DPR:$Dn, (f64 DPR:$Dm)))]>,
Sched<[WriteFPDIV64]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VDIVS : ASbI<0b11101, 0b00, 0, 0,
(outs SPR:$Sd), (ins SPR:$Sn, SPR:$Sm),
IIC_fpDIV32, "vdiv", ".f32\t$Sd, $Sn, $Sm",
- [(set SPR:$Sd, (fdiv SPR:$Sn, SPR:$Sm))]>,
+ [(set SPR:$Sd, (any_fdiv SPR:$Sn, SPR:$Sm))]>,
Sched<[WriteFPDIV32]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VDIVH : AHbI<0b11101, 0b00, 0, 0,
(outs HPR:$Sd), (ins HPR:$Sn, HPR:$Sm),
IIC_fpDIV16, "vdiv", ".f16\t$Sd, $Sn, $Sm",
- [(set (f16 HPR:$Sd), (fdiv (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
+ [(set (f16 HPR:$Sd), (any_fdiv (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
Sched<[WriteFPDIV32]>;
let TwoOperandAliasConstraint = "$Dn = $Dd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VMULD : ADbI<0b11100, 0b10, 0, 0,
(outs DPR:$Dd), (ins DPR:$Dn, DPR:$Dm),
IIC_fpMUL64, "vmul", ".f64\t$Dd, $Dn, $Dm",
- [(set DPR:$Dd, (fmul DPR:$Dn, (f64 DPR:$Dm)))]>,
+ [(set DPR:$Dd, (any_fmul DPR:$Dn, (f64 DPR:$Dm)))]>,
Sched<[WriteFPMUL64, ReadFPMUL, ReadFPMUL]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VMULS : ASbIn<0b11100, 0b10, 0, 0,
(outs SPR:$Sd), (ins SPR:$Sn, SPR:$Sm),
IIC_fpMUL32, "vmul", ".f32\t$Sd, $Sn, $Sm",
- [(set SPR:$Sd, (fmul SPR:$Sn, SPR:$Sm))]>,
+ [(set SPR:$Sd, (any_fmul SPR:$Sn, SPR:$Sm))]>,
Sched<[WriteFPMUL32, ReadFPMUL, ReadFPMUL]> {
// Some single precision VFP instructions may be executed on both NEON and
// VFP pipelines on A8.
@@ -528,21 +528,21 @@ let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FP
def VMULH : AHbI<0b11100, 0b10, 0, 0,
(outs HPR:$Sd), (ins HPR:$Sn, HPR:$Sm),
IIC_fpMUL16, "vmul", ".f16\t$Sd, $Sn, $Sm",
- [(set (f16 HPR:$Sd), (fmul (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
+ [(set (f16 HPR:$Sd), (any_fmul (f16 HPR:$Sn), (f16 HPR:$Sm)))]>,
Sched<[WriteFPMUL32, ReadFPMUL, ReadFPMUL]>;
let TwoOperandAliasConstraint = "$Dn = $Dd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VNMULD : ADbI<0b11100, 0b10, 1, 0,
(outs DPR:$Dd), (ins DPR:$Dn, DPR:$Dm),
IIC_fpMUL64, "vnmul", ".f64\t$Dd, $Dn, $Dm",
- [(set DPR:$Dd, (fneg (fmul DPR:$Dn, (f64 DPR:$Dm))))]>,
+ [(set DPR:$Dd, (fneg (any_fmul DPR:$Dn, (f64 DPR:$Dm))))]>,
Sched<[WriteFPMUL64, ReadFPMUL, ReadFPMUL]>;
let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VNMULS : ASbI<0b11100, 0b10, 1, 0,
(outs SPR:$Sd), (ins SPR:$Sn, SPR:$Sm),
IIC_fpMUL32, "vnmul", ".f32\t$Sd, $Sn, $Sm",
- [(set SPR:$Sd, (fneg (fmul SPR:$Sn, SPR:$Sm)))]>,
+ [(set SPR:$Sd, (fneg (any_fmul SPR:$Sn, SPR:$Sm)))]>,
Sched<[WriteFPMUL32, ReadFPMUL, ReadFPMUL]> {
// Some single precision VFP instructions may be executed on both NEON and
// VFP pipelines on A8.
@@ -553,7 +553,7 @@ let TwoOperandAliasConstraint = "$Sn = $Sd", mayRaiseFPException = 1, Uses = [FP
def VNMULH : AHbI<0b11100, 0b10, 1, 0,
(outs HPR:$Sd), (ins HPR:$Sn, HPR:$Sm),
IIC_fpMUL16, "vnmul", ".f16\t$Sd, $Sn, $Sm",
- [(set (f16 HPR:$Sd), (fneg (fmul (f16 HPR:$Sn), (f16 HPR:$Sm))))]>,
+ [(set (f16 HPR:$Sd), (fneg (any_fmul (f16 HPR:$Sn), (f16 HPR:$Sm))))]>,
Sched<[WriteFPMUL32, ReadFPMUL, ReadFPMUL]>;
multiclass vsel_inst<string op, bits<2> opc, int CC> {
@@ -587,7 +587,7 @@ defm VSELGE : vsel_inst<"ge", 0b10, 10>;
defm VSELEQ : vsel_inst<"eq", 0b00, 0>;
defm VSELVS : vsel_inst<"vs", 0b01, 6>;
-multiclass vmaxmin_inst<string op, bit opc, SDNode SD> {
+multiclass vmaxmin_inst<string op, bit opc, PatFrags SD> {
let DecoderNamespace = "VFPV8", PostEncoderMethod = "",
isUnpredicable = 1, mayRaiseFPException = 1 in {
def H : AHbInp<0b11101, 0b00, opc,
@@ -610,8 +610,8 @@ multiclass vmaxmin_inst<string op, bit opc, SDNode SD> {
}
}
-defm VFP_VMAXNM : vmaxmin_inst<"vmaxnm", 0, fmaxnum>;
-defm VFP_VMINNM : vmaxmin_inst<"vminnm", 1, fminnum>;
+defm VFP_VMAXNM : vmaxmin_inst<"vmaxnm", 0, any_fmaxnum>;
+defm VFP_VMINNM : vmaxmin_inst<"vminnm", 1, any_fminnum>;
// Match reassociated forms only if not sign dependent rounding.
def : Pat<(fmul (fneg DPR:$a), (f64 DPR:$b)),
@@ -746,7 +746,7 @@ let mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VCVTDS : ASuI<0b11101, 0b11, 0b0111, 0b11, 0,
(outs DPR:$Dd), (ins SPR:$Sm),
IIC_fpCVTDS, "vcvt", ".f64.f32\t$Dd, $Sm", "",
- [(set DPR:$Dd, (fpextend SPR:$Sm))]>,
+ [(set DPR:$Dd, (any_fpextend SPR:$Sm))]>,
Sched<[WriteFPCVT]> {
// Instruction operands.
bits<5> Dd;
@@ -766,7 +766,7 @@ def VCVTDS : ASuI<0b11101, 0b11, 0b0111, 0b11, 0,
let mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VCVTSD : VFPAI<(outs SPR:$Sd), (ins DPR:$Dm), VFPUnaryFrm,
IIC_fpCVTSD, "vcvt", ".f32.f64\t$Sd, $Dm", "",
- [(set SPR:$Sd, (fpround DPR:$Dm))]>,
+ [(set SPR:$Sd, (any_fpround DPR:$Dm))]>,
Sched<[WriteFPCVT]> {
// Instruction operands.
bits<5> Sd;
@@ -796,7 +796,7 @@ def VCVTBHS: ASuI<0b11101, 0b11, 0b0010, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sm),
Requires<[HasFP16]>,
Sched<[WriteFPCVT]>;
-def : FP16Pat<(f32 (fpextend (f16 HPR:$Sm))),
+def : FP16Pat<(f32 (any_fpextend (f16 HPR:$Sm))),
(VCVTBHS (COPY_TO_REGCLASS (f16 HPR:$Sm), SPR))>;
def : FP16Pat<(f16_to_fp GPR:$a),
(VCVTBHS (COPY_TO_REGCLASS GPR:$a, SPR))>;
@@ -808,16 +808,16 @@ def VCVTBSH: ASuI<0b11101, 0b11, 0b0011, 0b01, 0, (outs SPR:$Sd), (ins SPR:$Sda,
Requires<[HasFP16]>,
Sched<[WriteFPCVT]>;
-def : FP16Pat<(f16 (fpround SPR:$Sm)),
+def : FP16Pat<(f16 (any_fpround SPR:$Sm)),
(COPY_TO_REGCLASS (VCVTBSH (IMPLICIT_DEF), SPR:$Sm), HPR)>;
def : FP16Pat<(fp_to_f16 SPR:$a),
(i32 (COPY_TO_REGCLASS (VCVTBSH (IMPLICIT_DEF), SPR:$a), GPR))>;
-def : FP16Pat<(insertelt (v8f16 MQPR:$src1), (f16 (fpround (f32 SPR:$src2))), imm_even:$lane),
+def : FP16Pat<(insertelt (v8f16 MQPR:$src1), (f16 (any_fpround (f32 SPR:$src2))), imm_even:$lane),
(v8f16 (INSERT_SUBREG (v8f16 MQPR:$src1),
(VCVTBSH (EXTRACT_SUBREG (v8f16 MQPR:$src1), (SSubReg_f16_reg imm:$lane)),
SPR:$src2),
(SSubReg_f16_reg imm:$lane)))>;
-def : FP16Pat<(insertelt (v4f16 DPR:$src1), (f16 (fpround (f32 SPR:$src2))), imm_even:$lane),
+def : FP16Pat<(insertelt (v4f16 DPR:$src1), (f16 (any_fpround (f32 SPR:$src2))), imm_even:$lane),
(v4f16 (INSERT_SUBREG (v4f16 DPR:$src1),
(VCVTBSH (EXTRACT_SUBREG (v4f16 DPR:$src1), (SSubReg_f16_reg imm:$lane)),
SPR:$src2),
@@ -830,9 +830,9 @@ def VCVTTHS: ASuI<0b11101, 0b11, 0b0010, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sm),
Requires<[HasFP16]>,
Sched<[WriteFPCVT]>;
-def : FP16Pat<(f32 (fpextend (extractelt (v8f16 MQPR:$src), imm_odd:$lane))),
+def : FP16Pat<(f32 (any_fpextend (extractelt (v8f16 MQPR:$src), imm_odd:$lane))),
(VCVTTHS (EXTRACT_SUBREG MQPR:$src, (SSubReg_f16_reg imm_odd:$lane)))>;
-def : FP16Pat<(f32 (fpextend (extractelt (v4f16 DPR:$src), imm_odd:$lane))),
+def : FP16Pat<(f32 (any_fpextend (extractelt (v4f16 DPR:$src), imm_odd:$lane))),
(VCVTTHS (EXTRACT_SUBREG
(v2f32 (COPY_TO_REGCLASS (v4f16 DPR:$src), DPR_VFP2)),
(SSubReg_f16_reg imm_odd:$lane)))>;
@@ -844,12 +844,12 @@ def VCVTTSH: ASuI<0b11101, 0b11, 0b0011, 0b11, 0, (outs SPR:$Sd), (ins SPR:$Sda,
Requires<[HasFP16]>,
Sched<[WriteFPCVT]>;
-def : FP16Pat<(insertelt (v8f16 MQPR:$src1), (f16 (fpround (f32 SPR:$src2))), imm_odd:$lane),
+def : FP16Pat<(insertelt (v8f16 MQPR:$src1), (f16 (any_fpround (f32 SPR:$src2))), imm_odd:$lane),
(v8f16 (INSERT_SUBREG (v8f16 MQPR:$src1),
(VCVTTSH (EXTRACT_SUBREG (v8f16 MQPR:$src1), (SSubReg_f16_reg imm:$lane)),
SPR:$src2),
(SSubReg_f16_reg imm:$lane)))>;
-def : FP16Pat<(insertelt (v4f16 DPR:$src1), (f16 (fpround (f32 SPR:$src2))), imm_odd:$lane),
+def : FP16Pat<(insertelt (v4f16 DPR:$src1), (f16 (any_fpround (f32 SPR:$src2))), imm_odd:$lane),
(v4f16 (INSERT_SUBREG (v4f16 DPR:$src1),
(VCVTTSH (EXTRACT_SUBREG (v4f16 DPR:$src1), (SSubReg_f16_reg imm:$lane)),
SPR:$src2),
@@ -872,7 +872,7 @@ def VCVTBHD : ADuI<0b11101, 0b11, 0b0010, 0b01, 0,
let hasSideEffects = 0;
}
-def : FullFP16Pat<(f64 (fpextend (f16 HPR:$Sm))),
+def : FullFP16Pat<(f64 (any_fpextend (f16 HPR:$Sm))),
(VCVTBHD (COPY_TO_REGCLASS (f16 HPR:$Sm), SPR))>,
Requires<[HasFPARMv8, HasDPVFP]>;
def : FP16Pat<(f64 (f16_to_fp GPR:$a)),
@@ -898,7 +898,7 @@ def VCVTBDH : ADuI<0b11101, 0b11, 0b0011, 0b01, 0,
let hasSideEffects = 0;
}
-def : FullFP16Pat<(f16 (fpround DPR:$Dm)),
+def : FullFP16Pat<(f16 (any_fpround DPR:$Dm)),
(COPY_TO_REGCLASS (VCVTBDH (IMPLICIT_DEF), DPR:$Dm), HPR)>,
Requires<[HasFPARMv8, HasDPVFP]>;
def : FP16Pat<(fp_to_f16 (f64 DPR:$a)),
@@ -1007,41 +1007,41 @@ multiclass vcvt_inst<string opc, bits<2> rm,
let Predicates = [HasFPARMv8] in {
let Predicates = [HasFullFP16] in {
- def : Pat<(i32 (fp_to_sint (node (f16 HPR:$a)))),
+ def : Pat<(i32 (any_fp_to_sint (node (f16 HPR:$a)))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"SH") (f16 HPR:$a)),
GPR)>;
- def : Pat<(i32 (fp_to_uint (node (f16 HPR:$a)))),
+ def : Pat<(i32 (any_fp_to_uint (node (f16 HPR:$a)))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"UH") (f16 HPR:$a)),
GPR)>;
}
- def : Pat<(i32 (fp_to_sint (node SPR:$a))),
+ def : Pat<(i32 (any_fp_to_sint (node SPR:$a))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"SS") SPR:$a),
GPR)>;
- def : Pat<(i32 (fp_to_uint (node SPR:$a))),
+ def : Pat<(i32 (any_fp_to_uint (node SPR:$a))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"US") SPR:$a),
GPR)>;
}
let Predicates = [HasFPARMv8, HasDPVFP] in {
- def : Pat<(i32 (fp_to_sint (node (f64 DPR:$a)))),
+ def : Pat<(i32 (any_fp_to_sint (node (f64 DPR:$a)))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"SD") DPR:$a),
GPR)>;
- def : Pat<(i32 (fp_to_uint (node (f64 DPR:$a)))),
+ def : Pat<(i32 (any_fp_to_uint (node (f64 DPR:$a)))),
(COPY_TO_REGCLASS
(!cast<Instruction>(NAME#"UD") DPR:$a),
GPR)>;
}
}
-defm VCVTA : vcvt_inst<"a", 0b00, fround>;
+defm VCVTA : vcvt_inst<"a", 0b00, any_fround>;
defm VCVTN : vcvt_inst<"n", 0b01>;
-defm VCVTP : vcvt_inst<"p", 0b10, fceil>;
-defm VCVTM : vcvt_inst<"m", 0b11, ffloor>;
+defm VCVTP : vcvt_inst<"p", 0b10, any_fceil>;
+defm VCVTM : vcvt_inst<"m", 0b11, any_ffloor>;
def VNEGD : ADuI<0b11101, 0b11, 0b0001, 0b01, 0,
(outs DPR:$Dd), (ins DPR:$Dm),
@@ -1103,9 +1103,9 @@ multiclass vrint_inst_zrx<string opc, bit op, bit op2, SDPatternOperator node,
Requires<[HasFPARMv8,HasDPVFP]>;
}
-defm VRINTZ : vrint_inst_zrx<"z", 0, 1, ftrunc, [], 0>;
-defm VRINTR : vrint_inst_zrx<"r", 0, 0, fnearbyint, [FPSCR_RM], 0>;
-defm VRINTX : vrint_inst_zrx<"x", 1, 0, frint, [FPSCR_RM], 1>;
+defm VRINTZ : vrint_inst_zrx<"z", 0, 1, any_ftrunc, [], 0>;
+defm VRINTR : vrint_inst_zrx<"r", 0, 0, any_fnearbyint, [FPSCR_RM], 0>;
+defm VRINTX : vrint_inst_zrx<"x", 1, 0, any_frint, [FPSCR_RM], 1>;
multiclass vrint_inst_anpm<string opc, bits<2> rm,
SDPatternOperator node = null_frag> {
@@ -1145,30 +1145,31 @@ multiclass vrint_inst_anpm<string opc, bits<2> rm,
Requires<[HasFPARMv8,HasDPVFP]>;
}
-defm VRINTA : vrint_inst_anpm<"a", 0b00, fround>;
-defm VRINTN : vrint_inst_anpm<"n", 0b01, froundeven>;
-defm VRINTP : vrint_inst_anpm<"p", 0b10, fceil>;
-defm VRINTM : vrint_inst_anpm<"m", 0b11, ffloor>;
+defm VRINTA : vrint_inst_anpm<"a", 0b00, any_fround>;
+defm VRINTN : vrint_inst_anpm<"n", 0b01, any_froundeven>;
+defm VRINTP : vrint_inst_anpm<"p", 0b10, any_fceil>;
+defm VRINTM : vrint_inst_anpm<"m", 0b11, any_ffloor>;
+
let mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VSQRTD : ADuI<0b11101, 0b11, 0b0001, 0b11, 0,
(outs DPR:$Dd), (ins DPR:$Dm),
IIC_fpSQRT64, "vsqrt", ".f64\t$Dd, $Dm", "",
- [(set DPR:$Dd, (fsqrt (f64 DPR:$Dm)))]>,
+ [(set DPR:$Dd, (any_fsqrt (f64 DPR:$Dm)))]>,
Sched<[WriteFPSQRT64]>;
let mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VSQRTS : ASuI<0b11101, 0b11, 0b0001, 0b11, 0,
(outs SPR:$Sd), (ins SPR:$Sm),
IIC_fpSQRT32, "vsqrt", ".f32\t$Sd, $Sm", "",
- [(set SPR:$Sd, (fsqrt SPR:$Sm))]>,
+ [(set SPR:$Sd, (any_fsqrt SPR:$Sm))]>,
Sched<[WriteFPSQRT32]>;
let mayRaiseFPException = 1, Uses = [FPSCR_RM] in
def VSQRTH : AHuI<0b11101, 0b11, 0b0001, 0b11, 0,
(outs HPR:$Sd), (ins HPR:$Sm),
IIC_fpSQRT16, "vsqrt", ".f16\t$Sd, $Sm",
- [(set (f16 HPR:$Sd), (fsqrt (f16 HPR:$Sm)))]>;
+ [(set (f16 HPR:$Sd), (any_fsqrt (f16 HPR:$Sm)))]>;
let hasSideEffects = 0 in {
let isMoveReg = 1 in {
@@ -1509,10 +1510,10 @@ def VSITOD : AVConv1IDs_Encode<0b11101, 0b11, 0b1000, 0b1011,
}
let Predicates=[HasVFP2, HasDPVFP] in {
- def : VFPPat<(f64 (sint_to_fp GPR:$a)),
+ def : VFPPat<(f64 (any_sint_to_fp GPR:$a)),
(VSITOD (COPY_TO_REGCLASS GPR:$a, SPR))>;
- def : VFPPat<(f64 (sint_to_fp (i32 (alignedload32 addrmode5:$a)))),
+ def : VFPPat<(f64 (any_sint_to_fp (i32 (alignedload32 addrmode5:$a)))),
(VSITOD (VLDRS addrmode5:$a))>;
}
@@ -1529,10 +1530,10 @@ def VSITOS : AVConv1InSs_Encode<0b11101, 0b11, 0b1000, 0b1010,
let D = VFPNeonA8Domain;
}
-def : VFPNoNEONPat<(f32 (sint_to_fp GPR:$a)),
+def : VFPNoNEONPat<(f32 (any_sint_to_fp GPR:$a)),
(VSITOS (COPY_TO_REGCLASS GPR:$a, SPR))>;
-def : VFPNoNEONPat<(f32 (sint_to_fp (i32 (alignedload32 addrmode5:$a)))),
+def : VFPNoNEONPat<(f32 (any_sint_to_fp (i32 (alignedload32 addrmode5:$a)))),
(VSITOS (VLDRS addrmode5:$a))>;
let mayRaiseFPException = 1 in
@@ -1545,7 +1546,7 @@ def VSITOH : AVConv1IHs_Encode<0b11101, 0b11, 0b1000, 0b1001,
let isUnpredicable = 1;
}
-def : VFPNoNEONPat<(f16 (sint_to_fp GPR:$a)),
+def : VFPNoNEONPat<(f16 (any_sint_to_fp GPR:$a)),
(VSITOH (COPY_TO_REGCLASS GPR:$a, SPR))>;
let mayRaiseFPException = 1 in
@@ -1558,10 +1559,10 @@ def VUITOD : AVConv1IDs_Encode<0b11101, 0b11, 0b1000, 0b1011,
}
let Predicates=[HasVFP2, HasDPVFP] in {
- def : VFPPat<(f64 (uint_to_fp GPR:$a)),
+ def : VFPPat<(f64 (any_uint_to_fp GPR:$a)),
(VUITOD (COPY_TO_REGCLASS GPR:$a, SPR))>;
- def : VFPPat<(f64 (uint_to_fp (i32 (alignedload32 addrmode5:$a)))),
+ def : VFPPat<(f64 (any_uint_to_fp (i32 (alignedload32 addrmode5:$a)))),
(VUITOD (VLDRS addrmode5:$a))>;
}
@@ -1578,10 +1579,10 @@ def VUITOS : AVConv1InSs_Encode<0b11101, 0b11, 0b1000, 0b1010,
let D = VFPNeonA8Domain;
}
-def : VFPNoNEONPat<(f32 (uint_to_fp GPR:$a)),
+def : VFPNoNEONPat<(f32 (any_uint_to_fp GPR:$a)),
(VUITOS (COPY_TO_REGCLASS GPR:$a, SPR))>;
-def : VFPNoNEONPat<(f32 (uint_to_fp (i32 (alignedload32 addrmode5:$a)))),
+def : VFPNoNEONPat<(f32 (any_uint_to_fp (i32 (alignedload32 addrmode5:$a)))),
(VUITOS (VLDRS addrmode5:$a))>;
let mayRaiseFPException = 1 in
@@ -1594,7 +1595,7 @@ def VUITOH : AVConv1IHs_Encode<0b11101, 0b11, 0b1000, 0b1001,
let isUnpredicable = 1;
}
-def : VFPNoNEONPat<(f16 (uint_to_fp GPR:$a)),
+def : VFPNoNEONPat<(f16 (any_uint_to_fp GPR:$a)),
(VUITOH (COPY_TO_REGCLASS GPR:$a, SPR))>;
// FP -> Int:
@@ -1669,12 +1670,12 @@ def VTOSIZD : AVConv1IsD_Encode<0b11101, 0b11, 0b1101, 0b1011,
}
let Predicates=[HasVFP2, HasDPVFP] in {
- def : VFPPat<(i32 (fp_to_sint (f64 DPR:$a))),
+ def : VFPPat<(i32 (any_fp_to_sint (f64 DPR:$a))),
(COPY_TO_REGCLASS (VTOSIZD DPR:$a), GPR)>;
def : VFPPat<(i32 (fp_to_sint_sat (f64 DPR:$a), i32)),
(COPY_TO_REGCLASS (VTOSIZD DPR:$a), GPR)>;
- def : VFPPat<(alignedstore32 (i32 (fp_to_sint (f64 DPR:$a))), addrmode5:$ptr),
+ def : VFPPat<(alignedstore32 (i32 (any_fp_to_sint (f64 DPR:$a))), addrmode5:$ptr),
(VSTRS (VTOSIZD DPR:$a), addrmode5:$ptr)>;
def : VFPPat<(alignedstore32 (i32 (fp_to_sint_sat (f64 DPR:$a), i32)), addrmode5:$ptr),
(VSTRS (VTOSIZD DPR:$a), addrmode5:$ptr)>;
@@ -1693,12 +1694,12 @@ def VTOSIZS : AVConv1InsS_Encode<0b11101, 0b11, 0b1101, 0b1010,
let D = VFPNeonA8Domain;
}
-def : VFPNoNEONPat<(i32 (fp_to_sint SPR:$a)),
+def : VFPNoNEONPat<(i32 (any_fp_to_sint SPR:$a)),
(COPY_TO_REGCLASS (VTOSIZS SPR:$a), GPR)>;
def : VFPPat<(i32 (fp_to_sint_sat SPR:$a, i32)),
(COPY_TO_REGCLASS (VTOSIZS SPR:$a), GPR)>;
-def : VFPNoNEONPat<(alignedstore32 (i32 (fp_to_sint (f32 SPR:$a))),
+def : VFPNoNEONPat<(alignedstore32 (i32 (any_fp_to_sint (f32 SPR:$a))),
addrmode5:$ptr),
(VSTRS (VTOSIZS SPR:$a), addrmode5:$ptr)>;
def : VFPPat<(alignedstore32 (i32 (fp_to_sint_sat (f32 SPR:$a), i32)),
@@ -1715,7 +1716,7 @@ def VTOSIZH : AVConv1IsH_Encode<0b11101, 0b11, 0b1101, 0b1001,
let isUnpredicable = 1;
}
-def : VFPNoNEONPat<(i32 (fp_to_sint (f16 HPR:$a))),
+def : VFPNoNEONPat<(i32 (any_fp_to_sint (f16 HPR:$a))),
(COPY_TO_REGCLASS (VTOSIZH (f16 HPR:$a)), GPR)>;
def : VFPPat<(i32 (fp_to_sint_sat (f16 HPR:$a), i32)),
(COPY_TO_REGCLASS (VTOSIZH (f16 HPR:$a)), GPR)>;
@@ -1730,12 +1731,12 @@ def VTOUIZD : AVConv1IsD_Encode<0b11101, 0b11, 0b1100, 0b1011,
}
let Predicates=[HasVFP2, HasDPVFP] in {
- def : VFPPat<(i32 (fp_to_uint (f64 DPR:$a))),
+ def : VFPPat<(i32 (any_fp_to_uint (f64 DPR:$a))),
(COPY_TO_REGCLASS (VTOUIZD DPR:$a), GPR)>;
def : VFPPat<(i32 (fp_to_uint_sat (f64 DPR:$a), i32)),
(COPY_TO_REGCLASS (VTOUIZD DPR:$a), GPR)>;
- def : VFPPat<(alignedstore32 (i32 (fp_to_uint (f64 DPR:$a))), addrmode5:$ptr),
+ def : VFPPat<(alignedstore32 (i32 (any_fp_to_uint (f64 DPR:$a))), addrmode5:$ptr),
(VSTRS (VTOUIZD DPR:$a), addrmode5:$ptr)>;
def : VFPPat<(alignedstore32 (i32 (fp_to_uint_sat (f64 DPR:$a), i32)), addrmode5:$ptr),
(VSTRS (VTOUIZD DPR:$a), addrmode5:$ptr)>;
@@ -1754,12 +1755,12 @@ def VTOUIZS : AVConv1InsS_Encode<0b11101, 0b11, 0b1100, 0b1010,
let D = VFPNeonA8Domain;
}
-def : VFPNoNEONPat<(i32 (fp_to_uint SPR:$a)),
+def : VFPNoNEONPat<(i32 (any_fp_to_uint SPR:$a)),
(COPY_TO_REGCLASS (VTOUIZS SPR:$a), GPR)>;
def : VFPPat<(i32 (fp_to_uint_sat SPR:$a, i32)),
(COPY_TO_REGCLASS (VTOUIZS SPR:$a), GPR)>;
-def : VFPNoNEONPat<(alignedstore32 (i32 (fp_to_uint (f32 SPR:$a))),
+def : VFPNoNEONPat<(alignedstore32 (i32 (any_fp_to_uint (f32 SPR:$a))),
addrmode5:$ptr),
(VSTRS (VTOUIZS SPR:$a), addrmode5:$ptr)>;
def : VFPPat<(alignedstore32 (i32 (fp_to_uint_sat (f32 SPR:$a), i32)),
@@ -1776,7 +1777,7 @@ def VTOUIZH : AVConv1IsH_Encode<0b11101, 0b11, 0b1100, 0b1001,
let isUnpredicable = 1;
}
-def : VFPNoNEONPat<(i32 (fp_to_uint (f16 HPR:$a))),
+def : VFPNoNEONPat<(i32 (any_fp_to_uint (f16 HPR:$a))),
(COPY_TO_REGCLASS (VTOUIZH (f16 HPR:$a)), GPR)>;
def : VFPPat<(i32 (fp_to_uint_sat (f16 HPR:$a), i32)),
(COPY_TO_REGCLASS (VTOUIZH (f16 HPR:$a)), GPR)>;
@@ -2320,13 +2321,13 @@ def : Pat<(fadd_mlx HPR:$dstin, (fmul_su (f16 HPR:$a), HPR:$b)),
// Match @llvm.fma.* intrinsics
// (fma x, y, z) -> (vfms z, x, y)
-def : Pat<(f64 (fma DPR:$Dn, DPR:$Dm, DPR:$Ddin)),
+def : Pat<(f64 (any_fma DPR:$Dn, DPR:$Dm, DPR:$Ddin)),
(VFMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(f32 (fma SPR:$Sn, SPR:$Sm, SPR:$Sdin)),
+def : Pat<(f32 (any_fma SPR:$Sn, SPR:$Sm, SPR:$Sdin)),
(VFMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(f16 (fma HPR:$Sn, HPR:$Sm, (f16 HPR:$Sdin))),
+def : Pat<(f16 (any_fma HPR:$Sn, HPR:$Sm, (f16 HPR:$Sdin))),
(VFMAH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
@@ -2375,13 +2376,13 @@ def : Pat<(fsub_mlx HPR:$dstin, (fmul_su (f16 HPR:$a), HPR:$b)),
// Match @llvm.fma.* intrinsics
// (fma (fneg x), y, z) -> (vfms z, x, y)
-def : Pat<(f64 (fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin)),
+def : Pat<(f64 (any_fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin)),
(VFMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(f32 (fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin)),
+def : Pat<(f32 (any_fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin)),
(VFMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(f16 (fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (f16 HPR:$Sdin))),
+def : Pat<(f16 (any_fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (f16 HPR:$Sdin))),
(VFMSH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
@@ -2427,23 +2428,23 @@ def : Pat<(fsub_mlx (fneg (fmul_su SPR:$a, SPR:$b)), SPR:$dstin),
// Match @llvm.fma.* intrinsics
// (fneg (fma x, y, z)) -> (vfnma z, x, y)
-def : Pat<(fneg (fma (f64 DPR:$Dn), (f64 DPR:$Dm), (f64 DPR:$Ddin))),
+def : Pat<(fneg (any_fma (f64 DPR:$Dn), (f64 DPR:$Dm), (f64 DPR:$Ddin))),
(VFNMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(fneg (fma (f32 SPR:$Sn), (f32 SPR:$Sm), (f32 SPR:$Sdin))),
+def : Pat<(fneg (any_fma (f32 SPR:$Sn), (f32 SPR:$Sm), (f32 SPR:$Sdin))),
(VFNMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(fneg (fma (f16 HPR:$Sn), (f16 HPR:$Sm), (f16 (f16 HPR:$Sdin)))),
+def : Pat<(fneg (any_fma (f16 HPR:$Sn), (f16 HPR:$Sm), (f16 (f16 HPR:$Sdin)))),
(VFNMAH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
// (fma (fneg x), y, (fneg z)) -> (vfnma z, x, y)
-def : Pat<(f64 (fma (fneg DPR:$Dn), DPR:$Dm, (fneg DPR:$Ddin))),
+def : Pat<(f64 (any_fma (fneg DPR:$Dn), DPR:$Dm, (fneg DPR:$Ddin))),
(VFNMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(f32 (fma (fneg SPR:$Sn), SPR:$Sm, (fneg SPR:$Sdin))),
+def : Pat<(f32 (any_fma (fneg SPR:$Sn), SPR:$Sm, (fneg SPR:$Sdin))),
(VFNMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(f16 (fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (fneg (f16 HPR:$Sdin)))),
+def : Pat<(f16 (any_fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (fneg (f16 HPR:$Sdin)))),
(VFNMAH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
@@ -2488,23 +2489,23 @@ def : Pat<(fsub_mlx (fmul_su SPR:$a, SPR:$b), SPR:$dstin),
// Match @llvm.fma.* intrinsics
// (fma x, y, (fneg z)) -> (vfnms z, x, y))
-def : Pat<(f64 (fma DPR:$Dn, DPR:$Dm, (fneg DPR:$Ddin))),
+def : Pat<(f64 (any_fma DPR:$Dn, DPR:$Dm, (fneg DPR:$Ddin))),
(VFNMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(f32 (fma SPR:$Sn, SPR:$Sm, (fneg SPR:$Sdin))),
+def : Pat<(f32 (any_fma SPR:$Sn, SPR:$Sm, (fneg SPR:$Sdin))),
(VFNMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(f16 (fma (f16 HPR:$Sn), (f16 HPR:$Sm), (fneg (f16 HPR:$Sdin)))),
+def : Pat<(f16 (any_fma (f16 HPR:$Sn), (f16 HPR:$Sm), (fneg (f16 HPR:$Sdin)))),
(VFNMSH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
// (fneg (fma (fneg x), y, z)) -> (vfnms z, x, y)
-def : Pat<(fneg (f64 (fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin))),
+def : Pat<(fneg (f64 (any_fma (fneg DPR:$Dn), DPR:$Dm, DPR:$Ddin))),
(VFNMSD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
Requires<[HasVFP4,HasDPVFP]>;
-def : Pat<(fneg (f32 (fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin))),
+def : Pat<(fneg (f32 (any_fma (fneg SPR:$Sn), SPR:$Sm, SPR:$Sdin))),
(VFNMSS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
Requires<[HasVFP4]>;
-def : Pat<(fneg (f16 (fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (f16 HPR:$Sdin)))),
+def : Pat<(fneg (f16 (any_fma (fneg (f16 HPR:$Sn)), (f16 HPR:$Sm), (f16 HPR:$Sdin)))),
(VFNMSH (f16 HPR:$Sdin), (f16 HPR:$Sn), (f16 HPR:$Sm))>,
Requires<[HasFullFP16]>;
generated by cgit v1.1 at 2025-10-30 05:17:00 +0000