aboutsummaryrefslogtreecommitdiff
path: root/llvm/lib/Target/AArch64
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/lib/Target/AArch64')
-rw-r--r--llvm/lib/Target/AArch64/AArch64ExpandPseudoInsts.cpp66
-rw-r--r--llvm/lib/Target/AArch64/AArch64ISelLowering.cpp18
-rw-r--r--llvm/lib/Target/AArch64/AArch64ISelLowering.h5
-rw-r--r--llvm/lib/Target/AArch64/AArch64InstrInfo.td27
-rw-r--r--llvm/lib/Target/AArch64/AArch64MIPeepholeOpt.cpp26
-rw-r--r--llvm/lib/Target/AArch64/AArch64Subtarget.cpp2
-rw-r--r--llvm/lib/Target/AArch64/AArch64Subtarget.h3
-rw-r--r--llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp8
8 files changed, 89 insertions, 66 deletions
diff --git a/llvm/lib/Target/AArch64/AArch64ExpandPseudoInsts.cpp b/llvm/lib/Target/AArch64/AArch64ExpandPseudoInsts.cpp
index 12fc976..201bfe0 100644
--- a/llvm/lib/Target/AArch64/AArch64ExpandPseudoInsts.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ExpandPseudoInsts.cpp
@@ -1205,32 +1205,36 @@ bool AArch64ExpandPseudo::expandMI(MachineBasicBlock &MBB,
Register DstReg = MI.getOperand(0).getReg();
if (DstReg == MI.getOperand(3).getReg()) {
// Expand to BIT
- BuildMI(MBB, MBBI, MI.getDebugLoc(),
- TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BITv8i8
- : AArch64::BITv16i8))
- .add(MI.getOperand(0))
- .add(MI.getOperand(3))
- .add(MI.getOperand(2))
- .add(MI.getOperand(1));
+ auto I = BuildMI(MBB, MBBI, MI.getDebugLoc(),
+ TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BITv8i8
+ : AArch64::BITv16i8))
+ .add(MI.getOperand(0))
+ .add(MI.getOperand(3))
+ .add(MI.getOperand(2))
+ .add(MI.getOperand(1));
+ transferImpOps(MI, I, I);
} else if (DstReg == MI.getOperand(2).getReg()) {
// Expand to BIF
- BuildMI(MBB, MBBI, MI.getDebugLoc(),
- TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BIFv8i8
- : AArch64::BIFv16i8))
- .add(MI.getOperand(0))
- .add(MI.getOperand(2))
- .add(MI.getOperand(3))
- .add(MI.getOperand(1));
+ auto I = BuildMI(MBB, MBBI, MI.getDebugLoc(),
+ TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BIFv8i8
+ : AArch64::BIFv16i8))
+ .add(MI.getOperand(0))
+ .add(MI.getOperand(2))
+ .add(MI.getOperand(3))
+ .add(MI.getOperand(1));
+ transferImpOps(MI, I, I);
} else {
// Expand to BSL, use additional move if required
if (DstReg == MI.getOperand(1).getReg()) {
- BuildMI(MBB, MBBI, MI.getDebugLoc(),
- TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BSLv8i8
- : AArch64::BSLv16i8))
- .add(MI.getOperand(0))
- .add(MI.getOperand(1))
- .add(MI.getOperand(2))
- .add(MI.getOperand(3));
+ auto I =
+ BuildMI(MBB, MBBI, MI.getDebugLoc(),
+ TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BSLv8i8
+ : AArch64::BSLv16i8))
+ .add(MI.getOperand(0))
+ .add(MI.getOperand(1))
+ .add(MI.getOperand(2))
+ .add(MI.getOperand(3));
+ transferImpOps(MI, I, I);
} else {
BuildMI(MBB, MBBI, MI.getDebugLoc(),
TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::ORRv8i8
@@ -1240,15 +1244,17 @@ bool AArch64ExpandPseudo::expandMI(MachineBasicBlock &MBB,
getRenamableRegState(MI.getOperand(0).isRenamable()))
.add(MI.getOperand(1))
.add(MI.getOperand(1));
- BuildMI(MBB, MBBI, MI.getDebugLoc(),
- TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BSLv8i8
- : AArch64::BSLv16i8))
- .add(MI.getOperand(0))
- .addReg(DstReg,
- RegState::Kill |
- getRenamableRegState(MI.getOperand(0).isRenamable()))
- .add(MI.getOperand(2))
- .add(MI.getOperand(3));
+ auto I2 =
+ BuildMI(MBB, MBBI, MI.getDebugLoc(),
+ TII->get(Opcode == AArch64::BSPv8i8 ? AArch64::BSLv8i8
+ : AArch64::BSLv16i8))
+ .add(MI.getOperand(0))
+ .addReg(DstReg,
+ RegState::Kill | getRenamableRegState(
+ MI.getOperand(0).isRenamable()))
+ .add(MI.getOperand(2))
+ .add(MI.getOperand(3));
+ transferImpOps(MI, I2, I2);
}
}
MI.eraseFromParent();
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index ff23f76..02ee517 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -6439,7 +6439,9 @@ bool AArch64TargetLowering::isVectorLoadExtDesirable(SDValue ExtVal) const {
}
}
- return true;
+ EVT PreExtScalarVT = ExtVal->getOperand(0).getValueType().getScalarType();
+ return PreExtScalarVT == MVT::i8 || PreExtScalarVT == MVT::i16 ||
+ PreExtScalarVT == MVT::i32 || PreExtScalarVT == MVT::i64;
}
unsigned getGatherVecOpcode(bool IsScaled, bool IsSigned, bool NeedsExtend) {
@@ -17155,7 +17157,7 @@ static Function *getStructuredStoreFunction(Module *M, unsigned Factor,
/// %vec0 = extractelement { <4 x i32>, <4 x i32> } %ld2, i32 0
/// %vec1 = extractelement { <4 x i32>, <4 x i32> } %ld2, i32 1
bool AArch64TargetLowering::lowerInterleavedLoad(
- LoadInst *LI, ArrayRef<ShuffleVectorInst *> Shuffles,
+ Instruction *Load, Value *Mask, ArrayRef<ShuffleVectorInst *> Shuffles,
ArrayRef<unsigned> Indices, unsigned Factor) const {
assert(Factor >= 2 && Factor <= getMaxSupportedInterleaveFactor() &&
"Invalid interleave factor");
@@ -17163,6 +17165,11 @@ bool AArch64TargetLowering::lowerInterleavedLoad(
assert(Shuffles.size() == Indices.size() &&
"Unmatched number of shufflevectors and indices");
+ auto *LI = dyn_cast<LoadInst>(Load);
+ if (!LI)
+ return false;
+ assert(!Mask && "Unexpected mask on a load");
+
const DataLayout &DL = LI->getDataLayout();
VectorType *VTy = Shuffles[0]->getType();
@@ -17336,12 +17343,17 @@ bool hasNearbyPairedStore(Iter It, Iter End, Value *Ptr, const DataLayout &DL) {
/// %sub.v1 = shuffle <32 x i32> %v0, <32 x i32> v1, <32, 33, 34, 35>
/// %sub.v2 = shuffle <32 x i32> %v0, <32 x i32> v1, <16, 17, 18, 19>
/// call void llvm.aarch64.neon.st3(%sub.v0, %sub.v1, %sub.v2, %ptr)
-bool AArch64TargetLowering::lowerInterleavedStore(StoreInst *SI,
+bool AArch64TargetLowering::lowerInterleavedStore(Instruction *Store,
+ Value *LaneMask,
ShuffleVectorInst *SVI,
unsigned Factor) const {
assert(Factor >= 2 && Factor <= getMaxSupportedInterleaveFactor() &&
"Invalid interleave factor");
+ auto *SI = dyn_cast<StoreInst>(Store);
+ if (!SI)
+ return false;
+ assert(!LaneMask && "Unexpected mask on store");
auto *VecTy = cast<FixedVectorType>(SVI->getType());
assert(VecTy->getNumElements() % Factor == 0 && "Invalid interleaved store");
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.h b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
index 7b1de3d..d8403c2 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -211,11 +211,12 @@ public:
unsigned getMaxSupportedInterleaveFactor() const override { return 4; }
- bool lowerInterleavedLoad(LoadInst *LI,
+ bool lowerInterleavedLoad(Instruction *Load, Value *Mask,
ArrayRef<ShuffleVectorInst *> Shuffles,
ArrayRef<unsigned> Indices,
unsigned Factor) const override;
- bool lowerInterleavedStore(StoreInst *SI, ShuffleVectorInst *SVI,
+ bool lowerInterleavedStore(Instruction *Store, Value *Mask,
+ ShuffleVectorInst *SVI,
unsigned Factor) const override;
bool lowerDeinterleaveIntrinsicToLoad(Instruction *Load, Value *Mask,
diff --git a/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index 6c46b18..9f8a257 100644
--- a/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -1053,13 +1053,6 @@ def AArch64umaxv : SDNode<"AArch64ISD::UMAXV", SDT_AArch64UnaryVec>;
def AArch64uaddlv : SDNode<"AArch64ISD::UADDLV", SDT_AArch64uaddlp>;
def AArch64saddlv : SDNode<"AArch64ISD::SADDLV", SDT_AArch64uaddlp>;
-def AArch64uabd : PatFrags<(ops node:$lhs, node:$rhs),
- [(abdu node:$lhs, node:$rhs),
- (int_aarch64_neon_uabd node:$lhs, node:$rhs)]>;
-def AArch64sabd : PatFrags<(ops node:$lhs, node:$rhs),
- [(abds node:$lhs, node:$rhs),
- (int_aarch64_neon_sabd node:$lhs, node:$rhs)]>;
-
// Add Pairwise of two vectors
def AArch64addp_n : SDNode<"AArch64ISD::ADDP", SDT_AArch64Zip>;
// Add Long Pairwise
@@ -5667,8 +5660,7 @@ let Predicates = [HasFullFP16] in {
// Advanced SIMD two vector instructions.
//===----------------------------------------------------------------------===//
-defm UABDL : SIMDLongThreeVectorBHSabdl<1, 0b0111, "uabdl",
- AArch64uabd>;
+defm UABDL : SIMDLongThreeVectorBHSabdl<1, 0b0111, "uabdl", abdu>;
// Match UABDL in log2-shuffle patterns.
def : Pat<(abs (v8i16 (sub (zext (v8i8 V64:$opA)),
(zext (v8i8 V64:$opB))))),
@@ -6018,8 +6010,8 @@ defm MLS : SIMDThreeSameVectorBHSTied<1, 0b10010, "mls", null_frag>;
defm MUL : SIMDThreeSameVectorBHS<0, 0b10011, "mul", mul>;
defm PMUL : SIMDThreeSameVectorB<1, 0b10011, "pmul", int_aarch64_neon_pmul>;
defm SABA : SIMDThreeSameVectorBHSTied<0, 0b01111, "saba",
- TriOpFrag<(add node:$LHS, (AArch64sabd node:$MHS, node:$RHS))> >;
-defm SABD : SIMDThreeSameVectorBHS<0,0b01110,"sabd", AArch64sabd>;
+ TriOpFrag<(add node:$LHS, (abds node:$MHS, node:$RHS))> >;
+defm SABD : SIMDThreeSameVectorBHS<0,0b01110,"sabd", abds>;
defm SHADD : SIMDThreeSameVectorBHS<0,0b00000,"shadd", avgfloors>;
defm SHSUB : SIMDThreeSameVectorBHS<0,0b00100,"shsub", int_aarch64_neon_shsub>;
defm SMAXP : SIMDThreeSameVectorBHS<0,0b10100,"smaxp", int_aarch64_neon_smaxp>;
@@ -6037,8 +6029,8 @@ defm SRSHL : SIMDThreeSameVector<0,0b01010,"srshl", int_aarch64_neon_srshl>;
defm SSHL : SIMDThreeSameVector<0,0b01000,"sshl", int_aarch64_neon_sshl>;
defm SUB : SIMDThreeSameVector<1,0b10000,"sub", sub>;
defm UABA : SIMDThreeSameVectorBHSTied<1, 0b01111, "uaba",
- TriOpFrag<(add node:$LHS, (AArch64uabd node:$MHS, node:$RHS))> >;
-defm UABD : SIMDThreeSameVectorBHS<1,0b01110,"uabd", AArch64uabd>;
+ TriOpFrag<(add node:$LHS, (abdu node:$MHS, node:$RHS))> >;
+defm UABD : SIMDThreeSameVectorBHS<1,0b01110,"uabd", abdu>;
defm UHADD : SIMDThreeSameVectorBHS<1,0b00000,"uhadd", avgflooru>;
defm UHSUB : SIMDThreeSameVectorBHS<1,0b00100,"uhsub", int_aarch64_neon_uhsub>;
defm UMAXP : SIMDThreeSameVectorBHS<1,0b10100,"umaxp", int_aarch64_neon_umaxp>;
@@ -6759,10 +6751,8 @@ defm SUBHN : SIMDNarrowThreeVectorBHS<0,0b0110,"subhn", int_aarch64_neon_subhn>
defm RADDHN : SIMDNarrowThreeVectorBHS<1,0b0100,"raddhn",int_aarch64_neon_raddhn>;
defm RSUBHN : SIMDNarrowThreeVectorBHS<1,0b0110,"rsubhn",int_aarch64_neon_rsubhn>;
defm PMULL : SIMDDifferentThreeVectorBD<0,0b1110,"pmull", AArch64pmull>;
-defm SABAL : SIMDLongThreeVectorTiedBHSabal<0,0b0101,"sabal",
- AArch64sabd>;
-defm SABDL : SIMDLongThreeVectorBHSabdl<0, 0b0111, "sabdl",
- AArch64sabd>;
+defm SABAL : SIMDLongThreeVectorTiedBHSabal<0,0b0101,"sabal", abds>;
+defm SABDL : SIMDLongThreeVectorBHSabdl<0, 0b0111, "sabdl", abds>;
defm SADDL : SIMDLongThreeVectorBHS< 0, 0b0000, "saddl",
BinOpFrag<(add (sext node:$LHS), (sext node:$RHS))>>;
defm SADDW : SIMDWideThreeVectorBHS< 0, 0b0001, "saddw",
@@ -6780,8 +6770,7 @@ defm SSUBL : SIMDLongThreeVectorBHS<0, 0b0010, "ssubl",
BinOpFrag<(sub (sext node:$LHS), (sext node:$RHS))>>;
defm SSUBW : SIMDWideThreeVectorBHS<0, 0b0011, "ssubw",
BinOpFrag<(sub node:$LHS, (sext node:$RHS))>>;
-defm UABAL : SIMDLongThreeVectorTiedBHSabal<1, 0b0101, "uabal",
- AArch64uabd>;
+defm UABAL : SIMDLongThreeVectorTiedBHSabal<1, 0b0101, "uabal", abdu>;
defm UADDL : SIMDLongThreeVectorBHS<1, 0b0000, "uaddl",
BinOpFrag<(add (zanyext node:$LHS), (zanyext node:$RHS))>>;
defm UADDW : SIMDWideThreeVectorBHS<1, 0b0001, "uaddw",
diff --git a/llvm/lib/Target/AArch64/AArch64MIPeepholeOpt.cpp b/llvm/lib/Target/AArch64/AArch64MIPeepholeOpt.cpp
index 0ddd17c..abcd550 100644
--- a/llvm/lib/Target/AArch64/AArch64MIPeepholeOpt.cpp
+++ b/llvm/lib/Target/AArch64/AArch64MIPeepholeOpt.cpp
@@ -8,8 +8,8 @@
//
// This pass performs below peephole optimizations on MIR level.
//
-// 1. MOVi32imm + ANDWrr ==> ANDWri + ANDWri
-// MOVi64imm + ANDXrr ==> ANDXri + ANDXri
+// 1. MOVi32imm + ANDS?Wrr ==> ANDWri + ANDS?Wri
+// MOVi64imm + ANDS?Xrr ==> ANDXri + ANDS?Xri
//
// 2. MOVi32imm + ADDWrr ==> ADDWRi + ADDWRi
// MOVi64imm + ADDXrr ==> ANDXri + ANDXri
@@ -126,7 +126,7 @@ struct AArch64MIPeepholeOpt : public MachineFunctionPass {
bool visitADDSSUBS(OpcodePair PosOpcs, OpcodePair NegOpcs, MachineInstr &MI);
template <typename T>
- bool visitAND(unsigned Opc, MachineInstr &MI);
+ bool visitAND(unsigned Opc, MachineInstr &MI, unsigned OtherOpc = 0);
bool visitORR(MachineInstr &MI);
bool visitCSEL(MachineInstr &MI);
bool visitINSERT(MachineInstr &MI);
@@ -194,12 +194,12 @@ static bool splitBitmaskImm(T Imm, unsigned RegSize, T &Imm1Enc, T &Imm2Enc) {
}
template <typename T>
-bool AArch64MIPeepholeOpt::visitAND(
- unsigned Opc, MachineInstr &MI) {
+bool AArch64MIPeepholeOpt::visitAND(unsigned Opc, MachineInstr &MI,
+ unsigned OtherOpc) {
// Try below transformation.
//
- // MOVi32imm + ANDWrr ==> ANDWri + ANDWri
- // MOVi64imm + ANDXrr ==> ANDXri + ANDXri
+ // MOVi32imm + ANDS?Wrr ==> ANDWri + ANDS?Wri
+ // MOVi64imm + ANDS?Xrr ==> ANDXri + ANDS?Xri
//
// The mov pseudo instruction could be expanded to multiple mov instructions
// later. Let's try to split the constant operand of mov instruction into two
@@ -208,10 +208,10 @@ bool AArch64MIPeepholeOpt::visitAND(
return splitTwoPartImm<T>(
MI,
- [Opc](T Imm, unsigned RegSize, T &Imm0,
- T &Imm1) -> std::optional<OpcodePair> {
+ [Opc, OtherOpc](T Imm, unsigned RegSize, T &Imm0,
+ T &Imm1) -> std::optional<OpcodePair> {
if (splitBitmaskImm(Imm, RegSize, Imm0, Imm1))
- return std::make_pair(Opc, Opc);
+ return std::make_pair(Opc, !OtherOpc ? Opc : OtherOpc);
return std::nullopt;
},
[&TII = TII](MachineInstr &MI, OpcodePair Opcode, unsigned Imm0,
@@ -864,6 +864,12 @@ bool AArch64MIPeepholeOpt::runOnMachineFunction(MachineFunction &MF) {
case AArch64::ANDXrr:
Changed |= visitAND<uint64_t>(AArch64::ANDXri, MI);
break;
+ case AArch64::ANDSWrr:
+ Changed |= visitAND<uint32_t>(AArch64::ANDWri, MI, AArch64::ANDSWri);
+ break;
+ case AArch64::ANDSXrr:
+ Changed |= visitAND<uint64_t>(AArch64::ANDXri, MI, AArch64::ANDSXri);
+ break;
case AArch64::ORRWrs:
Changed |= visitORR(MI);
break;
diff --git a/llvm/lib/Target/AArch64/AArch64Subtarget.cpp b/llvm/lib/Target/AArch64/AArch64Subtarget.cpp
index 2409cc8..0f4f012 100644
--- a/llvm/lib/Target/AArch64/AArch64Subtarget.cpp
+++ b/llvm/lib/Target/AArch64/AArch64Subtarget.cpp
@@ -534,7 +534,7 @@ unsigned AArch64Subtarget::classifyGlobalFunctionReference(
}
void AArch64Subtarget::overrideSchedPolicy(MachineSchedPolicy &Policy,
- unsigned NumRegionInstrs) const {
+ const SchedRegion &Region) const {
// LNT run (at least on Cyclone) showed reasonably significant gains for
// bi-directional scheduling. 253.perlbmk.
Policy.OnlyTopDown = false;
diff --git a/llvm/lib/Target/AArch64/AArch64Subtarget.h b/llvm/lib/Target/AArch64/AArch64Subtarget.h
index 154db3c..061ed61 100644
--- a/llvm/lib/Target/AArch64/AArch64Subtarget.h
+++ b/llvm/lib/Target/AArch64/AArch64Subtarget.h
@@ -343,7 +343,8 @@ public:
}
void overrideSchedPolicy(MachineSchedPolicy &Policy,
- unsigned NumRegionInstrs) const override;
+ const SchedRegion &Region) const override;
+
void adjustSchedDependency(SUnit *Def, int DefOpIdx, SUnit *Use, int UseOpIdx,
SDep &Dep,
const TargetSchedModel *SchedModel) const override;
diff --git a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp
index 473ba5e..bb0f667b 100644
--- a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp
+++ b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp
@@ -287,6 +287,10 @@ AArch64LegalizerInfo::AArch64LegalizerInfo(const AArch64Subtarget &ST)
.moreElementsToNextPow2(0)
.lower();
+ getActionDefinitionsBuilder({G_ABDS, G_ABDU})
+ .legalFor({v8s8, v16s8, v4s16, v8s16, v2s32, v4s32})
+ .lower();
+
getActionDefinitionsBuilder(
{G_SADDE, G_SSUBE, G_UADDE, G_USUBE, G_SADDO, G_SSUBO, G_UADDO, G_USUBO})
.legalFor({{s32, s32}, {s64, s32}})
@@ -1794,6 +1798,10 @@ bool AArch64LegalizerInfo::legalizeIntrinsic(LegalizerHelper &Helper,
return LowerBinOp(AArch64::G_SMULL);
case Intrinsic::aarch64_neon_umull:
return LowerBinOp(AArch64::G_UMULL);
+ case Intrinsic::aarch64_neon_sabd:
+ return LowerBinOp(TargetOpcode::G_ABDS);
+ case Intrinsic::aarch64_neon_uabd:
+ return LowerBinOp(TargetOpcode::G_ABDU);
case Intrinsic::aarch64_neon_abs: {
// Lower the intrinsic to G_ABS.
MIB.buildInstr(TargetOpcode::G_ABS, {MI.getOperand(0)}, {MI.getOperand(2)});
generated by cgit v1.1 at 2025-07-23 09:17:16 +0000