X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FAArch64%2FAArch64InstrInfo.td;h=2ee1299b6fd97a0d6b975c620292f22bcbe0d4fd;hb=0e0cc9459a89ac4f5e4cb5003d223a974c2d4d94;hp=bdef183d1a62b4abe218ee8363b7bf1c49d2c2c4;hpb=ad1b9dcb2c9f1a1a031ff5d026bd988999a54c4b;p=oota-llvm.git diff --git a/lib/Target/AArch64/AArch64InstrInfo.td b/lib/Target/AArch64/AArch64InstrInfo.td index bdef183d1a6..2ee1299b6fd 100644 --- a/lib/Target/AArch64/AArch64InstrInfo.td +++ b/lib/Target/AArch64/AArch64InstrInfo.td @@ -1,4 +1,4 @@ -//===----- AArch64InstrInfo.td - AArch64 Instruction Info ----*- tablegen -*-=// +//=- AArch64InstrInfo.td - Describe the AArch64 Instructions -*- tablegen -*-=// // // The LLVM Compiler Infrastructure // @@ -7,5121 +7,5991 @@ // //===----------------------------------------------------------------------===// // -// This file describes the AArch64 scalar instructions in TableGen format. +// AArch64 Instruction definitions. // //===----------------------------------------------------------------------===// -include "AArch64InstrFormats.td" - //===----------------------------------------------------------------------===// -// Target-specific ISD nodes and profiles +// ARM Instruction Predicate Definitions. +// +def HasV8_1a : Predicate<"Subtarget->hasV8_1aOps()">, + AssemblerPredicate<"HasV8_1aOps", "armv8.1a">; +def HasV8_2a : Predicate<"Subtarget->hasV8_2aOps()">, + AssemblerPredicate<"HasV8_2aOps", "armv8.2a">; +def HasFPARMv8 : Predicate<"Subtarget->hasFPARMv8()">, + AssemblerPredicate<"FeatureFPARMv8", "fp-armv8">; +def HasNEON : Predicate<"Subtarget->hasNEON()">, + AssemblerPredicate<"FeatureNEON", "neon">; +def HasCrypto : Predicate<"Subtarget->hasCrypto()">, + AssemblerPredicate<"FeatureCrypto", "crypto">; +def HasCRC : Predicate<"Subtarget->hasCRC()">, + AssemblerPredicate<"FeatureCRC", "crc">; +def HasPerfMon : Predicate<"Subtarget->hasPerfMon()">; +def HasFullFP16 : Predicate<"Subtarget->hasFullFP16()">, + AssemblerPredicate<"FeatureFullFP16", "fullfp16">; +def HasSPE : Predicate<"Subtarget->hasSPE()">, + AssemblerPredicate<"FeatureSPE", "spe">; + +def IsLE : Predicate<"Subtarget->isLittleEndian()">; +def IsBE : Predicate<"!Subtarget->isLittleEndian()">; +def IsCyclone : Predicate<"Subtarget->isCyclone()">; + //===----------------------------------------------------------------------===// +// AArch64-specific DAG Nodes. +// -def SDT_A64ret : SDTypeProfile<0, 0, []>; -def A64ret : SDNode<"AArch64ISD::Ret", SDT_A64ret, [SDNPHasChain, - SDNPOptInGlue, - SDNPVariadic]>; +// SDTBinaryArithWithFlagsOut - RES1, FLAGS = op LHS, RHS +def SDTBinaryArithWithFlagsOut : SDTypeProfile<2, 2, + [SDTCisSameAs<0, 2>, + SDTCisSameAs<0, 3>, + SDTCisInt<0>, SDTCisVT<1, i32>]>; + +// SDTBinaryArithWithFlagsIn - RES1, FLAGS = op LHS, RHS, FLAGS +def SDTBinaryArithWithFlagsIn : SDTypeProfile<1, 3, + [SDTCisSameAs<0, 1>, + SDTCisSameAs<0, 2>, + SDTCisInt<0>, + SDTCisVT<3, i32>]>; + +// SDTBinaryArithWithFlagsInOut - RES1, FLAGS = op LHS, RHS, FLAGS +def SDTBinaryArithWithFlagsInOut : SDTypeProfile<2, 3, + [SDTCisSameAs<0, 2>, + SDTCisSameAs<0, 3>, + SDTCisInt<0>, + SDTCisVT<1, i32>, + SDTCisVT<4, i32>]>; + +def SDT_AArch64Brcond : SDTypeProfile<0, 3, + [SDTCisVT<0, OtherVT>, SDTCisVT<1, i32>, + SDTCisVT<2, i32>]>; +def SDT_AArch64cbz : SDTypeProfile<0, 2, [SDTCisInt<0>, SDTCisVT<1, OtherVT>]>; +def SDT_AArch64tbz : SDTypeProfile<0, 3, [SDTCisInt<0>, SDTCisInt<1>, + SDTCisVT<2, OtherVT>]>; + + +def SDT_AArch64CSel : SDTypeProfile<1, 4, + [SDTCisSameAs<0, 1>, + SDTCisSameAs<0, 2>, + SDTCisInt<3>, + SDTCisVT<4, i32>]>; +def SDT_AArch64CCMP : SDTypeProfile<1, 5, + [SDTCisVT<0, i32>, + SDTCisInt<1>, + SDTCisSameAs<1, 2>, + SDTCisInt<3>, + SDTCisInt<4>, + SDTCisVT<5, i32>]>; +def SDT_AArch64FCCMP : SDTypeProfile<1, 5, + [SDTCisVT<0, i32>, + SDTCisFP<1>, + SDTCisSameAs<1, 2>, + SDTCisInt<3>, + SDTCisInt<4>, + SDTCisVT<5, i32>]>; +def SDT_AArch64FCmp : SDTypeProfile<0, 2, + [SDTCisFP<0>, + SDTCisSameAs<0, 1>]>; +def SDT_AArch64Dup : SDTypeProfile<1, 1, [SDTCisVec<0>]>; +def SDT_AArch64DupLane : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisInt<2>]>; +def SDT_AArch64Zip : SDTypeProfile<1, 2, [SDTCisVec<0>, + SDTCisSameAs<0, 1>, + SDTCisSameAs<0, 2>]>; +def SDT_AArch64MOVIedit : SDTypeProfile<1, 1, [SDTCisInt<1>]>; +def SDT_AArch64MOVIshift : SDTypeProfile<1, 2, [SDTCisInt<1>, SDTCisInt<2>]>; +def SDT_AArch64vecimm : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisInt<2>, SDTCisInt<3>]>; +def SDT_AArch64UnaryVec: SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0,1>]>; +def SDT_AArch64ExtVec: SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisSameAs<0,2>, SDTCisInt<3>]>; +def SDT_AArch64vshift : SDTypeProfile<1, 2, [SDTCisSameAs<0,1>, SDTCisInt<2>]>; + +def SDT_AArch64unvec : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisSameAs<0,1>]>; +def SDT_AArch64fcmpz : SDTypeProfile<1, 1, []>; +def SDT_AArch64fcmp : SDTypeProfile<1, 2, [SDTCisSameAs<1,2>]>; +def SDT_AArch64binvec : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisSameAs<0,2>]>; +def SDT_AArch64trivec : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>, + SDTCisSameAs<0,2>, + SDTCisSameAs<0,3>]>; +def SDT_AArch64TCRET : SDTypeProfile<0, 2, [SDTCisPtrTy<0>]>; +def SDT_AArch64PREFETCH : SDTypeProfile<0, 2, [SDTCisVT<0, i32>, SDTCisPtrTy<1>]>; + +def SDT_AArch64ITOF : SDTypeProfile<1, 1, [SDTCisFP<0>, SDTCisSameAs<0,1>]>; + +def SDT_AArch64TLSDescCall : SDTypeProfile<0, -2, [SDTCisPtrTy<0>, + SDTCisPtrTy<1>]>; + +// Generates the general dynamic sequences, i.e. +// adrp x0, :tlsdesc:var +// ldr x1, [x0, #:tlsdesc_lo12:var] +// add x0, x0, #:tlsdesc_lo12:var +// .tlsdesccall var +// blr x1 + +// (the TPIDR_EL0 offset is put directly in X0, hence no "result" here) +// number of operands (the variable) +def SDT_AArch64TLSDescCallSeq : SDTypeProfile<0,1, + [SDTCisPtrTy<0>]>; + +def SDT_AArch64WrapperLarge : SDTypeProfile<1, 4, + [SDTCisVT<0, i64>, SDTCisVT<1, i32>, + SDTCisSameAs<1, 2>, SDTCisSameAs<1, 3>, + SDTCisSameAs<1, 4>]>; + + +// Node definitions. +def AArch64adrp : SDNode<"AArch64ISD::ADRP", SDTIntUnaryOp, []>; +def AArch64addlow : SDNode<"AArch64ISD::ADDlow", SDTIntBinOp, []>; +def AArch64LOADgot : SDNode<"AArch64ISD::LOADgot", SDTIntUnaryOp>; +def AArch64callseq_start : SDNode<"ISD::CALLSEQ_START", + SDCallSeqStart<[ SDTCisVT<0, i32> ]>, + [SDNPHasChain, SDNPOutGlue]>; +def AArch64callseq_end : SDNode<"ISD::CALLSEQ_END", + SDCallSeqEnd<[ SDTCisVT<0, i32>, + SDTCisVT<1, i32> ]>, + [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue]>; +def AArch64call : SDNode<"AArch64ISD::CALL", + SDTypeProfile<0, -1, [SDTCisPtrTy<0>]>, + [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, + SDNPVariadic]>; +def AArch64brcond : SDNode<"AArch64ISD::BRCOND", SDT_AArch64Brcond, + [SDNPHasChain]>; +def AArch64cbz : SDNode<"AArch64ISD::CBZ", SDT_AArch64cbz, + [SDNPHasChain]>; +def AArch64cbnz : SDNode<"AArch64ISD::CBNZ", SDT_AArch64cbz, + [SDNPHasChain]>; +def AArch64tbz : SDNode<"AArch64ISD::TBZ", SDT_AArch64tbz, + [SDNPHasChain]>; +def AArch64tbnz : SDNode<"AArch64ISD::TBNZ", SDT_AArch64tbz, + [SDNPHasChain]>; + + +def AArch64csel : SDNode<"AArch64ISD::CSEL", SDT_AArch64CSel>; +def AArch64csinv : SDNode<"AArch64ISD::CSINV", SDT_AArch64CSel>; +def AArch64csneg : SDNode<"AArch64ISD::CSNEG", SDT_AArch64CSel>; +def AArch64csinc : SDNode<"AArch64ISD::CSINC", SDT_AArch64CSel>; +def AArch64retflag : SDNode<"AArch64ISD::RET_FLAG", SDTNone, + [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>; +def AArch64adc : SDNode<"AArch64ISD::ADC", SDTBinaryArithWithFlagsIn >; +def AArch64sbc : SDNode<"AArch64ISD::SBC", SDTBinaryArithWithFlagsIn>; +def AArch64add_flag : SDNode<"AArch64ISD::ADDS", SDTBinaryArithWithFlagsOut, + [SDNPCommutative]>; +def AArch64sub_flag : SDNode<"AArch64ISD::SUBS", SDTBinaryArithWithFlagsOut>; +def AArch64and_flag : SDNode<"AArch64ISD::ANDS", SDTBinaryArithWithFlagsOut, + [SDNPCommutative]>; +def AArch64adc_flag : SDNode<"AArch64ISD::ADCS", SDTBinaryArithWithFlagsInOut>; +def AArch64sbc_flag : SDNode<"AArch64ISD::SBCS", SDTBinaryArithWithFlagsInOut>; + +def AArch64ccmp : SDNode<"AArch64ISD::CCMP", SDT_AArch64CCMP>; +def AArch64ccmn : SDNode<"AArch64ISD::CCMN", SDT_AArch64CCMP>; +def AArch64fccmp : SDNode<"AArch64ISD::FCCMP", SDT_AArch64FCCMP>; + +def AArch64threadpointer : SDNode<"AArch64ISD::THREAD_POINTER", SDTPtrLeaf>; + +def AArch64fcmp : SDNode<"AArch64ISD::FCMP", SDT_AArch64FCmp>; + +def AArch64dup : SDNode<"AArch64ISD::DUP", SDT_AArch64Dup>; +def AArch64duplane8 : SDNode<"AArch64ISD::DUPLANE8", SDT_AArch64DupLane>; +def AArch64duplane16 : SDNode<"AArch64ISD::DUPLANE16", SDT_AArch64DupLane>; +def AArch64duplane32 : SDNode<"AArch64ISD::DUPLANE32", SDT_AArch64DupLane>; +def AArch64duplane64 : SDNode<"AArch64ISD::DUPLANE64", SDT_AArch64DupLane>; + +def AArch64zip1 : SDNode<"AArch64ISD::ZIP1", SDT_AArch64Zip>; +def AArch64zip2 : SDNode<"AArch64ISD::ZIP2", SDT_AArch64Zip>; +def AArch64uzp1 : SDNode<"AArch64ISD::UZP1", SDT_AArch64Zip>; +def AArch64uzp2 : SDNode<"AArch64ISD::UZP2", SDT_AArch64Zip>; +def AArch64trn1 : SDNode<"AArch64ISD::TRN1", SDT_AArch64Zip>; +def AArch64trn2 : SDNode<"AArch64ISD::TRN2", SDT_AArch64Zip>; + +def AArch64movi_edit : SDNode<"AArch64ISD::MOVIedit", SDT_AArch64MOVIedit>; +def AArch64movi_shift : SDNode<"AArch64ISD::MOVIshift", SDT_AArch64MOVIshift>; +def AArch64movi_msl : SDNode<"AArch64ISD::MOVImsl", SDT_AArch64MOVIshift>; +def AArch64mvni_shift : SDNode<"AArch64ISD::MVNIshift", SDT_AArch64MOVIshift>; +def AArch64mvni_msl : SDNode<"AArch64ISD::MVNImsl", SDT_AArch64MOVIshift>; +def AArch64movi : SDNode<"AArch64ISD::MOVI", SDT_AArch64MOVIedit>; +def AArch64fmov : SDNode<"AArch64ISD::FMOV", SDT_AArch64MOVIedit>; + +def AArch64rev16 : SDNode<"AArch64ISD::REV16", SDT_AArch64UnaryVec>; +def AArch64rev32 : SDNode<"AArch64ISD::REV32", SDT_AArch64UnaryVec>; +def AArch64rev64 : SDNode<"AArch64ISD::REV64", SDT_AArch64UnaryVec>; +def AArch64ext : SDNode<"AArch64ISD::EXT", SDT_AArch64ExtVec>; + +def AArch64vashr : SDNode<"AArch64ISD::VASHR", SDT_AArch64vshift>; +def AArch64vlshr : SDNode<"AArch64ISD::VLSHR", SDT_AArch64vshift>; +def AArch64vshl : SDNode<"AArch64ISD::VSHL", SDT_AArch64vshift>; +def AArch64sqshli : SDNode<"AArch64ISD::SQSHL_I", SDT_AArch64vshift>; +def AArch64uqshli : SDNode<"AArch64ISD::UQSHL_I", SDT_AArch64vshift>; +def AArch64sqshlui : SDNode<"AArch64ISD::SQSHLU_I", SDT_AArch64vshift>; +def AArch64srshri : SDNode<"AArch64ISD::SRSHR_I", SDT_AArch64vshift>; +def AArch64urshri : SDNode<"AArch64ISD::URSHR_I", SDT_AArch64vshift>; + +def AArch64not: SDNode<"AArch64ISD::NOT", SDT_AArch64unvec>; +def AArch64bit: SDNode<"AArch64ISD::BIT", SDT_AArch64trivec>; +def AArch64bsl: SDNode<"AArch64ISD::BSL", SDT_AArch64trivec>; + +def AArch64cmeq: SDNode<"AArch64ISD::CMEQ", SDT_AArch64binvec>; +def AArch64cmge: SDNode<"AArch64ISD::CMGE", SDT_AArch64binvec>; +def AArch64cmgt: SDNode<"AArch64ISD::CMGT", SDT_AArch64binvec>; +def AArch64cmhi: SDNode<"AArch64ISD::CMHI", SDT_AArch64binvec>; +def AArch64cmhs: SDNode<"AArch64ISD::CMHS", SDT_AArch64binvec>; + +def AArch64fcmeq: SDNode<"AArch64ISD::FCMEQ", SDT_AArch64fcmp>; +def AArch64fcmge: SDNode<"AArch64ISD::FCMGE", SDT_AArch64fcmp>; +def AArch64fcmgt: SDNode<"AArch64ISD::FCMGT", SDT_AArch64fcmp>; + +def AArch64cmeqz: SDNode<"AArch64ISD::CMEQz", SDT_AArch64unvec>; +def AArch64cmgez: SDNode<"AArch64ISD::CMGEz", SDT_AArch64unvec>; +def AArch64cmgtz: SDNode<"AArch64ISD::CMGTz", SDT_AArch64unvec>; +def AArch64cmlez: SDNode<"AArch64ISD::CMLEz", SDT_AArch64unvec>; +def AArch64cmltz: SDNode<"AArch64ISD::CMLTz", SDT_AArch64unvec>; +def AArch64cmtst : PatFrag<(ops node:$LHS, node:$RHS), + (AArch64not (AArch64cmeqz (and node:$LHS, node:$RHS)))>; + +def AArch64fcmeqz: SDNode<"AArch64ISD::FCMEQz", SDT_AArch64fcmpz>; +def AArch64fcmgez: SDNode<"AArch64ISD::FCMGEz", SDT_AArch64fcmpz>; +def AArch64fcmgtz: SDNode<"AArch64ISD::FCMGTz", SDT_AArch64fcmpz>; +def AArch64fcmlez: SDNode<"AArch64ISD::FCMLEz", SDT_AArch64fcmpz>; +def AArch64fcmltz: SDNode<"AArch64ISD::FCMLTz", SDT_AArch64fcmpz>; + +def AArch64bici: SDNode<"AArch64ISD::BICi", SDT_AArch64vecimm>; +def AArch64orri: SDNode<"AArch64ISD::ORRi", SDT_AArch64vecimm>; + +def AArch64neg : SDNode<"AArch64ISD::NEG", SDT_AArch64unvec>; + +def AArch64tcret: SDNode<"AArch64ISD::TC_RETURN", SDT_AArch64TCRET, + [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>; + +def AArch64Prefetch : SDNode<"AArch64ISD::PREFETCH", SDT_AArch64PREFETCH, + [SDNPHasChain, SDNPSideEffect]>; + +def AArch64sitof: SDNode<"AArch64ISD::SITOF", SDT_AArch64ITOF>; +def AArch64uitof: SDNode<"AArch64ISD::UITOF", SDT_AArch64ITOF>; + +def AArch64tlsdesc_callseq : SDNode<"AArch64ISD::TLSDESC_CALLSEQ", + SDT_AArch64TLSDescCallSeq, + [SDNPInGlue, SDNPOutGlue, SDNPHasChain, + SDNPVariadic]>; + + +def AArch64WrapperLarge : SDNode<"AArch64ISD::WrapperLarge", + SDT_AArch64WrapperLarge>; + +def AArch64NvCast : SDNode<"AArch64ISD::NVCAST", SDTUnaryOp>; + +def SDT_AArch64mull : SDTypeProfile<1, 2, [SDTCisInt<0>, SDTCisInt<1>, + SDTCisSameAs<1, 2>]>; +def AArch64smull : SDNode<"AArch64ISD::SMULL", SDT_AArch64mull>; +def AArch64umull : SDNode<"AArch64ISD::UMULL", SDT_AArch64mull>; + +def AArch64saddv : SDNode<"AArch64ISD::SADDV", SDT_AArch64UnaryVec>; +def AArch64uaddv : SDNode<"AArch64ISD::UADDV", SDT_AArch64UnaryVec>; +def AArch64sminv : SDNode<"AArch64ISD::SMINV", SDT_AArch64UnaryVec>; +def AArch64uminv : SDNode<"AArch64ISD::UMINV", SDT_AArch64UnaryVec>; +def AArch64smaxv : SDNode<"AArch64ISD::SMAXV", SDT_AArch64UnaryVec>; +def AArch64umaxv : SDNode<"AArch64ISD::UMAXV", SDT_AArch64UnaryVec>; -// (ins NZCV, Condition, Dest) -def SDT_A64br_cc : SDTypeProfile<0, 3, [SDTCisVT<0, i32>]>; -def A64br_cc : SDNode<"AArch64ISD::BR_CC", SDT_A64br_cc, [SDNPHasChain]>; +//===----------------------------------------------------------------------===// -// (outs Result), (ins NZCV, IfTrue, IfFalse, Condition) -def SDT_A64select_cc : SDTypeProfile<1, 4, [SDTCisVT<1, i32>, - SDTCisSameAs<0, 2>, - SDTCisSameAs<2, 3>]>; -def A64select_cc : SDNode<"AArch64ISD::SELECT_CC", SDT_A64select_cc>; +//===----------------------------------------------------------------------===// -// (outs NZCV), (ins LHS, RHS, Condition) -def SDT_A64setcc : SDTypeProfile<1, 3, [SDTCisVT<0, i32>, - SDTCisSameAs<1, 2>]>; -def A64setcc : SDNode<"AArch64ISD::SETCC", SDT_A64setcc>; +// AArch64 Instruction Predicate Definitions. +// +def HasZCZ : Predicate<"Subtarget->hasZeroCycleZeroing()">; +def NoZCZ : Predicate<"!Subtarget->hasZeroCycleZeroing()">; +def IsDarwin : Predicate<"Subtarget->isTargetDarwin()">; +def IsNotDarwin: Predicate<"!Subtarget->isTargetDarwin()">; +def ForCodeSize : Predicate<"ForCodeSize">; +def NotForCodeSize : Predicate<"!ForCodeSize">; +include "AArch64InstrFormats.td" -// (outs GPR64), (ins) -def A64threadpointer : SDNode<"AArch64ISD::THREAD_POINTER", SDTPtrLeaf>; +//===----------------------------------------------------------------------===// -// A64 compares don't care about the cond really (they set all flags) so a -// simple binary operator is useful. -def A64cmp : PatFrag<(ops node:$lhs, node:$rhs), - (A64setcc node:$lhs, node:$rhs, cond)>; +//===----------------------------------------------------------------------===// +// Miscellaneous instructions. +//===----------------------------------------------------------------------===// +let Defs = [SP], Uses = [SP], hasSideEffects = 1, isCodeGenOnly = 1 in { +def ADJCALLSTACKDOWN : Pseudo<(outs), (ins i32imm:$amt), + [(AArch64callseq_start timm:$amt)]>; +def ADJCALLSTACKUP : Pseudo<(outs), (ins i32imm:$amt1, i32imm:$amt2), + [(AArch64callseq_end timm:$amt1, timm:$amt2)]>; +} // Defs = [SP], Uses = [SP], hasSideEffects = 1, isCodeGenOnly = 1 + +let isReMaterializable = 1, isCodeGenOnly = 1 in { +// FIXME: The following pseudo instructions are only needed because remat +// cannot handle multiple instructions. When that changes, they can be +// removed, along with the AArch64Wrapper node. + +let AddedComplexity = 10 in +def LOADgot : Pseudo<(outs GPR64:$dst), (ins i64imm:$addr), + [(set GPR64:$dst, (AArch64LOADgot tglobaladdr:$addr))]>, + Sched<[WriteLDAdr]>; + +// The MOVaddr instruction should match only when the add is not folded +// into a load or store address. +def MOVaddr + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp tglobaladdr:$hi), + tglobaladdr:$low))]>, + Sched<[WriteAdrAdr]>; +def MOVaddrJT + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp tjumptable:$hi), + tjumptable:$low))]>, + Sched<[WriteAdrAdr]>; +def MOVaddrCP + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp tconstpool:$hi), + tconstpool:$low))]>, + Sched<[WriteAdrAdr]>; +def MOVaddrBA + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp tblockaddress:$hi), + tblockaddress:$low))]>, + Sched<[WriteAdrAdr]>; +def MOVaddrTLS + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp tglobaltlsaddr:$hi), + tglobaltlsaddr:$low))]>, + Sched<[WriteAdrAdr]>; +def MOVaddrEXT + : Pseudo<(outs GPR64:$dst), (ins i64imm:$hi, i64imm:$low), + [(set GPR64:$dst, (AArch64addlow (AArch64adrp texternalsym:$hi), + texternalsym:$low))]>, + Sched<[WriteAdrAdr]>; + +} // isReMaterializable, isCodeGenOnly + +def : Pat<(AArch64LOADgot tglobaltlsaddr:$addr), + (LOADgot tglobaltlsaddr:$addr)>; + +def : Pat<(AArch64LOADgot texternalsym:$addr), + (LOADgot texternalsym:$addr)>; + +def : Pat<(AArch64LOADgot tconstpool:$addr), + (LOADgot tconstpool:$addr)>; -// When matching a notional (CMP op1, (sub 0, op2)), we'd like to use a CMN -// instruction on the grounds that "op1 - (-op2) == op1 + op2". However, the C -// and V flags can be set differently by this operation. It comes down to -// whether "SInt(~op2)+1 == SInt(~op2+1)" (and the same for UInt). If they are -// then everything is fine. If not then the optimization is wrong. Thus general -// comparisons are only valid if op2 != 0. +//===----------------------------------------------------------------------===// +// System instructions. +//===----------------------------------------------------------------------===// -// So, finally, the only LLVM-native comparisons that don't mention C and V are -// SETEQ and SETNE. They're the only ones we can safely use CMN for in the -// absence of information about op2. -def equality_cond : PatLeaf<(cond), [{ - return N->get() == ISD::SETEQ || N->get() == ISD::SETNE; -}]>; +def HINT : HintI<"hint">; +def : InstAlias<"nop", (HINT 0b000)>; +def : InstAlias<"yield",(HINT 0b001)>; +def : InstAlias<"wfe", (HINT 0b010)>; +def : InstAlias<"wfi", (HINT 0b011)>; +def : InstAlias<"sev", (HINT 0b100)>; +def : InstAlias<"sevl", (HINT 0b101)>; -def A64cmn : PatFrag<(ops node:$lhs, node:$rhs), - (A64setcc node:$lhs, (sub 0, node:$rhs), equality_cond)>; +// v8.2a Statistical Profiling extension +def : InstAlias<"psb $op", (HINT psbhint_op:$op)>, Requires<[HasSPE]>; -// There are two layers of indirection here, driven by the following -// considerations. -// + TableGen does not know CodeModel or Reloc so that decision should be -// made for a variable/address at ISelLowering. -// + The output of ISelLowering should be selectable (hence the Wrapper, -// rather than a bare target opcode) -def SDTAArch64Wrapper : SDTypeProfile<1, 3, [SDTCisSameAs<0, 1>, - SDTCisSameAs<1, 2>, - SDTCisVT<3, i32>, - SDTCisPtrTy<0>]>; +// As far as LLVM is concerned this writes to the system's exclusive monitors. +let mayLoad = 1, mayStore = 1 in +def CLREX : CRmSystemI; -def A64WrapperSmall : SDNode<"AArch64ISD::WrapperSmall", SDTAArch64Wrapper>; +// NOTE: ideally, this would have mayStore = 0, mayLoad = 0, but we cannot +// model patterns with sufficiently fine granularity. +let mayLoad = ?, mayStore = ? in { +def DMB : CRmSystemI; +def DSB : CRmSystemI; -def SDTAArch64GOTLoad : SDTypeProfile<1, 1, [SDTCisPtrTy<0>, SDTCisPtrTy<1>]>; -def A64GOTLoad : SDNode<"AArch64ISD::GOTLoad", SDTAArch64GOTLoad, - [SDNPHasChain]>; +def ISB : CRmSystemI; +} +def : InstAlias<"clrex", (CLREX 0xf)>; +def : InstAlias<"isb", (ISB 0xf)>; -// (A64BFI LHS, RHS, LSB, Width) -def SDTA64BFI : SDTypeProfile<1, 4, [SDTCisSameAs<0, 1>, - SDTCisSameAs<1, 2>, - SDTCisVT<3, i64>, - SDTCisVT<4, i64>]>; +def MRS : MRSI; +def MSR : MSRI; +def MSRpstateImm1 : MSRpstateImm0_1; +def MSRpstateImm4 : MSRpstateImm0_15; -def A64Bfi : SDNode<"AArch64ISD::BFI", SDTA64BFI>; +// The thread pointer (on Linux, at least, where this has been implemented) is +// TPIDR_EL0. +def : Pat<(AArch64threadpointer), (MRS 0xde82)>; -// (A64EXTR HiReg, LoReg, LSB) -def SDTA64EXTR : SDTypeProfile<1, 3, [SDTCisSameAs<0, 1>, SDTCisSameAs<1, 2>, - SDTCisVT<3, i64>]>; -def A64Extr : SDNode<"AArch64ISD::EXTR", SDTA64EXTR>; +// The cycle counter PMC register is PMCCNTR_EL0. +let Predicates = [HasPerfMon] in +def : Pat<(readcyclecounter), (MRS 0xdce8)>; -// (A64[SU]BFX Field, ImmR, ImmS). -// -// Note that ImmR and ImmS are already encoded for the actual instructions. The -// more natural LSB and Width mix together to form ImmR and ImmS, something -// which TableGen can't handle. -def SDTA64BFX : SDTypeProfile<1, 3, [SDTCisVT<2, i64>, SDTCisVT<3, i64>]>; -def A64Sbfx : SDNode<"AArch64ISD::SBFX", SDTA64BFX>; +// Generic system instructions +def SYSxt : SystemXtI<0, "sys">; +def SYSLxt : SystemLXtI<1, "sysl">; -def A64Ubfx : SDNode<"AArch64ISD::UBFX", SDTA64BFX>; +def : InstAlias<"sys $op1, $Cn, $Cm, $op2", + (SYSxt imm0_7:$op1, sys_cr_op:$Cn, + sys_cr_op:$Cm, imm0_7:$op2, XZR)>; //===----------------------------------------------------------------------===// -// Call sequence pseudo-instructions +// Move immediate instructions. //===----------------------------------------------------------------------===// +defm MOVK : InsertImmediate<0b11, "movk">; +defm MOVN : MoveImmediate<0b00, "movn">; -def SDT_AArch64Call : SDTypeProfile<0, -1, [SDTCisPtrTy<0>]>; -def AArch64Call : SDNode<"AArch64ISD::Call", SDT_AArch64Call, - [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue, SDNPVariadic]>; - -def AArch64tcret : SDNode<"AArch64ISD::TC_RETURN", SDT_AArch64Call, - [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>; - -// The TLSDESCCALL node is a variant call which goes to an indirectly calculated -// destination but needs a relocation against a fixed symbol. As such it has two -// certain operands: the callee and the relocated variable. -// -// The TLS ABI only allows it to be selected to a BLR instructin (with -// appropriate relocation). -def SDTTLSDescCall : SDTypeProfile<0, -2, [SDTCisPtrTy<0>, SDTCisPtrTy<1>]>; - -def A64tlsdesc_blr : SDNode<"AArch64ISD::TLSDESCCALL", SDTTLSDescCall, - [SDNPInGlue, SDNPOutGlue, SDNPHasChain, - SDNPVariadic]>; - - -def SDT_AArch64CallSeqStart : SDCallSeqStart<[ SDTCisPtrTy<0> ]>; -def AArch64callseq_start : SDNode<"ISD::CALLSEQ_START", SDT_AArch64CallSeqStart, - [SDNPHasChain, SDNPOutGlue]>; - -def SDT_AArch64CallSeqEnd : SDCallSeqEnd<[ SDTCisPtrTy<0>, SDTCisPtrTy<1> ]>; -def AArch64callseq_end : SDNode<"ISD::CALLSEQ_END", SDT_AArch64CallSeqEnd, - [SDNPHasChain, SDNPOptInGlue, SDNPOutGlue]>; +let PostEncoderMethod = "fixMOVZ" in +defm MOVZ : MoveImmediate<0b10, "movz">; +// First group of aliases covers an implicit "lsl #0". +def : InstAlias<"movk $dst, $imm", (MOVKWi GPR32:$dst, imm0_65535:$imm, 0)>; +def : InstAlias<"movk $dst, $imm", (MOVKXi GPR64:$dst, imm0_65535:$imm, 0)>; +def : InstAlias<"movn $dst, $imm", (MOVNWi GPR32:$dst, imm0_65535:$imm, 0)>; +def : InstAlias<"movn $dst, $imm", (MOVNXi GPR64:$dst, imm0_65535:$imm, 0)>; +def : InstAlias<"movz $dst, $imm", (MOVZWi GPR32:$dst, imm0_65535:$imm, 0)>; +def : InstAlias<"movz $dst, $imm", (MOVZXi GPR64:$dst, imm0_65535:$imm, 0)>; +// Next, we have various ELF relocations with the ":XYZ_g0:sym" syntax. +def : InstAlias<"movz $Rd, $sym", (MOVZXi GPR64:$Rd, movz_symbol_g3:$sym, 48)>; +def : InstAlias<"movz $Rd, $sym", (MOVZXi GPR64:$Rd, movz_symbol_g2:$sym, 32)>; +def : InstAlias<"movz $Rd, $sym", (MOVZXi GPR64:$Rd, movz_symbol_g1:$sym, 16)>; +def : InstAlias<"movz $Rd, $sym", (MOVZXi GPR64:$Rd, movz_symbol_g0:$sym, 0)>; -// These pseudo-instructions have special semantics by virtue of being passed to -// the InstrInfo constructor. CALLSEQ_START/CALLSEQ_END are produced by -// LowerCall to (in our case) tell the back-end about stack adjustments for -// arguments passed on the stack. Here we select those markers to -// pseudo-instructions which explicitly set the stack, and finally in the -// RegisterInfo we convert them to a true stack adjustment. -let Defs = [XSP], Uses = [XSP] in { - def ADJCALLSTACKDOWN : PseudoInst<(outs), (ins i64imm:$amt), - [(AArch64callseq_start timm:$amt)]>; +def : InstAlias<"movn $Rd, $sym", (MOVNXi GPR64:$Rd, movz_symbol_g3:$sym, 48)>; +def : InstAlias<"movn $Rd, $sym", (MOVNXi GPR64:$Rd, movz_symbol_g2:$sym, 32)>; +def : InstAlias<"movn $Rd, $sym", (MOVNXi GPR64:$Rd, movz_symbol_g1:$sym, 16)>; +def : InstAlias<"movn $Rd, $sym", (MOVNXi GPR64:$Rd, movz_symbol_g0:$sym, 0)>; - def ADJCALLSTACKUP : PseudoInst<(outs), (ins i64imm:$amt1, i64imm:$amt2), - [(AArch64callseq_end timm:$amt1, timm:$amt2)]>; -} +def : InstAlias<"movk $Rd, $sym", (MOVKXi GPR64:$Rd, movk_symbol_g3:$sym, 48)>; +def : InstAlias<"movk $Rd, $sym", (MOVKXi GPR64:$Rd, movk_symbol_g2:$sym, 32)>; +def : InstAlias<"movk $Rd, $sym", (MOVKXi GPR64:$Rd, movk_symbol_g1:$sym, 16)>; +def : InstAlias<"movk $Rd, $sym", (MOVKXi GPR64:$Rd, movk_symbol_g0:$sym, 0)>; -//===----------------------------------------------------------------------===// -// Atomic operation pseudo-instructions -//===----------------------------------------------------------------------===// +def : InstAlias<"movz $Rd, $sym", (MOVZWi GPR32:$Rd, movz_symbol_g1:$sym, 16)>; +def : InstAlias<"movz $Rd, $sym", (MOVZWi GPR32:$Rd, movz_symbol_g0:$sym, 0)>; -let usesCustomInserter = 1, Defs = [NZCV] in { -multiclass AtomicSizes { - def _I8 : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$incr), - [(set GPR32:$dst, (!cast(opname # "_8") GPR64:$ptr, GPR32:$incr))]>; - def _I16 : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$incr), - [(set GPR32:$dst, (!cast(opname # "_16") GPR64:$ptr, GPR32:$incr))]>; - def _I32 : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$incr), - [(set GPR32:$dst, (!cast(opname # "_32") GPR64:$ptr, GPR32:$incr))]>; - def _I64 : PseudoInst<(outs GPR64:$dst), (ins GPR64:$ptr, GPR64:$incr), - [(set GPR64:$dst, (!cast(opname # "_64") GPR64:$ptr, GPR64:$incr))]>; -} -} +def : InstAlias<"movn $Rd, $sym", (MOVNWi GPR32:$Rd, movz_symbol_g1:$sym, 16)>; +def : InstAlias<"movn $Rd, $sym", (MOVNWi GPR32:$Rd, movz_symbol_g0:$sym, 0)>; -defm ATOMIC_LOAD_ADD : AtomicSizes<"atomic_load_add">; -defm ATOMIC_LOAD_SUB : AtomicSizes<"atomic_load_sub">; -defm ATOMIC_LOAD_AND : AtomicSizes<"atomic_load_and">; -defm ATOMIC_LOAD_OR : AtomicSizes<"atomic_load_or">; -defm ATOMIC_LOAD_XOR : AtomicSizes<"atomic_load_xor">; -defm ATOMIC_LOAD_NAND : AtomicSizes<"atomic_load_nand">; -defm ATOMIC_LOAD_MIN : AtomicSizes<"atomic_load_min">; -defm ATOMIC_LOAD_MAX : AtomicSizes<"atomic_load_max">; -defm ATOMIC_LOAD_UMIN : AtomicSizes<"atomic_load_umin">; -defm ATOMIC_LOAD_UMAX : AtomicSizes<"atomic_load_umax">; -defm ATOMIC_SWAP : AtomicSizes<"atomic_swap">; - -let usesCustomInserter = 1, Defs = [NZCV] in { -def ATOMIC_CMP_SWAP_I8 - : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$old, GPR32:$new), - [(set GPR32:$dst, - (atomic_cmp_swap_8 GPR64:$ptr, GPR32:$old, GPR32:$new))]>; -def ATOMIC_CMP_SWAP_I16 - : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$old, GPR32:$new), - [(set GPR32:$dst, - (atomic_cmp_swap_16 GPR64:$ptr, GPR32:$old, GPR32:$new))]>; -def ATOMIC_CMP_SWAP_I32 - : PseudoInst<(outs GPR32:$dst), (ins GPR64:$ptr, GPR32:$old, GPR32:$new), - [(set GPR32:$dst, - (atomic_cmp_swap_32 GPR64:$ptr, GPR32:$old, GPR32:$new))]>; -def ATOMIC_CMP_SWAP_I64 - : PseudoInst<(outs GPR64:$dst), (ins GPR64:$ptr, GPR64:$old, GPR64:$new), - [(set GPR64:$dst, - (atomic_cmp_swap_64 GPR64:$ptr, GPR64:$old, GPR64:$new))]>; -} +def : InstAlias<"movk $Rd, $sym", (MOVKWi GPR32:$Rd, movk_symbol_g1:$sym, 16)>; +def : InstAlias<"movk $Rd, $sym", (MOVKWi GPR32:$Rd, movk_symbol_g0:$sym, 0)>; -//===----------------------------------------------------------------------===// -// Add-subtract (extended register) instructions -//===----------------------------------------------------------------------===// -// Contains: ADD, ADDS, SUB, SUBS + aliases CMN, CMP - -// The RHS of these operations is conceptually a sign/zero-extended -// register, optionally shifted left by 1-4. The extension can be a -// NOP (e.g. "sxtx" sign-extending a 64-bit register to 64-bits) but -// must be specified with one exception: - -// If one of the registers is sp/wsp then LSL is an alias for UXTW in -// 32-bit instructions and UXTX in 64-bit versions, the shift amount -// is not optional in that case (but can explicitly be 0), and the -// entire suffix can be skipped (e.g. "add sp, x3, x2"). - -multiclass extend_operands { - def _asmoperand : AsmOperandClass { - let Name = PREFIX; - let RenderMethod = "addRegExtendOperands"; - let PredicateMethod = "isRegExtend"; - let DiagnosticType = "AddSubRegExtend" # Diag; - } - - def _operand : Operand, - ImmLeaf= 0 && Imm <= 4; }]> { - let PrintMethod = "printRegExtendOperand"; - let DecoderMethod = "DecodeRegExtendOperand"; - let ParserMatchClass = !cast(PREFIX # "_asmoperand"); - } -} +// Final group of aliases covers true "mov $Rd, $imm" cases. +multiclass movw_mov_alias { + def _asmoperand : AsmOperandClass { + let Name = basename # width # "_lsl" # shift # "MovAlias"; + let PredicateMethod = "is" # basename # "MovAlias<" # width # ", " + # shift # ">"; + let RenderMethod = "add" # basename # "MovAliasOperands<" # shift # ">"; + } -defm UXTB : extend_operands<"UXTB", "Small">; -defm UXTH : extend_operands<"UXTH", "Small">; -defm UXTW : extend_operands<"UXTW", "Small">; -defm UXTX : extend_operands<"UXTX", "Large">; -defm SXTB : extend_operands<"SXTB", "Small">; -defm SXTH : extend_operands<"SXTH", "Small">; -defm SXTW : extend_operands<"SXTW", "Small">; -defm SXTX : extend_operands<"SXTX", "Large">; - -def LSL_extasmoperand : AsmOperandClass { - let Name = "RegExtendLSL"; - let RenderMethod = "addRegExtendOperands"; - let DiagnosticType = "AddSubRegExtendLarge"; -} + def _movimm : Operand { + let ParserMatchClass = !cast(NAME # "_asmoperand"); + } -def LSL_extoperand : Operand { - let ParserMatchClass = LSL_extasmoperand; -} + def : InstAlias<"mov $Rd, $imm", + (INST GPR:$Rd, !cast(NAME # "_movimm"):$imm, shift)>; +} + +defm : movw_mov_alias<"MOVZ", MOVZWi, GPR32, 32, 0>; +defm : movw_mov_alias<"MOVZ", MOVZWi, GPR32, 32, 16>; + +defm : movw_mov_alias<"MOVZ", MOVZXi, GPR64, 64, 0>; +defm : movw_mov_alias<"MOVZ", MOVZXi, GPR64, 64, 16>; +defm : movw_mov_alias<"MOVZ", MOVZXi, GPR64, 64, 32>; +defm : movw_mov_alias<"MOVZ", MOVZXi, GPR64, 64, 48>; + +defm : movw_mov_alias<"MOVN", MOVNWi, GPR32, 32, 0>; +defm : movw_mov_alias<"MOVN", MOVNWi, GPR32, 32, 16>; + +defm : movw_mov_alias<"MOVN", MOVNXi, GPR64, 64, 0>; +defm : movw_mov_alias<"MOVN", MOVNXi, GPR64, 64, 16>; +defm : movw_mov_alias<"MOVN", MOVNXi, GPR64, 64, 32>; +defm : movw_mov_alias<"MOVN", MOVNXi, GPR64, 64, 48>; + +let isReMaterializable = 1, isCodeGenOnly = 1, isMoveImm = 1, + isAsCheapAsAMove = 1 in { +// FIXME: The following pseudo instructions are only needed because remat +// cannot handle multiple instructions. When that changes, we can select +// directly to the real instructions and get rid of these pseudos. + +def MOVi32imm + : Pseudo<(outs GPR32:$dst), (ins i32imm:$src), + [(set GPR32:$dst, imm:$src)]>, + Sched<[WriteImm]>; +def MOVi64imm + : Pseudo<(outs GPR64:$dst), (ins i64imm:$src), + [(set GPR64:$dst, imm:$src)]>, + Sched<[WriteImm]>; +} // isReMaterializable, isCodeGenOnly + +// If possible, we want to use MOVi32imm even for 64-bit moves. This gives the +// eventual expansion code fewer bits to worry about getting right. Marshalling +// the types is a little tricky though: +def i64imm_32bit : ImmLeaf(Imm); +}]>; +def trunc_imm : SDNodeXFormgetTargetConstant(N->getZExtValue(), SDLoc(N), MVT::i32); +}]>; -// The patterns for various sign-extensions are a little ugly and -// non-uniform because everything has already been promoted to the -// legal i64 and i32 types. We'll wrap the various variants up in a -// class for use later. -class extend_types { - dag uxtb; dag uxth; dag uxtw; dag uxtx; - dag sxtb; dag sxth; dag sxtw; dag sxtx; -} +def : Pat<(i64 i64imm_32bit:$src), + (SUBREG_TO_REG (i64 0), (MOVi32imm (trunc_imm imm:$src)), sub_32)>; -def extends_to_i64 : extend_types { - let uxtb = (and (anyext GPR32:$Rm), 255); - let uxth = (and (anyext GPR32:$Rm), 65535); - let uxtw = (zext GPR32:$Rm); - let uxtx = (i64 GPR64:$Rm); +// Materialize FP constants via MOVi32imm/MOVi64imm (MachO large code model). +def bitcast_fpimm_to_i32 : SDNodeXFormgetTargetConstant( + N->getValueAPF().bitcastToAPInt().getZExtValue(), SDLoc(N), MVT::i32); +}]>; - let sxtb = (sext_inreg (anyext GPR32:$Rm), i8); - let sxth = (sext_inreg (anyext GPR32:$Rm), i16); - let sxtw = (sext GPR32:$Rm); - let sxtx = (i64 GPR64:$Rm); -} +def bitcast_fpimm_to_i64 : SDNodeXFormgetTargetConstant( + N->getValueAPF().bitcastToAPInt().getZExtValue(), SDLoc(N), MVT::i64); +}]>; -def extends_to_i32 : extend_types { - let uxtb = (and GPR32:$Rm, 255); - let uxth = (and GPR32:$Rm, 65535); - let uxtw = (i32 GPR32:$Rm); - let uxtx = (i32 GPR32:$Rm); +def : Pat<(f32 fpimm:$in), + (COPY_TO_REGCLASS (MOVi32imm (bitcast_fpimm_to_i32 f32:$in)), FPR32)>; +def : Pat<(f64 fpimm:$in), + (COPY_TO_REGCLASS (MOVi64imm (bitcast_fpimm_to_i64 f64:$in)), FPR64)>; - let sxtb = (sext_inreg GPR32:$Rm, i8); - let sxth = (sext_inreg GPR32:$Rm, i16); - let sxtw = (i32 GPR32:$Rm); - let sxtx = (i32 GPR32:$Rm); -} -// Now, six of the extensions supported are easy and uniform: if the source size -// is 32-bits or less, then Rm is always a 32-bit register. We'll instantiate -// those instructions in one block. - -// The uxtx/sxtx could potentially be merged in, but three facts dissuaded me: -// + It would break the naming scheme: either ADDxx_uxtx or ADDww_uxtx would -// be impossible. -// + Patterns are very different as well. -// + Passing different registers would be ugly (more fields in extend_types -// would probably be the best option). -multiclass addsub_exts { - def w_uxtb : A64I_addsubext; - def w_uxth : A64I_addsubext; - def w_uxtw : A64I_addsubext; - - def w_sxtb : A64I_addsubext; - def w_sxth : A64I_addsubext; - def w_sxtw : A64I_addsubext; -} +// Deal with the various forms of (ELF) large addressing with MOVZ/MOVK +// sequences. +def : Pat<(AArch64WrapperLarge tglobaladdr:$g3, tglobaladdr:$g2, + tglobaladdr:$g1, tglobaladdr:$g0), + (MOVKXi (MOVKXi (MOVKXi (MOVZXi tglobaladdr:$g3, 48), + tglobaladdr:$g2, 32), + tglobaladdr:$g1, 16), + tglobaladdr:$g0, 0)>; -// These two could be merge in with the above, but their patterns aren't really -// necessary and the naming-scheme would necessarily break: -multiclass addsub_xxtx { - def x_uxtx : A64I_addsubext<0b1, op, S, 0b00, 0b011, - outs, - (ins GPR64xsp:$Rn, GPR64:$Rm, UXTX_operand:$Imm3), - !strconcat(asmop, "$Rn, $Rm, $Imm3"), - [(opfrag GPR64xsp:$Rn, (shl GPR64:$Rm, UXTX_operand:$Imm3))], - NoItinerary>; - - def x_sxtx : A64I_addsubext<0b1, op, S, 0b00, 0b111, - outs, - (ins GPR64xsp:$Rn, GPR64:$Rm, SXTX_operand:$Imm3), - !strconcat(asmop, "$Rn, $Rm, $Imm3"), - [/* No Pattern: same as uxtx */], - NoItinerary>; -} +def : Pat<(AArch64WrapperLarge tblockaddress:$g3, tblockaddress:$g2, + tblockaddress:$g1, tblockaddress:$g0), + (MOVKXi (MOVKXi (MOVKXi (MOVZXi tblockaddress:$g3, 48), + tblockaddress:$g2, 32), + tblockaddress:$g1, 16), + tblockaddress:$g0, 0)>; -multiclass addsub_wxtx { - def w_uxtx : A64I_addsubext<0b0, op, S, 0b00, 0b011, - outs, - (ins GPR32wsp:$Rn, GPR32:$Rm, UXTX_operand:$Imm3), - !strconcat(asmop, "$Rn, $Rm, $Imm3"), - [/* No pattern: probably same as uxtw */], - NoItinerary>; - - def w_sxtx : A64I_addsubext<0b0, op, S, 0b00, 0b111, - outs, - (ins GPR32wsp:$Rn, GPR32:$Rm, SXTX_operand:$Imm3), - !strconcat(asmop, "$Rn, $Rm, $Imm3"), - [/* No Pattern: probably same as uxtw */], - NoItinerary>; -} +def : Pat<(AArch64WrapperLarge tconstpool:$g3, tconstpool:$g2, + tconstpool:$g1, tconstpool:$g0), + (MOVKXi (MOVKXi (MOVKXi (MOVZXi tconstpool:$g3, 48), + tconstpool:$g2, 32), + tconstpool:$g1, 16), + tconstpool:$g0, 0)>; -class SetRD - : PatFrag<(ops node:$lhs, node:$rhs), (set RC:$Rd, (op node:$lhs, node:$rhs))>; -class SetNZCV - : PatFrag<(ops node:$lhs, node:$rhs), (set NZCV, (op node:$lhs, node:$rhs))>; - -defm ADDxx :addsub_exts<0b1, 0b0, 0b0, "add\t$Rd, ", SetRD, - (outs GPR64xsp:$Rd), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b0, 0b0, "add\t$Rd, ", SetRD, - (outs GPR64xsp:$Rd)>; -defm ADDww :addsub_exts<0b0, 0b0, 0b0, "add\t$Rd, ", SetRD, - (outs GPR32wsp:$Rd), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b0, 0b0, "add\t$Rd, ", - (outs GPR32wsp:$Rd)>; -defm SUBxx :addsub_exts<0b1, 0b1, 0b0, "sub\t$Rd, ", SetRD, - (outs GPR64xsp:$Rd), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b1, 0b0, "sub\t$Rd, ", SetRD, - (outs GPR64xsp:$Rd)>; -defm SUBww :addsub_exts<0b0, 0b1, 0b0, "sub\t$Rd, ", SetRD, - (outs GPR32wsp:$Rd), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b1, 0b0, "sub\t$Rd, ", - (outs GPR32wsp:$Rd)>; - -let Defs = [NZCV] in { -defm ADDSxx :addsub_exts<0b1, 0b0, 0b1, "adds\t$Rd, ", SetRD, - (outs GPR64:$Rd), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b0, 0b1, "adds\t$Rd, ", SetRD, - (outs GPR64:$Rd)>; -defm ADDSww :addsub_exts<0b0, 0b0, 0b1, "adds\t$Rd, ", SetRD, - (outs GPR32:$Rd), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b0, 0b1, "adds\t$Rd, ", - (outs GPR32:$Rd)>; -defm SUBSxx :addsub_exts<0b1, 0b1, 0b1, "subs\t$Rd, ", SetRD, - (outs GPR64:$Rd), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b1, 0b1, "subs\t$Rd, ", SetRD, - (outs GPR64:$Rd)>; -defm SUBSww :addsub_exts<0b0, 0b1, 0b1, "subs\t$Rd, ", SetRD, - (outs GPR32:$Rd), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b1, 0b1, "subs\t$Rd, ", - (outs GPR32:$Rd)>; - - -let Rd = 0b11111, isCompare = 1 in { -defm CMNx : addsub_exts<0b1, 0b0, 0b1, "cmn\t", SetNZCV, - (outs), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b0, 0b1, "cmn\t", SetNZCV, (outs)>; -defm CMNw : addsub_exts<0b0, 0b0, 0b1, "cmn\t", SetNZCV, - (outs), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b0, 0b1, "cmn\t", (outs)>; -defm CMPx : addsub_exts<0b1, 0b1, 0b1, "cmp\t", SetNZCV, - (outs), extends_to_i64, GPR64xsp>, - addsub_xxtx< 0b1, 0b1, "cmp\t", SetNZCV, (outs)>; -defm CMPw : addsub_exts<0b0, 0b1, 0b1, "cmp\t", SetNZCV, - (outs), extends_to_i32, GPR32wsp>, - addsub_wxtx< 0b1, 0b1, "cmp\t", (outs)>; -} -} +def : Pat<(AArch64WrapperLarge tjumptable:$g3, tjumptable:$g2, + tjumptable:$g1, tjumptable:$g0), + (MOVKXi (MOVKXi (MOVKXi (MOVZXi tjumptable:$g3, 48), + tjumptable:$g2, 32), + tjumptable:$g1, 16), + tjumptable:$g0, 0)>; -// Now patterns for the operation without a shift being needed. No patterns are -// created for uxtx/sxtx since they're non-uniform and it's expected that -// add/sub (shifted register) will handle those cases anyway. -multiclass addsubext_noshift_patterns { - def : Pat<(nodeop GPRsp:$Rn, exts.uxtb), - (!cast(prefix # "w_uxtb") GPRsp:$Rn, GPR32:$Rm, 0)>; - def : Pat<(nodeop GPRsp:$Rn, exts.uxth), - (!cast(prefix # "w_uxth") GPRsp:$Rn, GPR32:$Rm, 0)>; - def : Pat<(nodeop GPRsp:$Rn, exts.uxtw), - (!cast(prefix # "w_uxtw") GPRsp:$Rn, GPR32:$Rm, 0)>; - - def : Pat<(nodeop GPRsp:$Rn, exts.sxtb), - (!cast(prefix # "w_sxtb") GPRsp:$Rn, GPR32:$Rm, 0)>; - def : Pat<(nodeop GPRsp:$Rn, exts.sxth), - (!cast(prefix # "w_sxth") GPRsp:$Rn, GPR32:$Rm, 0)>; - def : Pat<(nodeop GPRsp:$Rn, exts.sxtw), - (!cast(prefix # "w_sxtw") GPRsp:$Rn, GPR32:$Rm, 0)>; -} -defm : addsubext_noshift_patterns<"ADDxx", add, GPR64xsp, extends_to_i64>; -defm : addsubext_noshift_patterns<"ADDww", add, GPR32wsp, extends_to_i32>; -defm : addsubext_noshift_patterns<"SUBxx", sub, GPR64xsp, extends_to_i64>; -defm : addsubext_noshift_patterns<"SUBww", sub, GPR32wsp, extends_to_i32>; +//===----------------------------------------------------------------------===// +// Arithmetic instructions. +//===----------------------------------------------------------------------===// -defm : addsubext_noshift_patterns<"CMNx", A64cmn, GPR64xsp, extends_to_i64>; -defm : addsubext_noshift_patterns<"CMNw", A64cmn, GPR32wsp, extends_to_i32>; -defm : addsubext_noshift_patterns<"CMPx", A64cmp, GPR64xsp, extends_to_i64>; -defm : addsubext_noshift_patterns<"CMPw", A64cmp, GPR32wsp, extends_to_i32>; +// Add/subtract with carry. +defm ADC : AddSubCarry<0, "adc", "adcs", AArch64adc, AArch64adc_flag>; +defm SBC : AddSubCarry<1, "sbc", "sbcs", AArch64sbc, AArch64sbc_flag>; + +def : InstAlias<"ngc $dst, $src", (SBCWr GPR32:$dst, WZR, GPR32:$src)>; +def : InstAlias<"ngc $dst, $src", (SBCXr GPR64:$dst, XZR, GPR64:$src)>; +def : InstAlias<"ngcs $dst, $src", (SBCSWr GPR32:$dst, WZR, GPR32:$src)>; +def : InstAlias<"ngcs $dst, $src", (SBCSXr GPR64:$dst, XZR, GPR64:$src)>; + +// Add/subtract +defm ADD : AddSub<0, "add", "sub", add>; +defm SUB : AddSub<1, "sub", "add">; + +def : InstAlias<"mov $dst, $src", + (ADDWri GPR32sponly:$dst, GPR32sp:$src, 0, 0)>; +def : InstAlias<"mov $dst, $src", + (ADDWri GPR32sp:$dst, GPR32sponly:$src, 0, 0)>; +def : InstAlias<"mov $dst, $src", + (ADDXri GPR64sponly:$dst, GPR64sp:$src, 0, 0)>; +def : InstAlias<"mov $dst, $src", + (ADDXri GPR64sp:$dst, GPR64sponly:$src, 0, 0)>; + +defm ADDS : AddSubS<0, "adds", AArch64add_flag, "cmn", "subs", "cmp">; +defm SUBS : AddSubS<1, "subs", AArch64sub_flag, "cmp", "adds", "cmn">; + +// Use SUBS instead of SUB to enable CSE between SUBS and SUB. +def : Pat<(sub GPR32sp:$Rn, addsub_shifted_imm32:$imm), + (SUBSWri GPR32sp:$Rn, addsub_shifted_imm32:$imm)>; +def : Pat<(sub GPR64sp:$Rn, addsub_shifted_imm64:$imm), + (SUBSXri GPR64sp:$Rn, addsub_shifted_imm64:$imm)>; +def : Pat<(sub GPR32:$Rn, GPR32:$Rm), + (SUBSWrr GPR32:$Rn, GPR32:$Rm)>; +def : Pat<(sub GPR64:$Rn, GPR64:$Rm), + (SUBSXrr GPR64:$Rn, GPR64:$Rm)>; +def : Pat<(sub GPR32:$Rn, arith_shifted_reg32:$Rm), + (SUBSWrs GPR32:$Rn, arith_shifted_reg32:$Rm)>; +def : Pat<(sub GPR64:$Rn, arith_shifted_reg64:$Rm), + (SUBSXrs GPR64:$Rn, arith_shifted_reg64:$Rm)>; +let AddedComplexity = 1 in { +def : Pat<(sub GPR32sp:$R2, arith_extended_reg32:$R3), + (SUBSWrx GPR32sp:$R2, arith_extended_reg32:$R3)>; +def : Pat<(sub GPR64sp:$R2, arith_extended_reg32to64:$R3), + (SUBSXrx GPR64sp:$R2, arith_extended_reg32to64:$R3)>; +} + +// Because of the immediate format for add/sub-imm instructions, the +// expression (add x, -1) must be transformed to (SUB{W,X}ri x, 1). +// These patterns capture that transformation. +let AddedComplexity = 1 in { +def : Pat<(add GPR32:$Rn, neg_addsub_shifted_imm32:$imm), + (SUBSWri GPR32:$Rn, neg_addsub_shifted_imm32:$imm)>; +def : Pat<(add GPR64:$Rn, neg_addsub_shifted_imm64:$imm), + (SUBSXri GPR64:$Rn, neg_addsub_shifted_imm64:$imm)>; +def : Pat<(sub GPR32:$Rn, neg_addsub_shifted_imm32:$imm), + (ADDWri GPR32:$Rn, neg_addsub_shifted_imm32:$imm)>; +def : Pat<(sub GPR64:$Rn, neg_addsub_shifted_imm64:$imm), + (ADDXri GPR64:$Rn, neg_addsub_shifted_imm64:$imm)>; +} + +// Because of the immediate format for add/sub-imm instructions, the +// expression (add x, -1) must be transformed to (SUB{W,X}ri x, 1). +// These patterns capture that transformation. +let AddedComplexity = 1 in { +def : Pat<(AArch64add_flag GPR32:$Rn, neg_addsub_shifted_imm32:$imm), + (SUBSWri GPR32:$Rn, neg_addsub_shifted_imm32:$imm)>; +def : Pat<(AArch64add_flag GPR64:$Rn, neg_addsub_shifted_imm64:$imm), + (SUBSXri GPR64:$Rn, neg_addsub_shifted_imm64:$imm)>; +def : Pat<(AArch64sub_flag GPR32:$Rn, neg_addsub_shifted_imm32:$imm), + (ADDSWri GPR32:$Rn, neg_addsub_shifted_imm32:$imm)>; +def : Pat<(AArch64sub_flag GPR64:$Rn, neg_addsub_shifted_imm64:$imm), + (ADDSXri GPR64:$Rn, neg_addsub_shifted_imm64:$imm)>; +} + +def : InstAlias<"neg $dst, $src", (SUBWrs GPR32:$dst, WZR, GPR32:$src, 0), 3>; +def : InstAlias<"neg $dst, $src", (SUBXrs GPR64:$dst, XZR, GPR64:$src, 0), 3>; +def : InstAlias<"neg $dst, $src$shift", + (SUBWrs GPR32:$dst, WZR, GPR32:$src, arith_shift32:$shift), 2>; +def : InstAlias<"neg $dst, $src$shift", + (SUBXrs GPR64:$dst, XZR, GPR64:$src, arith_shift64:$shift), 2>; + +def : InstAlias<"negs $dst, $src", (SUBSWrs GPR32:$dst, WZR, GPR32:$src, 0), 3>; +def : InstAlias<"negs $dst, $src", (SUBSXrs GPR64:$dst, XZR, GPR64:$src, 0), 3>; +def : InstAlias<"negs $dst, $src$shift", + (SUBSWrs GPR32:$dst, WZR, GPR32:$src, arith_shift32:$shift), 2>; +def : InstAlias<"negs $dst, $src$shift", + (SUBSXrs GPR64:$dst, XZR, GPR64:$src, arith_shift64:$shift), 2>; + + +// Unsigned/Signed divide +defm UDIV : Div<0, "udiv", udiv>; +defm SDIV : Div<1, "sdiv", sdiv>; +let isCodeGenOnly = 1 in { +defm UDIV_Int : Div<0, "udiv", int_aarch64_udiv>; +defm SDIV_Int : Div<1, "sdiv", int_aarch64_sdiv>; +} + +// Variable shift +defm ASRV : Shift<0b10, "asr", sra>; +defm LSLV : Shift<0b00, "lsl", shl>; +defm LSRV : Shift<0b01, "lsr", srl>; +defm RORV : Shift<0b11, "ror", rotr>; + +def : ShiftAlias<"asrv", ASRVWr, GPR32>; +def : ShiftAlias<"asrv", ASRVXr, GPR64>; +def : ShiftAlias<"lslv", LSLVWr, GPR32>; +def : ShiftAlias<"lslv", LSLVXr, GPR64>; +def : ShiftAlias<"lsrv", LSRVWr, GPR32>; +def : ShiftAlias<"lsrv", LSRVXr, GPR64>; +def : ShiftAlias<"rorv", RORVWr, GPR32>; +def : ShiftAlias<"rorv", RORVXr, GPR64>; + +// Multiply-add +let AddedComplexity = 7 in { +defm MADD : MulAccum<0, "madd", add>; +defm MSUB : MulAccum<1, "msub", sub>; + +def : Pat<(i32 (mul GPR32:$Rn, GPR32:$Rm)), + (MADDWrrr GPR32:$Rn, GPR32:$Rm, WZR)>; +def : Pat<(i64 (mul GPR64:$Rn, GPR64:$Rm)), + (MADDXrrr GPR64:$Rn, GPR64:$Rm, XZR)>; + +def : Pat<(i32 (ineg (mul GPR32:$Rn, GPR32:$Rm))), + (MSUBWrrr GPR32:$Rn, GPR32:$Rm, WZR)>; +def : Pat<(i64 (ineg (mul GPR64:$Rn, GPR64:$Rm))), + (MSUBXrrr GPR64:$Rn, GPR64:$Rm, XZR)>; +def : Pat<(i32 (mul (ineg GPR32:$Rn), GPR32:$Rm)), + (MSUBWrrr GPR32:$Rn, GPR32:$Rm, WZR)>; +def : Pat<(i64 (mul (ineg GPR64:$Rn), GPR64:$Rm)), + (MSUBXrrr GPR64:$Rn, GPR64:$Rm, XZR)>; +} // AddedComplexity = 7 + +let AddedComplexity = 5 in { +def SMADDLrrr : WideMulAccum<0, 0b001, "smaddl", add, sext>; +def SMSUBLrrr : WideMulAccum<1, 0b001, "smsubl", sub, sext>; +def UMADDLrrr : WideMulAccum<0, 0b101, "umaddl", add, zext>; +def UMSUBLrrr : WideMulAccum<1, 0b101, "umsubl", sub, zext>; + +def : Pat<(i64 (mul (sext GPR32:$Rn), (sext GPR32:$Rm))), + (SMADDLrrr GPR32:$Rn, GPR32:$Rm, XZR)>; +def : Pat<(i64 (mul (zext GPR32:$Rn), (zext GPR32:$Rm))), + (UMADDLrrr GPR32:$Rn, GPR32:$Rm, XZR)>; + +def : Pat<(i64 (ineg (mul (sext GPR32:$Rn), (sext GPR32:$Rm)))), + (SMSUBLrrr GPR32:$Rn, GPR32:$Rm, XZR)>; +def : Pat<(i64 (ineg (mul (zext GPR32:$Rn), (zext GPR32:$Rm)))), + (UMSUBLrrr GPR32:$Rn, GPR32:$Rm, XZR)>; +} // AddedComplexity = 5 + +def : MulAccumWAlias<"mul", MADDWrrr>; +def : MulAccumXAlias<"mul", MADDXrrr>; +def : MulAccumWAlias<"mneg", MSUBWrrr>; +def : MulAccumXAlias<"mneg", MSUBXrrr>; +def : WideMulAccumAlias<"smull", SMADDLrrr>; +def : WideMulAccumAlias<"smnegl", SMSUBLrrr>; +def : WideMulAccumAlias<"umull", UMADDLrrr>; +def : WideMulAccumAlias<"umnegl", UMSUBLrrr>; + +// Multiply-high +def SMULHrr : MulHi<0b010, "smulh", mulhs>; +def UMULHrr : MulHi<0b110, "umulh", mulhu>; + +// CRC32 +def CRC32Brr : BaseCRC32<0, 0b00, 0, GPR32, int_aarch64_crc32b, "crc32b">; +def CRC32Hrr : BaseCRC32<0, 0b01, 0, GPR32, int_aarch64_crc32h, "crc32h">; +def CRC32Wrr : BaseCRC32<0, 0b10, 0, GPR32, int_aarch64_crc32w, "crc32w">; +def CRC32Xrr : BaseCRC32<1, 0b11, 0, GPR64, int_aarch64_crc32x, "crc32x">; + +def CRC32CBrr : BaseCRC32<0, 0b00, 1, GPR32, int_aarch64_crc32cb, "crc32cb">; +def CRC32CHrr : BaseCRC32<0, 0b01, 1, GPR32, int_aarch64_crc32ch, "crc32ch">; +def CRC32CWrr : BaseCRC32<0, 0b10, 1, GPR32, int_aarch64_crc32cw, "crc32cw">; +def CRC32CXrr : BaseCRC32<1, 0b11, 1, GPR64, int_aarch64_crc32cx, "crc32cx">; + +// v8.1 atomic CAS +defm CAS : CompareAndSwap<0, 0, "">; +defm CASA : CompareAndSwap<1, 0, "a">; +defm CASL : CompareAndSwap<0, 1, "l">; +defm CASAL : CompareAndSwap<1, 1, "al">; + +// v8.1 atomic CASP +defm CASP : CompareAndSwapPair<0, 0, "">; +defm CASPA : CompareAndSwapPair<1, 0, "a">; +defm CASPL : CompareAndSwapPair<0, 1, "l">; +defm CASPAL : CompareAndSwapPair<1, 1, "al">; + +// v8.1 atomic SWP +defm SWP : Swap<0, 0, "">; +defm SWPA : Swap<1, 0, "a">; +defm SWPL : Swap<0, 1, "l">; +defm SWPAL : Swap<1, 1, "al">; + +// v8.1 atomic LD(register). Performs load and then ST(register) +defm LDADD : LDOPregister<0b000, "add", 0, 0, "">; +defm LDADDA : LDOPregister<0b000, "add", 1, 0, "a">; +defm LDADDL : LDOPregister<0b000, "add", 0, 1, "l">; +defm LDADDAL : LDOPregister<0b000, "add", 1, 1, "al">; + +defm LDCLR : LDOPregister<0b001, "clr", 0, 0, "">; +defm LDCLRA : LDOPregister<0b001, "clr", 1, 0, "a">; +defm LDCLRL : LDOPregister<0b001, "clr", 0, 1, "l">; +defm LDCLRAL : LDOPregister<0b001, "clr", 1, 1, "al">; + +defm LDEOR : LDOPregister<0b010, "eor", 0, 0, "">; +defm LDEORA : LDOPregister<0b010, "eor", 1, 0, "a">; +defm LDEORL : LDOPregister<0b010, "eor", 0, 1, "l">; +defm LDEORAL : LDOPregister<0b010, "eor", 1, 1, "al">; + +defm LDSET : LDOPregister<0b011, "set", 0, 0, "">; +defm LDSETA : LDOPregister<0b011, "set", 1, 0, "a">; +defm LDSETL : LDOPregister<0b011, "set", 0, 1, "l">; +defm LDSETAL : LDOPregister<0b011, "set", 1, 1, "al">; + +defm LDSMAX : LDOPregister<0b100, "smax", 0, 0, "">; +defm LDSMAXA : LDOPregister<0b100, "smax", 1, 0, "a">; +defm LDSMAXL : LDOPregister<0b100, "smax", 0, 1, "l">; +defm LDSMAXAL : LDOPregister<0b100, "smax", 1, 1, "al">; + +defm LDSMIN : LDOPregister<0b101, "smin", 0, 0, "">; +defm LDSMINA : LDOPregister<0b101, "smin", 1, 0, "a">; +defm LDSMINL : LDOPregister<0b101, "smin", 0, 1, "l">; +defm LDSMINAL : LDOPregister<0b101, "smin", 1, 1, "al">; + +defm LDUMAX : LDOPregister<0b110, "umax", 0, 0, "">; +defm LDUMAXA : LDOPregister<0b110, "umax", 1, 0, "a">; +defm LDUMAXL : LDOPregister<0b110, "umax", 0, 1, "l">; +defm LDUMAXAL : LDOPregister<0b110, "umax", 1, 1, "al">; + +defm LDUMIN : LDOPregister<0b111, "umin", 0, 0, "">; +defm LDUMINA : LDOPregister<0b111, "umin", 1, 0, "a">; +defm LDUMINL : LDOPregister<0b111, "umin", 0, 1, "l">; +defm LDUMINAL : LDOPregister<0b111, "umin", 1, 1, "al">; + +// v8.1 atomic ST(register) as aliases to "LD(register) when Rt=xZR" +defm : STOPregister<"stadd","LDADD">; // STADDx +defm : STOPregister<"stclr","LDCLR">; // STCLRx +defm : STOPregister<"steor","LDEOR">; // STEORx +defm : STOPregister<"stset","LDSET">; // STSETx +defm : STOPregister<"stsmax","LDSMAX">;// STSMAXx +defm : STOPregister<"stsmin","LDSMIN">;// STSMINx +defm : STOPregister<"stumax","LDUMAX">;// STUMAXx +defm : STOPregister<"stumin","LDUMIN">;// STUMINx -// An extend of "lsl #imm" is valid if and only if one of Rn and Rd is -// sp/wsp. It is synonymous with uxtx/uxtw depending on the size of the -// operation. Also permitted in this case is complete omission of the argument, -// which implies "lsl #0". -multiclass lsl_aliases { - def : InstAlias; +//===----------------------------------------------------------------------===// +// Logical instructions. +//===----------------------------------------------------------------------===// - def : InstAlias; +// (immediate) +defm ANDS : LogicalImmS<0b11, "ands", AArch64and_flag, "bics">; +defm AND : LogicalImm<0b00, "and", and, "bic">; +defm EOR : LogicalImm<0b10, "eor", xor, "eon">; +defm ORR : LogicalImm<0b01, "orr", or, "orn">; + +// FIXME: these aliases *are* canonical sometimes (when movz can't be +// used). Actually, it seems to be working right now, but putting logical_immXX +// here is a bit dodgy on the AsmParser side too. +def : InstAlias<"mov $dst, $imm", (ORRWri GPR32sp:$dst, WZR, + logical_imm32:$imm), 0>; +def : InstAlias<"mov $dst, $imm", (ORRXri GPR64sp:$dst, XZR, + logical_imm64:$imm), 0>; + + +// (register) +defm ANDS : LogicalRegS<0b11, 0, "ands", AArch64and_flag>; +defm BICS : LogicalRegS<0b11, 1, "bics", + BinOpFrag<(AArch64and_flag node:$LHS, (not node:$RHS))>>; +defm AND : LogicalReg<0b00, 0, "and", and>; +defm BIC : LogicalReg<0b00, 1, "bic", + BinOpFrag<(and node:$LHS, (not node:$RHS))>>; +defm EON : LogicalReg<0b10, 1, "eon", + BinOpFrag<(not (xor node:$LHS, node:$RHS))>>; +defm EOR : LogicalReg<0b10, 0, "eor", xor>; +defm ORN : LogicalReg<0b01, 1, "orn", + BinOpFrag<(or node:$LHS, (not node:$RHS))>>; +defm ORR : LogicalReg<0b01, 0, "orr", or>; + +def : InstAlias<"mov $dst, $src", (ORRWrs GPR32:$dst, WZR, GPR32:$src, 0), 2>; +def : InstAlias<"mov $dst, $src", (ORRXrs GPR64:$dst, XZR, GPR64:$src, 0), 2>; + +def : InstAlias<"mvn $Wd, $Wm", (ORNWrs GPR32:$Wd, WZR, GPR32:$Wm, 0), 3>; +def : InstAlias<"mvn $Xd, $Xm", (ORNXrs GPR64:$Xd, XZR, GPR64:$Xm, 0), 3>; + +def : InstAlias<"mvn $Wd, $Wm$sh", + (ORNWrs GPR32:$Wd, WZR, GPR32:$Wm, logical_shift32:$sh), 2>; +def : InstAlias<"mvn $Xd, $Xm$sh", + (ORNXrs GPR64:$Xd, XZR, GPR64:$Xm, logical_shift64:$sh), 2>; + +def : InstAlias<"tst $src1, $src2", + (ANDSWri WZR, GPR32:$src1, logical_imm32:$src2), 2>; +def : InstAlias<"tst $src1, $src2", + (ANDSXri XZR, GPR64:$src1, logical_imm64:$src2), 2>; + +def : InstAlias<"tst $src1, $src2", + (ANDSWrs WZR, GPR32:$src1, GPR32:$src2, 0), 3>; +def : InstAlias<"tst $src1, $src2", + (ANDSXrs XZR, GPR64:$src1, GPR64:$src2, 0), 3>; + +def : InstAlias<"tst $src1, $src2$sh", + (ANDSWrs WZR, GPR32:$src1, GPR32:$src2, logical_shift32:$sh), 2>; +def : InstAlias<"tst $src1, $src2$sh", + (ANDSXrs XZR, GPR64:$src1, GPR64:$src2, logical_shift64:$sh), 2>; + + +def : Pat<(not GPR32:$Wm), (ORNWrr WZR, GPR32:$Wm)>; +def : Pat<(not GPR64:$Xm), (ORNXrr XZR, GPR64:$Xm)>; -} -defm : lsl_aliases<"add", ADDxxx_uxtx, Rxsp, GPR64xsp, GPR64>; -defm : lsl_aliases<"add", ADDxxx_uxtx, GPR64xsp, Rxsp, GPR64>; -defm : lsl_aliases<"add", ADDwww_uxtw, Rwsp, GPR32wsp, GPR32>; -defm : lsl_aliases<"add", ADDwww_uxtw, GPR32wsp, Rwsp, GPR32>; -defm : lsl_aliases<"sub", SUBxxx_uxtx, Rxsp, GPR64xsp, GPR64>; -defm : lsl_aliases<"sub", SUBxxx_uxtx, GPR64xsp, Rxsp, GPR64>; -defm : lsl_aliases<"sub", SUBwww_uxtw, Rwsp, GPR32wsp, GPR32>; -defm : lsl_aliases<"sub", SUBwww_uxtw, GPR32wsp, Rwsp, GPR32>; - -// Rd cannot be sp for flag-setting variants so only half of the aliases are -// needed. -defm : lsl_aliases<"adds", ADDSxxx_uxtx, GPR64, Rxsp, GPR64>; -defm : lsl_aliases<"adds", ADDSwww_uxtw, GPR32, Rwsp, GPR32>; -defm : lsl_aliases<"subs", SUBSxxx_uxtx, GPR64, Rxsp, GPR64>; -defm : lsl_aliases<"subs", SUBSwww_uxtw, GPR32, Rwsp, GPR32>; - -// CMP unfortunately has to be different because the instruction doesn't have a -// dest register. -multiclass cmp_lsl_aliases { - def : InstAlias; - - def : InstAlias; -} +//===----------------------------------------------------------------------===// +// One operand data processing instructions. +//===----------------------------------------------------------------------===// -defm : cmp_lsl_aliases<"cmp", CMPxx_uxtx, Rxsp, GPR64>; -defm : cmp_lsl_aliases<"cmp", CMPww_uxtw, Rwsp, GPR32>; -defm : cmp_lsl_aliases<"cmn", CMNxx_uxtx, Rxsp, GPR64>; -defm : cmp_lsl_aliases<"cmn", CMNww_uxtw, Rwsp, GPR32>; +defm CLS : OneOperandData<0b101, "cls">; +defm CLZ : OneOperandData<0b100, "clz", ctlz>; +defm RBIT : OneOperandData<0b000, "rbit">; + +def : Pat<(int_aarch64_rbit GPR32:$Rn), (RBITWr $Rn)>; +def : Pat<(int_aarch64_rbit GPR64:$Rn), (RBITXr $Rn)>; + +def REV16Wr : OneWRegData<0b001, "rev16", + UnOpFrag<(rotr (bswap node:$LHS), (i64 16))>>; +def REV16Xr : OneXRegData<0b001, "rev16", null_frag>; + +def : Pat<(cttz GPR32:$Rn), + (CLZWr (RBITWr GPR32:$Rn))>; +def : Pat<(cttz GPR64:$Rn), + (CLZXr (RBITXr GPR64:$Rn))>; +def : Pat<(ctlz (or (shl (xor (sra GPR32:$Rn, (i64 31)), GPR32:$Rn), (i64 1)), + (i32 1))), + (CLSWr GPR32:$Rn)>; +def : Pat<(ctlz (or (shl (xor (sra GPR64:$Rn, (i64 63)), GPR64:$Rn), (i64 1)), + (i64 1))), + (CLSXr GPR64:$Rn)>; + +// Unlike the other one operand instructions, the instructions with the "rev" +// mnemonic do *not* just different in the size bit, but actually use different +// opcode bits for the different sizes. +def REVWr : OneWRegData<0b010, "rev", bswap>; +def REVXr : OneXRegData<0b011, "rev", bswap>; +def REV32Xr : OneXRegData<0b010, "rev32", + UnOpFrag<(rotr (bswap node:$LHS), (i64 32))>>; + +def : InstAlias<"rev64 $Rd, $Rn", (REVXr GPR64:$Rd, GPR64:$Rn), 0>; + +// The bswap commutes with the rotr so we want a pattern for both possible +// orders. +def : Pat<(bswap (rotr GPR32:$Rn, (i64 16))), (REV16Wr GPR32:$Rn)>; +def : Pat<(bswap (rotr GPR64:$Rn, (i64 32))), (REV32Xr GPR64:$Rn)>; //===----------------------------------------------------------------------===// -// Add-subtract (immediate) instructions +// Bitfield immediate extraction instruction. //===----------------------------------------------------------------------===// -// Contains: ADD, ADDS, SUB, SUBS + aliases CMN, CMP, MOV - -// These instructions accept a 12-bit unsigned immediate, optionally shifted -// left by 12 bits. Official assembly format specifies a 12 bit immediate with -// one of "", "LSL #0", "LSL #12" supplementary operands. - -// There are surprisingly few ways to make this work with TableGen, so this -// implementation has separate instructions for the "LSL #0" and "LSL #12" -// variants. - -// If the MCInst retained a single combined immediate (which could be 0x123000, -// for example) then both components (imm & shift) would have to be delegated to -// a single assembly operand. This would entail a separate operand parser -// (because the LSL would have to live in the same AArch64Operand as the -// immediate to be accessible); assembly parsing is rather complex and -// error-prone C++ code. -// -// By splitting the immediate, we can delegate handling this optional operand to -// an InstAlias. Supporting functions to generate the correct MCInst are still -// required, but these are essentially trivial and parsing can remain generic. -// -// Rejected plans with rationale: -// ------------------------------ -// -// In an ideal world you'de have two first class immediate operands (in -// InOperandList, specifying imm12 and shift). Unfortunately this is not -// selectable by any means I could discover. -// -// An Instruction with two MCOperands hidden behind a single entry in -// InOperandList (expanded by ComplexPatterns and MIOperandInfo) was functional, -// but required more C++ code to handle encoding/decoding. Parsing (the intended -// main beneficiary) ended up equally complex because of the optional nature of -// "LSL #0". -// -// Attempting to circumvent the need for a custom OperandParser above by giving -// InstAliases without the "lsl #0" failed. add/sub could be accommodated but -// the cmp/cmn aliases didn't use the MIOperandInfo to determine how operands -// should be parsed: there was no way to accommodate an "lsl #12". - -let ParserMethod = "ParseImmWithLSLOperand", - RenderMethod = "addImmWithLSLOperands" in { - // Derived PredicateMethod fields are different for each - def addsubimm_lsl0_asmoperand : AsmOperandClass { - let Name = "AddSubImmLSL0"; - // If an error is reported against this operand, instruction could also be a - // register variant. - let DiagnosticType = "AddSubSecondSource"; - } +let hasSideEffects = 0 in +defm EXTR : ExtractImm<"extr">; +def : InstAlias<"ror $dst, $src, $shift", + (EXTRWrri GPR32:$dst, GPR32:$src, GPR32:$src, imm0_31:$shift)>; +def : InstAlias<"ror $dst, $src, $shift", + (EXTRXrri GPR64:$dst, GPR64:$src, GPR64:$src, imm0_63:$shift)>; + +def : Pat<(rotr GPR32:$Rn, (i64 imm0_31:$imm)), + (EXTRWrri GPR32:$Rn, GPR32:$Rn, imm0_31:$imm)>; +def : Pat<(rotr GPR64:$Rn, (i64 imm0_63:$imm)), + (EXTRXrri GPR64:$Rn, GPR64:$Rn, imm0_63:$imm)>; - def addsubimm_lsl12_asmoperand : AsmOperandClass { - let Name = "AddSubImmLSL12"; - let DiagnosticType = "AddSubSecondSource"; - } +//===----------------------------------------------------------------------===// +// Other bitfield immediate instructions. +//===----------------------------------------------------------------------===// +let hasSideEffects = 0 in { +defm BFM : BitfieldImmWith2RegArgs<0b01, "bfm">; +defm SBFM : BitfieldImm<0b00, "sbfm">; +defm UBFM : BitfieldImm<0b10, "ubfm">; } -def shr_12_XFORM : SDNodeXFormgetTargetConstant(N->getSExtValue() >> 12, MVT::i32); +def i32shift_a : Operand, SDNodeXFormgetZExtValue()) & 0x1f; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); }]>; -def shr_12_neg_XFORM : SDNodeXFormgetTargetConstant((-N->getSExtValue()) >> 12, MVT::i32); +def i32shift_b : Operand, SDNodeXFormgetZExtValue(); + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); }]>; -def neg_XFORM : SDNodeXFormgetTargetConstant(-N->getSExtValue(), MVT::i32); +// min(7, 31 - shift_amt) +def i32shift_sext_i8 : Operand, SDNodeXFormgetZExtValue(); + enc = enc > 7 ? 7 : enc; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); }]>; +// min(15, 31 - shift_amt) +def i32shift_sext_i16 : Operand, SDNodeXFormgetZExtValue(); + enc = enc > 15 ? 15 : enc; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; -multiclass addsub_imm_operands { - let PrintMethod = "printAddSubImmLSL0Operand", - EncoderMethod = "getAddSubImmOpValue", - ParserMatchClass = addsubimm_lsl0_asmoperand in { - def _posimm_lsl0 : Operand, - ImmLeaf= 0 && (Imm & ~0xfff) == 0; }]>; - def _negimm_lsl0 : Operand, - ImmLeaf; - } - - let PrintMethod = "printAddSubImmLSL12Operand", - EncoderMethod = "getAddSubImmOpValue", - ParserMatchClass = addsubimm_lsl12_asmoperand in { - def _posimm_lsl12 : Operand, - ImmLeaf= 0 && (Imm & ~0xfff000) == 0; }], - shr_12_XFORM>; - - def _negimm_lsl12 : Operand, - ImmLeaf; - } -} +def i64shift_a : Operand, SDNodeXFormgetZExtValue()) & 0x3f; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; -// The add operands don't need any transformation -defm addsubimm_operand_i32 : addsub_imm_operands; -defm addsubimm_operand_i64 : addsub_imm_operands; - -multiclass addsubimm_varieties shift, - string asmop, string cmpasmop, - Operand imm_operand, Operand cmp_imm_operand, - RegisterClass GPR, RegisterClass GPRsp, - AArch64Reg ZR> { - // All registers for non-S variants allow SP - def _s : A64I_addsubimm; - - - // S variants can read SP but would write to ZR - def _S : A64I_addsubimm { - let Defs = [NZCV]; - } +def i64shift_b : Operand, SDNodeXFormgetZExtValue(); + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; - // Note that the pattern here for ADDS is subtle. Canonically CMP - // a, b becomes SUBS a, b. If b < 0 then this is equivalent to - // ADDS a, (-b). This is not true in general. - def _cmp : A64I_addsubimm { - let Rd = 0b11111; - let Defs = [NZCV]; - let isCompare = 1; - } -} +// min(7, 63 - shift_amt) +def i64shift_sext_i8 : Operand, SDNodeXFormgetZExtValue(); + enc = enc > 7 ? 7 : enc; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; +// min(15, 63 - shift_amt) +def i64shift_sext_i16 : Operand, SDNodeXFormgetZExtValue(); + enc = enc > 15 ? 15 : enc; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; -multiclass addsubimm_shifts { - defm _lsl0 : addsubimm_varieties(operand # "_lsl0"), - !cast(cmpoperand # "_lsl0"), - GPR, GPRsp, ZR>; - - defm _lsl12 : addsubimm_varieties(operand # "_lsl12"), - !cast(cmpoperand # "_lsl12"), - GPR, GPRsp, ZR>; -} +// min(31, 63 - shift_amt) +def i64shift_sext_i32 : Operand, SDNodeXFormgetZExtValue(); + enc = enc > 31 ? 31 : enc; + return CurDAG->getTargetConstant(enc, SDLoc(N), MVT::i64); +}]>; -defm ADDwwi : addsubimm_shifts<"ADDwi", 0b0, 0b0, "add", "cmn", - "addsubimm_operand_i32_posimm", - "addsubimm_operand_i32_negimm", - GPR32, GPR32wsp, WZR>; -defm ADDxxi : addsubimm_shifts<"ADDxi", 0b1, 0b0, "add", "cmn", - "addsubimm_operand_i64_posimm", - "addsubimm_operand_i64_negimm", - GPR64, GPR64xsp, XZR>; -defm SUBwwi : addsubimm_shifts<"SUBwi", 0b0, 0b1, "sub", "cmp", - "addsubimm_operand_i32_negimm", - "addsubimm_operand_i32_posimm", - GPR32, GPR32wsp, WZR>; -defm SUBxxi : addsubimm_shifts<"SUBxi", 0b1, 0b1, "sub", "cmp", - "addsubimm_operand_i64_negimm", - "addsubimm_operand_i64_posimm", - GPR64, GPR64xsp, XZR>; - -multiclass MOVsp { - def _fromsp : InstAlias<"mov $Rd, $Rn", - (addop GPRsp:$Rd, SP:$Rn, 0), - 0b1>; - - def _tosp : InstAlias<"mov $Rd, $Rn", - (addop SP:$Rd, GPRsp:$Rn, 0), - 0b1>; -} +def : Pat<(shl GPR32:$Rn, (i64 imm0_31:$imm)), + (UBFMWri GPR32:$Rn, (i64 (i32shift_a imm0_31:$imm)), + (i64 (i32shift_b imm0_31:$imm)))>; +def : Pat<(shl GPR64:$Rn, (i64 imm0_63:$imm)), + (UBFMXri GPR64:$Rn, (i64 (i64shift_a imm0_63:$imm)), + (i64 (i64shift_b imm0_63:$imm)))>; -// Recall Rxsp is a RegisterClass containing *just* xsp. -defm MOVxx : MOVsp; -defm MOVww : MOVsp; +let AddedComplexity = 10 in { +def : Pat<(sra GPR32:$Rn, (i64 imm0_31:$imm)), + (SBFMWri GPR32:$Rn, imm0_31:$imm, 31)>; +def : Pat<(sra GPR64:$Rn, (i64 imm0_63:$imm)), + (SBFMXri GPR64:$Rn, imm0_63:$imm, 63)>; +} + +def : InstAlias<"asr $dst, $src, $shift", + (SBFMWri GPR32:$dst, GPR32:$src, imm0_31:$shift, 31)>; +def : InstAlias<"asr $dst, $src, $shift", + (SBFMXri GPR64:$dst, GPR64:$src, imm0_63:$shift, 63)>; +def : InstAlias<"sxtb $dst, $src", (SBFMWri GPR32:$dst, GPR32:$src, 0, 7)>; +def : InstAlias<"sxtb $dst, $src", (SBFMXri GPR64:$dst, GPR64:$src, 0, 7)>; +def : InstAlias<"sxth $dst, $src", (SBFMWri GPR32:$dst, GPR32:$src, 0, 15)>; +def : InstAlias<"sxth $dst, $src", (SBFMXri GPR64:$dst, GPR64:$src, 0, 15)>; +def : InstAlias<"sxtw $dst, $src", (SBFMXri GPR64:$dst, GPR64:$src, 0, 31)>; + +def : Pat<(srl GPR32:$Rn, (i64 imm0_31:$imm)), + (UBFMWri GPR32:$Rn, imm0_31:$imm, 31)>; +def : Pat<(srl GPR64:$Rn, (i64 imm0_63:$imm)), + (UBFMXri GPR64:$Rn, imm0_63:$imm, 63)>; + +def : InstAlias<"lsr $dst, $src, $shift", + (UBFMWri GPR32:$dst, GPR32:$src, imm0_31:$shift, 31)>; +def : InstAlias<"lsr $dst, $src, $shift", + (UBFMXri GPR64:$dst, GPR64:$src, imm0_63:$shift, 63)>; +def : InstAlias<"uxtb $dst, $src", (UBFMWri GPR32:$dst, GPR32:$src, 0, 7)>; +def : InstAlias<"uxtb $dst, $src", (UBFMXri GPR64:$dst, GPR64:$src, 0, 7)>; +def : InstAlias<"uxth $dst, $src", (UBFMWri GPR32:$dst, GPR32:$src, 0, 15)>; +def : InstAlias<"uxth $dst, $src", (UBFMXri GPR64:$dst, GPR64:$src, 0, 15)>; +def : InstAlias<"uxtw $dst, $src", (UBFMXri GPR64:$dst, GPR64:$src, 0, 31)>; //===----------------------------------------------------------------------===// -// Add-subtract (shifted register) instructions +// Conditional comparison instructions. //===----------------------------------------------------------------------===// -// Contains: ADD, ADDS, SUB, SUBS + aliases CMN, CMP, NEG, NEGS - -//===------------------------------- -// 1. The "shifed register" operands. Shared with logical insts. -//===------------------------------- - -multiclass shift_operands { - def _asmoperand_i32 : AsmOperandClass { - let Name = "Shift" # form # "i32"; - let RenderMethod = "addShiftOperands"; - let PredicateMethod = "isShift"; - let DiagnosticType = "AddSubRegShift32"; - } - - // Note that the operand type is intentionally i64 because the DAGCombiner - // puts these into a canonical form. - def _i32 : Operand, ImmLeaf= 0 && Imm <= 31; }]> { - let ParserMatchClass - = !cast(prefix # "_asmoperand_i32"); - let PrintMethod = "printShiftOperand"; - let DecoderMethod = "Decode32BitShiftOperand"; - } - - def _asmoperand_i64 : AsmOperandClass { - let Name = "Shift" # form # "i64"; - let RenderMethod = "addShiftOperands"; - let PredicateMethod = "isShift"; - let DiagnosticType = "AddSubRegShift64"; - } +defm CCMN : CondComparison<0, "ccmn", AArch64ccmn>; +defm CCMP : CondComparison<1, "ccmp", AArch64ccmp>; - def _i64 : Operand, ImmLeaf= 0 && Imm <= 63; }]> { - let ParserMatchClass - = !cast(prefix # "_asmoperand_i64"); - let PrintMethod = "printShiftOperand"; - } -} - -defm lsl_operand : shift_operands<"lsl_operand", "LSL">; -defm lsr_operand : shift_operands<"lsr_operand", "LSR">; -defm asr_operand : shift_operands<"asr_operand", "ASR">; - -// Not used for add/sub, but defined here for completeness. The "logical -// (shifted register)" instructions *do* have an ROR variant. -defm ror_operand : shift_operands<"ror_operand", "ROR">; - -//===------------------------------- -// 2. The basic 3.5-operand ADD/SUB/ADDS/SUBS instructions. -//===------------------------------- - -// N.b. the commutable parameter is just !N. It will be first against the wall -// when the revolution comes. -multiclass addsub_shifts defs> { - let isCommutable = commutable, Defs = defs in { - def _lsl : A64I_addsubshift("lsl_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (shl GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _lsr : A64I_addsubshift("lsr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (srl GPR:$Rm, - !cast("lsr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _asr : A64I_addsubshift("asr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (sra GPR:$Rm, - !cast("asr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - } +//===----------------------------------------------------------------------===// +// Conditional select instructions. +//===----------------------------------------------------------------------===// +defm CSEL : CondSelect<0, 0b00, "csel">; + +def inc : PatFrag<(ops node:$in), (add node:$in, 1)>; +defm CSINC : CondSelectOp<0, 0b01, "csinc", inc>; +defm CSINV : CondSelectOp<1, 0b00, "csinv", not>; +defm CSNEG : CondSelectOp<1, 0b01, "csneg", ineg>; + +def : Pat<(AArch64csinv GPR32:$tval, GPR32:$fval, (i32 imm:$cc), NZCV), + (CSINVWr GPR32:$tval, GPR32:$fval, (i32 imm:$cc))>; +def : Pat<(AArch64csinv GPR64:$tval, GPR64:$fval, (i32 imm:$cc), NZCV), + (CSINVXr GPR64:$tval, GPR64:$fval, (i32 imm:$cc))>; +def : Pat<(AArch64csneg GPR32:$tval, GPR32:$fval, (i32 imm:$cc), NZCV), + (CSNEGWr GPR32:$tval, GPR32:$fval, (i32 imm:$cc))>; +def : Pat<(AArch64csneg GPR64:$tval, GPR64:$fval, (i32 imm:$cc), NZCV), + (CSNEGXr GPR64:$tval, GPR64:$fval, (i32 imm:$cc))>; +def : Pat<(AArch64csinc GPR32:$tval, GPR32:$fval, (i32 imm:$cc), NZCV), + (CSINCWr GPR32:$tval, GPR32:$fval, (i32 imm:$cc))>; +def : Pat<(AArch64csinc GPR64:$tval, GPR64:$fval, (i32 imm:$cc), NZCV), + (CSINCXr GPR64:$tval, GPR64:$fval, (i32 imm:$cc))>; + +def : Pat<(AArch64csel (i32 0), (i32 1), (i32 imm:$cc), NZCV), + (CSINCWr WZR, WZR, (i32 imm:$cc))>; +def : Pat<(AArch64csel (i64 0), (i64 1), (i32 imm:$cc), NZCV), + (CSINCXr XZR, XZR, (i32 imm:$cc))>; +def : Pat<(AArch64csel (i32 0), (i32 -1), (i32 imm:$cc), NZCV), + (CSINVWr WZR, WZR, (i32 imm:$cc))>; +def : Pat<(AArch64csel (i64 0), (i64 -1), (i32 imm:$cc), NZCV), + (CSINVXr XZR, XZR, (i32 imm:$cc))>; + +// The inverse of the condition code from the alias instruction is what is used +// in the aliased instruction. The parser all ready inverts the condition code +// for these aliases. +def : InstAlias<"cset $dst, $cc", + (CSINCWr GPR32:$dst, WZR, WZR, inv_ccode:$cc)>; +def : InstAlias<"cset $dst, $cc", + (CSINCXr GPR64:$dst, XZR, XZR, inv_ccode:$cc)>; + +def : InstAlias<"csetm $dst, $cc", + (CSINVWr GPR32:$dst, WZR, WZR, inv_ccode:$cc)>; +def : InstAlias<"csetm $dst, $cc", + (CSINVXr GPR64:$dst, XZR, XZR, inv_ccode:$cc)>; + +def : InstAlias<"cinc $dst, $src, $cc", + (CSINCWr GPR32:$dst, GPR32:$src, GPR32:$src, inv_ccode:$cc)>; +def : InstAlias<"cinc $dst, $src, $cc", + (CSINCXr GPR64:$dst, GPR64:$src, GPR64:$src, inv_ccode:$cc)>; + +def : InstAlias<"cinv $dst, $src, $cc", + (CSINVWr GPR32:$dst, GPR32:$src, GPR32:$src, inv_ccode:$cc)>; +def : InstAlias<"cinv $dst, $src, $cc", + (CSINVXr GPR64:$dst, GPR64:$src, GPR64:$src, inv_ccode:$cc)>; + +def : InstAlias<"cneg $dst, $src, $cc", + (CSNEGWr GPR32:$dst, GPR32:$src, GPR32:$src, inv_ccode:$cc)>; +def : InstAlias<"cneg $dst, $src, $cc", + (CSNEGXr GPR64:$dst, GPR64:$src, GPR64:$src, inv_ccode:$cc)>; - def _noshift - : InstAlias(prefix # "_lsl") GPR:$Rd, GPR:$Rn, - GPR:$Rm, 0)>; +//===----------------------------------------------------------------------===// +// PC-relative instructions. +//===----------------------------------------------------------------------===// +let isReMaterializable = 1 in { +let hasSideEffects = 0, mayStore = 0, mayLoad = 0 in { +def ADR : ADRI<0, "adr", adrlabel, []>; +} // hasSideEffects = 0 - def : Pat<(opfrag GPR:$Rn, GPR:$Rm), - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; -} +def ADRP : ADRI<1, "adrp", adrplabel, + [(set GPR64:$Xd, (AArch64adrp tglobaladdr:$label))]>; +} // isReMaterializable = 1 -multiclass addsub_sizes defs> { - defm xxx : addsub_shifts; - defm www : addsub_shifts; -} +// page address of a constant pool entry, block address +def : Pat<(AArch64adrp tconstpool:$cp), (ADRP tconstpool:$cp)>; +def : Pat<(AArch64adrp tblockaddress:$cp), (ADRP tblockaddress:$cp)>; +//===----------------------------------------------------------------------===// +// Unconditional branch (register) instructions. +//===----------------------------------------------------------------------===// -defm ADD : addsub_sizes<"ADD", 0b0, 0b0, 0b1, "add", add, []>; -defm SUB : addsub_sizes<"SUB", 0b1, 0b0, 0b0, "sub", sub, []>; +let isReturn = 1, isTerminator = 1, isBarrier = 1 in { +def RET : BranchReg<0b0010, "ret", []>; +def DRPS : SpecialReturn<0b0101, "drps">; +def ERET : SpecialReturn<0b0100, "eret">; +} // isReturn = 1, isTerminator = 1, isBarrier = 1 -defm ADDS : addsub_sizes<"ADDS", 0b0, 0b1, 0b1, "adds", addc, [NZCV]>; -defm SUBS : addsub_sizes<"SUBS", 0b1, 0b1, 0b0, "subs", subc, [NZCV]>; +// Default to the LR register. +def : InstAlias<"ret", (RET LR)>; -//===------------------------------- -// 1. The NEG/NEGS aliases -//===------------------------------- +let isCall = 1, Defs = [LR], Uses = [SP] in { +def BLR : BranchReg<0b0001, "blr", [(AArch64call GPR64:$Rn)]>; +} // isCall -multiclass neg_alias { - def : InstAlias<"neg $Rd, $Rm, $Imm6", - (INST GPR:$Rd, ZR, GPR:$Rm, shift_operand:$Imm6)>; +let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in { +def BR : BranchReg<0b0000, "br", [(brind GPR64:$Rn)]>; +} // isBranch, isTerminator, isBarrier, isIndirectBranch - def : Pat<(sub 0, (shiftop GPR:$Rm, shift_operand:$Imm6)), - (INST ZR, GPR:$Rm, shift_operand:$Imm6)>; +// Create a separate pseudo-instruction for codegen to use so that we don't +// flag lr as used in every function. It'll be restored before the RET by the +// epilogue if it's legitimately used. +def RET_ReallyLR : Pseudo<(outs), (ins), [(AArch64retflag)]> { + let isTerminator = 1; + let isBarrier = 1; + let isReturn = 1; } -defm : neg_alias; -defm : neg_alias; -defm : neg_alias; -def : InstAlias<"neg $Rd, $Rm", (SUBwww_lsl GPR32:$Rd, WZR, GPR32:$Rm, 0)>; -def : Pat<(sub 0, GPR32:$Rm), (SUBwww_lsl WZR, GPR32:$Rm, 0)>; - -defm : neg_alias; -defm : neg_alias; -defm : neg_alias; -def : InstAlias<"neg $Rd, $Rm", (SUBxxx_lsl GPR64:$Rd, XZR, GPR64:$Rm, 0)>; -def : Pat<(sub 0, GPR64:$Rm), (SUBxxx_lsl XZR, GPR64:$Rm, 0)>; - -// NEGS doesn't get any patterns yet: defining multiple outputs means C++ has to -// be involved. -class negs_alias - : InstAlias<"negs $Rd, $Rm, $Imm6", - (INST GPR:$Rd, ZR, GPR:$Rm, shift_operand:$Imm6)>; - -def : negs_alias; -def : negs_alias; -def : negs_alias; -def : InstAlias<"negs $Rd, $Rm", (SUBSwww_lsl GPR32:$Rd, WZR, GPR32:$Rm, 0)>; - -def : negs_alias; -def : negs_alias; -def : negs_alias; -def : InstAlias<"negs $Rd, $Rm", (SUBSxxx_lsl GPR64:$Rd, XZR, GPR64:$Rm, 0)>; - -//===------------------------------- -// 1. The CMP/CMN aliases -//===------------------------------- - -multiclass cmp_shifts { - let isCommutable = commutable, Rd = 0b11111, Defs = [NZCV] in { - def _lsl : A64I_addsubshift("lsl_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rn, $Rm, $Imm6"), - [(set NZCV, (opfrag GPR:$Rn, (shl GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _lsr : A64I_addsubshift("lsr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rn, $Rm, $Imm6"), - [(set NZCV, (opfrag GPR:$Rn, (srl GPR:$Rm, - !cast("lsr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _asr : A64I_addsubshift("asr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rn, $Rm, $Imm6"), - [(set NZCV, (opfrag GPR:$Rn, (sra GPR:$Rm, - !cast("asr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - } - - def _noshift - : InstAlias(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; - - def : Pat<(opfrag GPR:$Rn, GPR:$Rm), - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; +// This is a directive-like pseudo-instruction. The purpose is to insert an +// R_AARCH64_TLSDESC_CALL relocation at the offset of the following instruction +// (which in the usual case is a BLR). +let hasSideEffects = 1 in +def TLSDESCCALL : Pseudo<(outs), (ins i64imm:$sym), []> { + let AsmString = ".tlsdesccall $sym"; } -defm CMPww : cmp_shifts<"CMPww", 0b0, 0b1, 0b0, "cmp", A64cmp, "i32", GPR32>; -defm CMPxx : cmp_shifts<"CMPxx", 0b1, 0b1, 0b0, "cmp", A64cmp, "i64", GPR64>; - -defm CMNww : cmp_shifts<"CMNww", 0b0, 0b0, 0b1, "cmn", A64cmn, "i32", GPR32>; -defm CMNxx : cmp_shifts<"CMNxx", 0b1, 0b0, 0b1, "cmn", A64cmn, "i64", GPR64>; +// FIXME: maybe the scratch register used shouldn't be fixed to X1? +// FIXME: can "hasSideEffects be dropped? +let isCall = 1, Defs = [LR, X0, X1], hasSideEffects = 1, + isCodeGenOnly = 1 in +def TLSDESC_CALLSEQ + : Pseudo<(outs), (ins i64imm:$sym), + [(AArch64tlsdesc_callseq tglobaltlsaddr:$sym)]>; +def : Pat<(AArch64tlsdesc_callseq texternalsym:$sym), + (TLSDESC_CALLSEQ texternalsym:$sym)>; //===----------------------------------------------------------------------===// -// Add-subtract (with carry) instructions +// Conditional branch (immediate) instruction. //===----------------------------------------------------------------------===// -// Contains: ADC, ADCS, SBC, SBCS + aliases NGC, NGCS - -multiclass A64I_addsubcarrySizes { - let Uses = [NZCV] in { - def www : A64I_addsubcarry<0b0, op, s, 0b000000, - (outs GPR32:$Rd), (ins GPR32:$Rn, GPR32:$Rm), - !strconcat(asmop, "\t$Rd, $Rn, $Rm"), - [], NoItinerary>; - - def xxx : A64I_addsubcarry<0b1, op, s, 0b000000, - (outs GPR64:$Rd), (ins GPR64:$Rn, GPR64:$Rm), - !strconcat(asmop, "\t$Rd, $Rn, $Rm"), - [], NoItinerary>; - } -} - -let isCommutable = 1 in { - defm ADC : A64I_addsubcarrySizes<0b0, 0b0, "adc">; -} - -defm SBC : A64I_addsubcarrySizes<0b1, 0b0, "sbc">; +def Bcc : BranchCond; -let Defs = [NZCV] in { - let isCommutable = 1 in { - defm ADCS : A64I_addsubcarrySizes<0b0, 0b1, "adcs">; - } +//===----------------------------------------------------------------------===// +// Compare-and-branch instructions. +//===----------------------------------------------------------------------===// +defm CBZ : CmpBranch<0, "cbz", AArch64cbz>; +defm CBNZ : CmpBranch<1, "cbnz", AArch64cbnz>; - defm SBCS : A64I_addsubcarrySizes<0b1, 0b1, "sbcs">; -} +//===----------------------------------------------------------------------===// +// Test-bit-and-branch instructions. +//===----------------------------------------------------------------------===// +defm TBZ : TestBranch<0, "tbz", AArch64tbz>; +defm TBNZ : TestBranch<1, "tbnz", AArch64tbnz>; -def : InstAlias<"ngc $Rd, $Rm", (SBCwww GPR32:$Rd, WZR, GPR32:$Rm)>; -def : InstAlias<"ngc $Rd, $Rm", (SBCxxx GPR64:$Rd, XZR, GPR64:$Rm)>; -def : InstAlias<"ngcs $Rd, $Rm", (SBCSwww GPR32:$Rd, WZR, GPR32:$Rm)>; -def : InstAlias<"ngcs $Rd, $Rm", (SBCSxxx GPR64:$Rd, XZR, GPR64:$Rm)>; +//===----------------------------------------------------------------------===// +// Unconditional branch (immediate) instructions. +//===----------------------------------------------------------------------===// +let isBranch = 1, isTerminator = 1, isBarrier = 1 in { +def B : BranchImm<0, "b", [(br bb:$addr)]>; +} // isBranch, isTerminator, isBarrier -// Note that adde and sube can form a chain longer than two (e.g. for 256-bit -// addition). So the flag-setting instructions are appropriate. -def : Pat<(adde GPR32:$Rn, GPR32:$Rm), (ADCSwww GPR32:$Rn, GPR32:$Rm)>; -def : Pat<(adde GPR64:$Rn, GPR64:$Rm), (ADCSxxx GPR64:$Rn, GPR64:$Rm)>; -def : Pat<(sube GPR32:$Rn, GPR32:$Rm), (SBCSwww GPR32:$Rn, GPR32:$Rm)>; -def : Pat<(sube GPR64:$Rn, GPR64:$Rm), (SBCSxxx GPR64:$Rn, GPR64:$Rm)>; +let isCall = 1, Defs = [LR], Uses = [SP] in { +def BL : CallImm<1, "bl", [(AArch64call tglobaladdr:$addr)]>; +} // isCall +def : Pat<(AArch64call texternalsym:$func), (BL texternalsym:$func)>; //===----------------------------------------------------------------------===// -// Bitfield +// Exception generation instructions. //===----------------------------------------------------------------------===// -// Contains: SBFM, BFM, UBFM, [SU]XT[BHW], ASR, LSR, LSL, SBFI[ZX], BFI, BFXIL, -// UBFIZ, UBFX - -// Because of the rather complicated nearly-overlapping aliases, the decoding of -// this range of instructions is handled manually. The architectural -// instructions are BFM, SBFM and UBFM but a disassembler should never produce -// these. -// -// In the end, the best option was to use BFM instructions for decoding under -// almost all circumstances, but to create aliasing *Instructions* for each of -// the canonical forms and specify a completely custom decoder which would -// substitute the correct MCInst as needed. -// -// This also simplifies instruction selection, parsing etc because the MCInsts -// have a shape that's closer to their use in code. +def BRK : ExceptionGeneration<0b001, 0b00, "brk">; +def DCPS1 : ExceptionGeneration<0b101, 0b01, "dcps1">; +def DCPS2 : ExceptionGeneration<0b101, 0b10, "dcps2">; +def DCPS3 : ExceptionGeneration<0b101, 0b11, "dcps3">; +def HLT : ExceptionGeneration<0b010, 0b00, "hlt">; +def HVC : ExceptionGeneration<0b000, 0b10, "hvc">; +def SMC : ExceptionGeneration<0b000, 0b11, "smc">; +def SVC : ExceptionGeneration<0b000, 0b01, "svc">; + +// DCPSn defaults to an immediate operand of zero if unspecified. +def : InstAlias<"dcps1", (DCPS1 0)>; +def : InstAlias<"dcps2", (DCPS2 0)>; +def : InstAlias<"dcps3", (DCPS3 0)>; -//===------------------------------- -// 1. The architectural BFM instructions -//===------------------------------- +//===----------------------------------------------------------------------===// +// Load instructions. +//===----------------------------------------------------------------------===// -def uimm5_asmoperand : AsmOperandClass { - let Name = "UImm5"; - let PredicateMethod = "isUImm<5>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm5"; +// Pair (indexed, offset) +defm LDPW : LoadPairOffset<0b00, 0, GPR32, simm7s4, "ldp">; +defm LDPX : LoadPairOffset<0b10, 0, GPR64, simm7s8, "ldp">; +defm LDPS : LoadPairOffset<0b00, 1, FPR32, simm7s4, "ldp">; +defm LDPD : LoadPairOffset<0b01, 1, FPR64, simm7s8, "ldp">; +defm LDPQ : LoadPairOffset<0b10, 1, FPR128, simm7s16, "ldp">; + +defm LDPSW : LoadPairOffset<0b01, 0, GPR64, simm7s4, "ldpsw">; + +// Pair (pre-indexed) +def LDPWpre : LoadPairPreIdx<0b00, 0, GPR32, simm7s4, "ldp">; +def LDPXpre : LoadPairPreIdx<0b10, 0, GPR64, simm7s8, "ldp">; +def LDPSpre : LoadPairPreIdx<0b00, 1, FPR32, simm7s4, "ldp">; +def LDPDpre : LoadPairPreIdx<0b01, 1, FPR64, simm7s8, "ldp">; +def LDPQpre : LoadPairPreIdx<0b10, 1, FPR128, simm7s16, "ldp">; + +def LDPSWpre : LoadPairPreIdx<0b01, 0, GPR64, simm7s4, "ldpsw">; + +// Pair (post-indexed) +def LDPWpost : LoadPairPostIdx<0b00, 0, GPR32, simm7s4, "ldp">; +def LDPXpost : LoadPairPostIdx<0b10, 0, GPR64, simm7s8, "ldp">; +def LDPSpost : LoadPairPostIdx<0b00, 1, FPR32, simm7s4, "ldp">; +def LDPDpost : LoadPairPostIdx<0b01, 1, FPR64, simm7s8, "ldp">; +def LDPQpost : LoadPairPostIdx<0b10, 1, FPR128, simm7s16, "ldp">; + +def LDPSWpost : LoadPairPostIdx<0b01, 0, GPR64, simm7s4, "ldpsw">; + + +// Pair (no allocate) +defm LDNPW : LoadPairNoAlloc<0b00, 0, GPR32, simm7s4, "ldnp">; +defm LDNPX : LoadPairNoAlloc<0b10, 0, GPR64, simm7s8, "ldnp">; +defm LDNPS : LoadPairNoAlloc<0b00, 1, FPR32, simm7s4, "ldnp">; +defm LDNPD : LoadPairNoAlloc<0b01, 1, FPR64, simm7s8, "ldnp">; +defm LDNPQ : LoadPairNoAlloc<0b10, 1, FPR128, simm7s16, "ldnp">; + +//--- +// (register offset) +//--- + +// Integer +defm LDRBB : Load8RO<0b00, 0, 0b01, GPR32, "ldrb", i32, zextloadi8>; +defm LDRHH : Load16RO<0b01, 0, 0b01, GPR32, "ldrh", i32, zextloadi16>; +defm LDRW : Load32RO<0b10, 0, 0b01, GPR32, "ldr", i32, load>; +defm LDRX : Load64RO<0b11, 0, 0b01, GPR64, "ldr", i64, load>; + +// Floating-point +defm LDRB : Load8RO<0b00, 1, 0b01, FPR8, "ldr", untyped, load>; +defm LDRH : Load16RO<0b01, 1, 0b01, FPR16, "ldr", f16, load>; +defm LDRS : Load32RO<0b10, 1, 0b01, FPR32, "ldr", f32, load>; +defm LDRD : Load64RO<0b11, 1, 0b01, FPR64, "ldr", f64, load>; +defm LDRQ : Load128RO<0b00, 1, 0b11, FPR128, "ldr", f128, load>; + +// Load sign-extended half-word +defm LDRSHW : Load16RO<0b01, 0, 0b11, GPR32, "ldrsh", i32, sextloadi16>; +defm LDRSHX : Load16RO<0b01, 0, 0b10, GPR64, "ldrsh", i64, sextloadi16>; + +// Load sign-extended byte +defm LDRSBW : Load8RO<0b00, 0, 0b11, GPR32, "ldrsb", i32, sextloadi8>; +defm LDRSBX : Load8RO<0b00, 0, 0b10, GPR64, "ldrsb", i64, sextloadi8>; + +// Load sign-extended word +defm LDRSW : Load32RO<0b10, 0, 0b10, GPR64, "ldrsw", i64, sextloadi32>; + +// Pre-fetch. +defm PRFM : PrefetchRO<0b11, 0, 0b10, "prfm">; + +// For regular load, we do not have any alignment requirement. +// Thus, it is safe to directly map the vector loads with interesting +// addressing modes. +// FIXME: We could do the same for bitconvert to floating point vectors. +multiclass ScalToVecROLoadPat { + def : Pat<(VecTy (scalar_to_vector (ScalTy + (loadop (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$offset))))), + (INSERT_SUBREG (VecTy (IMPLICIT_DEF)), + (LOADW GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$offset), + sub)>; + + def : Pat<(VecTy (scalar_to_vector (ScalTy + (loadop (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$offset))))), + (INSERT_SUBREG (VecTy (IMPLICIT_DEF)), + (LOADX GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$offset), + sub)>; } -def uimm6_asmoperand : AsmOperandClass { - let Name = "UImm6"; - let PredicateMethod = "isUImm<6>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm6"; -} +let AddedComplexity = 10 in { +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; -def bitfield32_imm : Operand, - ImmLeaf= 0 && Imm < 32; }]> { - let ParserMatchClass = uimm5_asmoperand; +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; - let DecoderMethod = "DecodeBitfield32ImmOperand"; -} +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; -def bitfield64_imm : Operand, - ImmLeaf= 0 && Imm < 64; }]> { - let ParserMatchClass = uimm6_asmoperand; +defm : ScalToVecROLoadPat; +defm : ScalToVecROLoadPat; - // Default decoder works in 64-bit case: the 6-bit field can take any value. -} +defm : ScalToVecROLoadPat; -multiclass A64I_bitfieldSizes opc, string asmop> { - def wwii : A64I_bitfield<0b0, opc, 0b0, (outs GPR32:$Rd), - (ins GPR32:$Rn, bitfield32_imm:$ImmR, bitfield32_imm:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [], NoItinerary> { - let DecoderMethod = "DecodeBitfieldInstruction"; - } +defm : ScalToVecROLoadPat; - def xxii : A64I_bitfield<0b1, opc, 0b1, (outs GPR64:$Rd), - (ins GPR64:$Rn, bitfield64_imm:$ImmR, bitfield64_imm:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [], NoItinerary> { - let DecoderMethod = "DecodeBitfieldInstruction"; - } -} -defm SBFM : A64I_bitfieldSizes<0b00, "sbfm">; -defm UBFM : A64I_bitfieldSizes<0b10, "ubfm">; - -// BFM instructions modify the destination register rather than defining it -// completely. -def BFMwwii : - A64I_bitfield<0b0, 0b01, 0b0, (outs GPR32:$Rd), - (ins GPR32:$src, GPR32:$Rn, bitfield32_imm:$ImmR, bitfield32_imm:$ImmS), - "bfm\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - let DecoderMethod = "DecodeBitfieldInstruction"; - let Constraints = "$src = $Rd"; -} +def : Pat <(v1i64 (scalar_to_vector (i64 + (load (ro_Windexed64 GPR64sp:$Rn, GPR32:$Rm, + ro_Wextend64:$extend))))), + (LDRDroW GPR64sp:$Rn, GPR32:$Rm, ro_Wextend64:$extend)>; -def BFMxxii : - A64I_bitfield<0b1, 0b01, 0b1, (outs GPR64:$Rd), - (ins GPR64:$src, GPR64:$Rn, bitfield64_imm:$ImmR, bitfield64_imm:$ImmS), - "bfm\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - let DecoderMethod = "DecodeBitfieldInstruction"; - let Constraints = "$src = $Rd"; +def : Pat <(v1i64 (scalar_to_vector (i64 + (load (ro_Xindexed64 GPR64sp:$Rn, GPR64:$Rm, + ro_Xextend64:$extend))))), + (LDRDroX GPR64sp:$Rn, GPR64:$Rm, ro_Xextend64:$extend)>; } +// Match all load 64 bits width whose type is compatible with FPR64 +multiclass VecROLoadPat { -//===------------------------------- -// 2. Extend aliases to 64-bit dest -//===------------------------------- - -// Unfortunately the extensions that end up as 64-bits cannot be handled by an -// instruction alias: their syntax is (for example) "SXTB x0, w0", which needs -// to be mapped to "SBFM x0, x0, #0, 7" (changing the class of Rn). InstAlias is -// not capable of such a map as far as I'm aware - -// Note that these instructions are strictly more specific than the -// BFM ones (in ImmR) so they can handle their own decoding. -class A64I_bf_ext opc, RegisterClass GPRDest, string asmop, - bits<6> imms, dag pattern> - : A64I_bitfield { - let ImmR = 0b000000; - let ImmS = imms; -} + def : Pat<(VecTy (load (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend))), + (LOADW GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)>; -// Signed extensions -def SXTBxw : A64I_bf_ext<0b1, 0b00, GPR64, "sxtb", 7, - (sext_inreg (anyext GPR32:$Rn), i8)>; -def SXTBww : A64I_bf_ext<0b0, 0b00, GPR32, "sxtb", 7, - (sext_inreg GPR32:$Rn, i8)>; -def SXTHxw : A64I_bf_ext<0b1, 0b00, GPR64, "sxth", 15, - (sext_inreg (anyext GPR32:$Rn), i16)>; -def SXTHww : A64I_bf_ext<0b0, 0b00, GPR32, "sxth", 15, - (sext_inreg GPR32:$Rn, i16)>; -def SXTWxw : A64I_bf_ext<0b1, 0b00, GPR64, "sxtw", 31, (sext GPR32:$Rn)>; - -// Unsigned extensions -def UXTBww : A64I_bf_ext<0b0, 0b10, GPR32, "uxtb", 7, - (and GPR32:$Rn, 255)>; -def UXTHww : A64I_bf_ext<0b0, 0b10, GPR32, "uxth", 15, - (and GPR32:$Rn, 65535)>; - -// The 64-bit unsigned variants are not strictly architectural but recommended -// for consistency. -let isAsmParserOnly = 1 in { - def UXTBxw : A64I_bf_ext<0b0, 0b10, GPR64, "uxtb", 7, - (and (anyext GPR32:$Rn), 255)>; - def UXTHxw : A64I_bf_ext<0b0, 0b10, GPR64, "uxth", 15, - (and (anyext GPR32:$Rn), 65535)>; + def : Pat<(VecTy (load (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend))), + (LOADX GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)>; } -// Extra patterns for when the source register is actually 64-bits -// too. There's no architectural difference here, it's just LLVM -// shinanigans. There's no need for equivalent zero-extension patterns -// because they'll already be caught by logical (immediate) matching. -def : Pat<(sext_inreg GPR64:$Rn, i8), - (SXTBxw (EXTRACT_SUBREG GPR64:$Rn, sub_32))>; -def : Pat<(sext_inreg GPR64:$Rn, i16), - (SXTHxw (EXTRACT_SUBREG GPR64:$Rn, sub_32))>; -def : Pat<(sext_inreg GPR64:$Rn, i32), - (SXTWxw (EXTRACT_SUBREG GPR64:$Rn, sub_32))>; - - -//===------------------------------- -// 3. Aliases for ASR and LSR (the simple shifts) -//===------------------------------- - -// These also handle their own decoding because ImmS being set makes -// them take precedence over BFM. -multiclass A64I_shift opc, string asmop, SDNode opnode> { - def wwi : A64I_bitfield<0b0, opc, 0b0, - (outs GPR32:$Rd), (ins GPR32:$Rn, bitfield32_imm:$ImmR), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR"), - [(set GPR32:$Rd, (opnode GPR32:$Rn, bitfield32_imm:$ImmR))], - NoItinerary> { - let ImmS = 31; - } - - def xxi : A64I_bitfield<0b1, opc, 0b1, - (outs GPR64:$Rd), (ins GPR64:$Rn, bitfield64_imm:$ImmR), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR"), - [(set GPR64:$Rd, (opnode GPR64:$Rn, bitfield64_imm:$ImmR))], - NoItinerary> { - let ImmS = 63; - } - +let AddedComplexity = 10 in { +let Predicates = [IsLE] in { + // We must do vector loads with LD1 in big-endian. + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; +} + +defm : VecROLoadPat; +defm : VecROLoadPat; + +// Match all load 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + // We must do vector loads with LD1 in big-endian. + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; + defm : VecROLoadPat; +} +} // AddedComplexity = 10 + +// zextload -> i64 +multiclass ExtLoadTo64ROPat { + def : Pat<(i64 (loadop (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend))), + (SUBREG_TO_REG (i64 0), + (INSTW GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend), + sub_32)>; + + def : Pat<(i64 (loadop (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend))), + (SUBREG_TO_REG (i64 0), + (INSTX GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend), + sub_32)>; } -defm ASR : A64I_shift<0b00, "asr", sra>; -defm LSR : A64I_shift<0b10, "lsr", srl>; - -//===------------------------------- -// 4. Aliases for LSL -//===------------------------------- +let AddedComplexity = 10 in { + defm : ExtLoadTo64ROPat; + defm : ExtLoadTo64ROPat; + defm : ExtLoadTo64ROPat; -// Unfortunately LSL and subsequent aliases are much more complicated. We need -// to be able to say certain output instruction fields depend in a complex -// manner on combinations of input assembly fields). -// -// MIOperandInfo *might* have been able to do it, but at the cost of -// significantly more C++ code. - -// N.b. contrary to usual practice these operands store the shift rather than -// the machine bits in an MCInst. The complexity overhead of consistency -// outweighed the benefits in this case (custom asmparser, printer and selection -// vs custom encoder). -def bitfield32_lsl_imm : Operand, - ImmLeaf= 0 && Imm <= 31; }]> { - let ParserMatchClass = uimm5_asmoperand; - let EncoderMethod = "getBitfield32LSLOpValue"; -} + // zextloadi1 -> zextloadi8 + defm : ExtLoadTo64ROPat; -def bitfield64_lsl_imm : Operand, - ImmLeaf= 0 && Imm <= 63; }]> { - let ParserMatchClass = uimm6_asmoperand; - let EncoderMethod = "getBitfield64LSLOpValue"; -} + // extload -> zextload + defm : ExtLoadTo64ROPat; + defm : ExtLoadTo64ROPat; + defm : ExtLoadTo64ROPat; -class A64I_bitfield_lsl - : A64I_bitfield { - bits<12> FullImm; - let ImmR = FullImm{5-0}; - let ImmS = FullImm{11-6}; - - // No disassembler allowed because it would overlap with BFM which does the - // actual work. - let isAsmParserOnly = 1; + // extloadi1 -> zextloadi8 + defm : ExtLoadTo64ROPat; } -def LSLwwi : A64I_bitfield_lsl<0b0, GPR32, bitfield32_lsl_imm>; -def LSLxxi : A64I_bitfield_lsl<0b1, GPR64, bitfield64_lsl_imm>; -//===------------------------------- -// 5. Aliases for bitfield extract instructions -//===------------------------------- +// zextload -> i64 +multiclass ExtLoadTo32ROPat { + def : Pat<(i32 (loadop (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend))), + (INSTW GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)>; -def bfx32_width_asmoperand : AsmOperandClass { - let Name = "BFX32Width"; - let PredicateMethod = "isBitfieldWidth<32>"; - let RenderMethod = "addBFXWidthOperands"; - let DiagnosticType = "Width32"; -} - -def bfx32_width : Operand, ImmLeaf { - let PrintMethod = "printBFXWidthOperand"; - let ParserMatchClass = bfx32_width_asmoperand; -} + def : Pat<(i32 (loadop (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend))), + (INSTX GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)>; -def bfx64_width_asmoperand : AsmOperandClass { - let Name = "BFX64Width"; - let PredicateMethod = "isBitfieldWidth<64>"; - let RenderMethod = "addBFXWidthOperands"; - let DiagnosticType = "Width64"; } -def bfx64_width : Operand { - let PrintMethod = "printBFXWidthOperand"; - let ParserMatchClass = bfx64_width_asmoperand; +let AddedComplexity = 10 in { + // extload -> zextload + defm : ExtLoadTo32ROPat; + defm : ExtLoadTo32ROPat; + defm : ExtLoadTo32ROPat; + + // zextloadi1 -> zextloadi8 + defm : ExtLoadTo32ROPat; +} + +//--- +// (unsigned immediate) +//--- +defm LDRX : LoadUI<0b11, 0, 0b01, GPR64, uimm12s8, "ldr", + [(set GPR64:$Rt, + (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)))]>; +defm LDRW : LoadUI<0b10, 0, 0b01, GPR32, uimm12s4, "ldr", + [(set GPR32:$Rt, + (load (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset)))]>; +defm LDRB : LoadUI<0b00, 1, 0b01, FPR8, uimm12s1, "ldr", + [(set FPR8:$Rt, + (load (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset)))]>; +defm LDRH : LoadUI<0b01, 1, 0b01, FPR16, uimm12s2, "ldr", + [(set (f16 FPR16:$Rt), + (load (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset)))]>; +defm LDRS : LoadUI<0b10, 1, 0b01, FPR32, uimm12s4, "ldr", + [(set (f32 FPR32:$Rt), + (load (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset)))]>; +defm LDRD : LoadUI<0b11, 1, 0b01, FPR64, uimm12s8, "ldr", + [(set (f64 FPR64:$Rt), + (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)))]>; +defm LDRQ : LoadUI<0b00, 1, 0b11, FPR128, uimm12s16, "ldr", + [(set (f128 FPR128:$Rt), + (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)))]>; + +// For regular load, we do not have any alignment requirement. +// Thus, it is safe to directly map the vector loads with interesting +// addressing modes. +// FIXME: We could do the same for bitconvert to floating point vectors. +def : Pat <(v8i8 (scalar_to_vector (i32 + (extloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))))), + (INSERT_SUBREG (v8i8 (IMPLICIT_DEF)), + (LDRBui GPR64sp:$Rn, uimm12s1:$offset), bsub)>; +def : Pat <(v16i8 (scalar_to_vector (i32 + (extloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))))), + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (LDRBui GPR64sp:$Rn, uimm12s1:$offset), bsub)>; +def : Pat <(v4i16 (scalar_to_vector (i32 + (extloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))))), + (INSERT_SUBREG (v4i16 (IMPLICIT_DEF)), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset), hsub)>; +def : Pat <(v8i16 (scalar_to_vector (i32 + (extloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))))), + (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset), hsub)>; +def : Pat <(v2i32 (scalar_to_vector (i32 + (load (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))))), + (INSERT_SUBREG (v2i32 (IMPLICIT_DEF)), + (LDRSui GPR64sp:$Rn, uimm12s4:$offset), ssub)>; +def : Pat <(v4i32 (scalar_to_vector (i32 + (load (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))))), + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), + (LDRSui GPR64sp:$Rn, uimm12s4:$offset), ssub)>; +def : Pat <(v1i64 (scalar_to_vector (i64 + (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; +def : Pat <(v2i64 (scalar_to_vector (i64 + (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))))), + (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset), dsub)>; + +// Match all load 64 bits width whose type is compatible with FPR64 +let Predicates = [IsLE] in { + // We must use LD1 to perform vector loads in big-endian. + def : Pat<(v2f32 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(v8i8 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(v4i16 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(v2i32 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(v4f16 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; +} +def : Pat<(v1f64 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; +def : Pat<(v1i64 (load (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))), + (LDRDui GPR64sp:$Rn, uimm12s8:$offset)>; + +// Match all load 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + // We must use LD1 to perform vector loads in big-endian. + def : Pat<(v4f32 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v2f64 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v16i8 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v8i16 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v4i32 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v2i64 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(v8f16 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; +} +def : Pat<(f128 (load (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset))), + (LDRQui GPR64sp:$Rn, uimm12s16:$offset)>; + +defm LDRHH : LoadUI<0b01, 0, 0b01, GPR32, uimm12s2, "ldrh", + [(set GPR32:$Rt, + (zextloadi16 (am_indexed16 GPR64sp:$Rn, + uimm12s2:$offset)))]>; +defm LDRBB : LoadUI<0b00, 0, 0b01, GPR32, uimm12s1, "ldrb", + [(set GPR32:$Rt, + (zextloadi8 (am_indexed8 GPR64sp:$Rn, + uimm12s1:$offset)))]>; +// zextload -> i64 +def : Pat<(i64 (zextloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (SUBREG_TO_REG (i64 0), (LDRBBui GPR64sp:$Rn, uimm12s1:$offset), sub_32)>; +def : Pat<(i64 (zextloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))), + (SUBREG_TO_REG (i64 0), (LDRHHui GPR64sp:$Rn, uimm12s2:$offset), sub_32)>; + +// zextloadi1 -> zextloadi8 +def : Pat<(i32 (zextloadi1 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (LDRBBui GPR64sp:$Rn, uimm12s1:$offset)>; +def : Pat<(i64 (zextloadi1 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (SUBREG_TO_REG (i64 0), (LDRBBui GPR64sp:$Rn, uimm12s1:$offset), sub_32)>; + +// extload -> zextload +def : Pat<(i32 (extloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))), + (LDRHHui GPR64sp:$Rn, uimm12s2:$offset)>; +def : Pat<(i32 (extloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (LDRBBui GPR64sp:$Rn, uimm12s1:$offset)>; +def : Pat<(i32 (extloadi1 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (LDRBBui GPR64sp:$Rn, uimm12s1:$offset)>; +def : Pat<(i64 (extloadi32 (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))), + (SUBREG_TO_REG (i64 0), (LDRWui GPR64sp:$Rn, uimm12s4:$offset), sub_32)>; +def : Pat<(i64 (extloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))), + (SUBREG_TO_REG (i64 0), (LDRHHui GPR64sp:$Rn, uimm12s2:$offset), sub_32)>; +def : Pat<(i64 (extloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (SUBREG_TO_REG (i64 0), (LDRBBui GPR64sp:$Rn, uimm12s1:$offset), sub_32)>; +def : Pat<(i64 (extloadi1 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))), + (SUBREG_TO_REG (i64 0), (LDRBBui GPR64sp:$Rn, uimm12s1:$offset), sub_32)>; + +// load sign-extended half-word +defm LDRSHW : LoadUI<0b01, 0, 0b11, GPR32, uimm12s2, "ldrsh", + [(set GPR32:$Rt, + (sextloadi16 (am_indexed16 GPR64sp:$Rn, + uimm12s2:$offset)))]>; +defm LDRSHX : LoadUI<0b01, 0, 0b10, GPR64, uimm12s2, "ldrsh", + [(set GPR64:$Rt, + (sextloadi16 (am_indexed16 GPR64sp:$Rn, + uimm12s2:$offset)))]>; + +// load sign-extended byte +defm LDRSBW : LoadUI<0b00, 0, 0b11, GPR32, uimm12s1, "ldrsb", + [(set GPR32:$Rt, + (sextloadi8 (am_indexed8 GPR64sp:$Rn, + uimm12s1:$offset)))]>; +defm LDRSBX : LoadUI<0b00, 0, 0b10, GPR64, uimm12s1, "ldrsb", + [(set GPR64:$Rt, + (sextloadi8 (am_indexed8 GPR64sp:$Rn, + uimm12s1:$offset)))]>; + +// load sign-extended word +defm LDRSW : LoadUI<0b10, 0, 0b10, GPR64, uimm12s4, "ldrsw", + [(set GPR64:$Rt, + (sextloadi32 (am_indexed32 GPR64sp:$Rn, + uimm12s4:$offset)))]>; + +// load zero-extended word +def : Pat<(i64 (zextloadi32 (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))), + (SUBREG_TO_REG (i64 0), (LDRWui GPR64sp:$Rn, uimm12s4:$offset), sub_32)>; + +// Pre-fetch. +def PRFMui : PrefetchUI<0b11, 0, 0b10, "prfm", + [(AArch64Prefetch imm:$Rt, + (am_indexed64 GPR64sp:$Rn, + uimm12s8:$offset))]>; + +def : InstAlias<"prfm $Rt, [$Rn]", (PRFMui prfop:$Rt, GPR64sp:$Rn, 0)>; + +//--- +// (literal) +def LDRWl : LoadLiteral<0b00, 0, GPR32, "ldr">; +def LDRXl : LoadLiteral<0b01, 0, GPR64, "ldr">; +def LDRSl : LoadLiteral<0b00, 1, FPR32, "ldr">; +def LDRDl : LoadLiteral<0b01, 1, FPR64, "ldr">; +def LDRQl : LoadLiteral<0b10, 1, FPR128, "ldr">; + +// load sign-extended word +def LDRSWl : LoadLiteral<0b10, 0, GPR64, "ldrsw">; + +// prefetch +def PRFMl : PrefetchLiteral<0b11, 0, "prfm", []>; +// [(AArch64Prefetch imm:$Rt, tglobaladdr:$label)]>; + +//--- +// (unscaled immediate) +defm LDURX : LoadUnscaled<0b11, 0, 0b01, GPR64, "ldur", + [(set GPR64:$Rt, + (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURW : LoadUnscaled<0b10, 0, 0b01, GPR32, "ldur", + [(set GPR32:$Rt, + (load (am_unscaled32 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURB : LoadUnscaled<0b00, 1, 0b01, FPR8, "ldur", + [(set FPR8:$Rt, + (load (am_unscaled8 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURH : LoadUnscaled<0b01, 1, 0b01, FPR16, "ldur", + [(set FPR16:$Rt, + (load (am_unscaled16 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURS : LoadUnscaled<0b10, 1, 0b01, FPR32, "ldur", + [(set (f32 FPR32:$Rt), + (load (am_unscaled32 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURD : LoadUnscaled<0b11, 1, 0b01, FPR64, "ldur", + [(set (f64 FPR64:$Rt), + (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURQ : LoadUnscaled<0b00, 1, 0b11, FPR128, "ldur", + [(set (f128 FPR128:$Rt), + (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset)))]>; + +defm LDURHH + : LoadUnscaled<0b01, 0, 0b01, GPR32, "ldurh", + [(set GPR32:$Rt, + (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURBB + : LoadUnscaled<0b00, 0, 0b01, GPR32, "ldurb", + [(set GPR32:$Rt, + (zextloadi8 (am_unscaled16 GPR64sp:$Rn, simm9:$offset)))]>; + +// Match all load 64 bits width whose type is compatible with FPR64 +let Predicates = [IsLE] in { + def : Pat<(v2f32 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v2i32 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v4i16 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v8i8 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v4f16 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; +} +def : Pat<(v1f64 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(v1i64 (load (am_unscaled64 GPR64sp:$Rn, simm9:$offset))), + (LDURDi GPR64sp:$Rn, simm9:$offset)>; + +// Match all load 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + def : Pat<(v2f64 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v2i64 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v4f32 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v4i32 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v8i16 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v16i8 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(v8f16 (load (am_unscaled128 GPR64sp:$Rn, simm9:$offset))), + (LDURQi GPR64sp:$Rn, simm9:$offset)>; +} + +// anyext -> zext +def : Pat<(i32 (extloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))), + (LDURHHi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i32 (extloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (LDURBBi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i32 (extloadi1 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (LDURBBi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i64 (extloadi32 (am_unscaled32 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURWi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (extloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURHHi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (extloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURBBi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (extloadi1 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURBBi GPR64sp:$Rn, simm9:$offset), sub_32)>; +// unscaled zext +def : Pat<(i32 (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))), + (LDURHHi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i32 (zextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (LDURBBi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i32 (zextloadi1 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (LDURBBi GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(i64 (zextloadi32 (am_unscaled32 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURWi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURHHi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (zextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURBBi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (zextloadi1 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURBBi GPR64sp:$Rn, simm9:$offset), sub_32)>; + + +//--- +// LDR mnemonics fall back to LDUR for negative or unaligned offsets. + +// Define new assembler match classes as we want to only match these when +// the don't otherwise match the scaled addressing mode for LDR/STR. Don't +// associate a DiagnosticType either, as we want the diagnostic for the +// canonical form (the scaled operand) to take precedence. +class SImm9OffsetOperand : AsmOperandClass { + let Name = "SImm9OffsetFB" # Width; + let PredicateMethod = "isSImm9OffsetFB<" # Width # ">"; + let RenderMethod = "addImmOperands"; } +def SImm9OffsetFB8Operand : SImm9OffsetOperand<8>; +def SImm9OffsetFB16Operand : SImm9OffsetOperand<16>; +def SImm9OffsetFB32Operand : SImm9OffsetOperand<32>; +def SImm9OffsetFB64Operand : SImm9OffsetOperand<64>; +def SImm9OffsetFB128Operand : SImm9OffsetOperand<128>; + +def simm9_offset_fb8 : Operand { + let ParserMatchClass = SImm9OffsetFB8Operand; +} +def simm9_offset_fb16 : Operand { + let ParserMatchClass = SImm9OffsetFB16Operand; +} +def simm9_offset_fb32 : Operand { + let ParserMatchClass = SImm9OffsetFB32Operand; +} +def simm9_offset_fb64 : Operand { + let ParserMatchClass = SImm9OffsetFB64Operand; +} +def simm9_offset_fb128 : Operand { + let ParserMatchClass = SImm9OffsetFB128Operand; +} + +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURXi GPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb64:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURWi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb32:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURBi FPR8:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURHi FPR16:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURSi FPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb32:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURDi FPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb64:$offset), 0>; +def : InstAlias<"ldr $Rt, [$Rn, $offset]", + (LDURQi FPR128:$Rt, GPR64sp:$Rn, simm9_offset_fb128:$offset), 0>; + +// zextload -> i64 +def : Pat<(i64 (zextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURBBi GPR64sp:$Rn, simm9:$offset), sub_32)>; +def : Pat<(i64 (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))), + (SUBREG_TO_REG (i64 0), (LDURHHi GPR64sp:$Rn, simm9:$offset), sub_32)>; + +// load sign-extended half-word +defm LDURSHW + : LoadUnscaled<0b01, 0, 0b11, GPR32, "ldursh", + [(set GPR32:$Rt, + (sextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURSHX + : LoadUnscaled<0b01, 0, 0b10, GPR64, "ldursh", + [(set GPR64:$Rt, + (sextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset)))]>; + +// load sign-extended byte +defm LDURSBW + : LoadUnscaled<0b00, 0, 0b11, GPR32, "ldursb", + [(set GPR32:$Rt, + (sextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset)))]>; +defm LDURSBX + : LoadUnscaled<0b00, 0, 0b10, GPR64, "ldursb", + [(set GPR64:$Rt, + (sextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset)))]>; + +// load sign-extended word +defm LDURSW + : LoadUnscaled<0b10, 0, 0b10, GPR64, "ldursw", + [(set GPR64:$Rt, + (sextloadi32 (am_unscaled32 GPR64sp:$Rn, simm9:$offset)))]>; + +// zero and sign extending aliases from generic LDR* mnemonics to LDUR*. +def : InstAlias<"ldrb $Rt, [$Rn, $offset]", + (LDURBBi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"ldrh $Rt, [$Rn, $offset]", + (LDURHHi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; +def : InstAlias<"ldrsb $Rt, [$Rn, $offset]", + (LDURSBWi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"ldrsb $Rt, [$Rn, $offset]", + (LDURSBXi GPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"ldrsh $Rt, [$Rn, $offset]", + (LDURSHWi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; +def : InstAlias<"ldrsh $Rt, [$Rn, $offset]", + (LDURSHXi GPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; +def : InstAlias<"ldrsw $Rt, [$Rn, $offset]", + (LDURSWi GPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb32:$offset), 0>; + +// Pre-fetch. +defm PRFUM : PrefetchUnscaled<0b11, 0, 0b10, "prfum", + [(AArch64Prefetch imm:$Rt, + (am_unscaled64 GPR64sp:$Rn, simm9:$offset))]>; + +//--- +// (unscaled immediate, unprivileged) +defm LDTRX : LoadUnprivileged<0b11, 0, 0b01, GPR64, "ldtr">; +defm LDTRW : LoadUnprivileged<0b10, 0, 0b01, GPR32, "ldtr">; + +defm LDTRH : LoadUnprivileged<0b01, 0, 0b01, GPR32, "ldtrh">; +defm LDTRB : LoadUnprivileged<0b00, 0, 0b01, GPR32, "ldtrb">; + +// load sign-extended half-word +defm LDTRSHW : LoadUnprivileged<0b01, 0, 0b11, GPR32, "ldtrsh">; +defm LDTRSHX : LoadUnprivileged<0b01, 0, 0b10, GPR64, "ldtrsh">; + +// load sign-extended byte +defm LDTRSBW : LoadUnprivileged<0b00, 0, 0b11, GPR32, "ldtrsb">; +defm LDTRSBX : LoadUnprivileged<0b00, 0, 0b10, GPR64, "ldtrsb">; + +// load sign-extended word +defm LDTRSW : LoadUnprivileged<0b10, 0, 0b10, GPR64, "ldtrsw">; + +//--- +// (immediate pre-indexed) +def LDRWpre : LoadPreIdx<0b10, 0, 0b01, GPR32, "ldr">; +def LDRXpre : LoadPreIdx<0b11, 0, 0b01, GPR64, "ldr">; +def LDRBpre : LoadPreIdx<0b00, 1, 0b01, FPR8, "ldr">; +def LDRHpre : LoadPreIdx<0b01, 1, 0b01, FPR16, "ldr">; +def LDRSpre : LoadPreIdx<0b10, 1, 0b01, FPR32, "ldr">; +def LDRDpre : LoadPreIdx<0b11, 1, 0b01, FPR64, "ldr">; +def LDRQpre : LoadPreIdx<0b00, 1, 0b11, FPR128, "ldr">; + +// load sign-extended half-word +def LDRSHWpre : LoadPreIdx<0b01, 0, 0b11, GPR32, "ldrsh">; +def LDRSHXpre : LoadPreIdx<0b01, 0, 0b10, GPR64, "ldrsh">; + +// load sign-extended byte +def LDRSBWpre : LoadPreIdx<0b00, 0, 0b11, GPR32, "ldrsb">; +def LDRSBXpre : LoadPreIdx<0b00, 0, 0b10, GPR64, "ldrsb">; + +// load zero-extended byte +def LDRBBpre : LoadPreIdx<0b00, 0, 0b01, GPR32, "ldrb">; +def LDRHHpre : LoadPreIdx<0b01, 0, 0b01, GPR32, "ldrh">; + +// load sign-extended word +def LDRSWpre : LoadPreIdx<0b10, 0, 0b10, GPR64, "ldrsw">; + +//--- +// (immediate post-indexed) +def LDRWpost : LoadPostIdx<0b10, 0, 0b01, GPR32, "ldr">; +def LDRXpost : LoadPostIdx<0b11, 0, 0b01, GPR64, "ldr">; +def LDRBpost : LoadPostIdx<0b00, 1, 0b01, FPR8, "ldr">; +def LDRHpost : LoadPostIdx<0b01, 1, 0b01, FPR16, "ldr">; +def LDRSpost : LoadPostIdx<0b10, 1, 0b01, FPR32, "ldr">; +def LDRDpost : LoadPostIdx<0b11, 1, 0b01, FPR64, "ldr">; +def LDRQpost : LoadPostIdx<0b00, 1, 0b11, FPR128, "ldr">; + +// load sign-extended half-word +def LDRSHWpost : LoadPostIdx<0b01, 0, 0b11, GPR32, "ldrsh">; +def LDRSHXpost : LoadPostIdx<0b01, 0, 0b10, GPR64, "ldrsh">; + +// load sign-extended byte +def LDRSBWpost : LoadPostIdx<0b00, 0, 0b11, GPR32, "ldrsb">; +def LDRSBXpost : LoadPostIdx<0b00, 0, 0b10, GPR64, "ldrsb">; + +// load zero-extended byte +def LDRBBpost : LoadPostIdx<0b00, 0, 0b01, GPR32, "ldrb">; +def LDRHHpost : LoadPostIdx<0b01, 0, 0b01, GPR32, "ldrh">; + +// load sign-extended word +def LDRSWpost : LoadPostIdx<0b10, 0, 0b10, GPR64, "ldrsw">; -multiclass A64I_bitfield_extract opc, string asmop, SDNode op> { - def wwii : A64I_bitfield<0b0, opc, 0b0, (outs GPR32:$Rd), - (ins GPR32:$Rn, bitfield32_imm:$ImmR, bfx32_width:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [(set GPR32:$Rd, (op GPR32:$Rn, imm:$ImmR, imm:$ImmS))], - NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - } +//===----------------------------------------------------------------------===// +// Store instructions. +//===----------------------------------------------------------------------===// - def xxii : A64I_bitfield<0b1, opc, 0b1, (outs GPR64:$Rd), - (ins GPR64:$Rn, bitfield64_imm:$ImmR, bfx64_width:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [(set GPR64:$Rd, (op GPR64:$Rn, imm:$ImmR, imm:$ImmS))], - NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - } +// Pair (indexed, offset) +// FIXME: Use dedicated range-checked addressing mode operand here. +defm STPW : StorePairOffset<0b00, 0, GPR32, simm7s4, "stp">; +defm STPX : StorePairOffset<0b10, 0, GPR64, simm7s8, "stp">; +defm STPS : StorePairOffset<0b00, 1, FPR32, simm7s4, "stp">; +defm STPD : StorePairOffset<0b01, 1, FPR64, simm7s8, "stp">; +defm STPQ : StorePairOffset<0b10, 1, FPR128, simm7s16, "stp">; + +// Pair (pre-indexed) +def STPWpre : StorePairPreIdx<0b00, 0, GPR32, simm7s4, "stp">; +def STPXpre : StorePairPreIdx<0b10, 0, GPR64, simm7s8, "stp">; +def STPSpre : StorePairPreIdx<0b00, 1, FPR32, simm7s4, "stp">; +def STPDpre : StorePairPreIdx<0b01, 1, FPR64, simm7s8, "stp">; +def STPQpre : StorePairPreIdx<0b10, 1, FPR128, simm7s16, "stp">; + +// Pair (pre-indexed) +def STPWpost : StorePairPostIdx<0b00, 0, GPR32, simm7s4, "stp">; +def STPXpost : StorePairPostIdx<0b10, 0, GPR64, simm7s8, "stp">; +def STPSpost : StorePairPostIdx<0b00, 1, FPR32, simm7s4, "stp">; +def STPDpost : StorePairPostIdx<0b01, 1, FPR64, simm7s8, "stp">; +def STPQpost : StorePairPostIdx<0b10, 1, FPR128, simm7s16, "stp">; + +// Pair (no allocate) +defm STNPW : StorePairNoAlloc<0b00, 0, GPR32, simm7s4, "stnp">; +defm STNPX : StorePairNoAlloc<0b10, 0, GPR64, simm7s8, "stnp">; +defm STNPS : StorePairNoAlloc<0b00, 1, FPR32, simm7s4, "stnp">; +defm STNPD : StorePairNoAlloc<0b01, 1, FPR64, simm7s8, "stnp">; +defm STNPQ : StorePairNoAlloc<0b10, 1, FPR128, simm7s16, "stnp">; + +//--- +// (Register offset) + +// Integer +defm STRBB : Store8RO< 0b00, 0, 0b00, GPR32, "strb", i32, truncstorei8>; +defm STRHH : Store16RO<0b01, 0, 0b00, GPR32, "strh", i32, truncstorei16>; +defm STRW : Store32RO<0b10, 0, 0b00, GPR32, "str", i32, store>; +defm STRX : Store64RO<0b11, 0, 0b00, GPR64, "str", i64, store>; + + +// Floating-point +defm STRB : Store8RO< 0b00, 1, 0b00, FPR8, "str", untyped, store>; +defm STRH : Store16RO<0b01, 1, 0b00, FPR16, "str", f16, store>; +defm STRS : Store32RO<0b10, 1, 0b00, FPR32, "str", f32, store>; +defm STRD : Store64RO<0b11, 1, 0b00, FPR64, "str", f64, store>; +defm STRQ : Store128RO<0b00, 1, 0b10, FPR128, "str", f128, store>; + +multiclass TruncStoreFrom64ROPat { + + def : Pat<(storeop GPR64:$Rt, + (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)), + (STRW (EXTRACT_SUBREG GPR64:$Rt, sub_32), + GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)>; + + def : Pat<(storeop GPR64:$Rt, + (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)), + (STRX (EXTRACT_SUBREG GPR64:$Rt, sub_32), + GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)>; } -defm SBFX : A64I_bitfield_extract<0b00, "sbfx", A64Sbfx>; -defm UBFX : A64I_bitfield_extract<0b10, "ubfx", A64Ubfx>; - -// Again, variants based on BFM modify Rd so need it as an input too. -def BFXILwwii : A64I_bitfield<0b0, 0b01, 0b0, (outs GPR32:$Rd), - (ins GPR32:$src, GPR32:$Rn, bitfield32_imm:$ImmR, bfx32_width:$ImmS), - "bfxil\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - let Constraints = "$src = $Rd"; +let AddedComplexity = 10 in { + // truncstore i64 + defm : TruncStoreFrom64ROPat; + defm : TruncStoreFrom64ROPat; + defm : TruncStoreFrom64ROPat; } -def BFXILxxii : A64I_bitfield<0b1, 0b01, 0b1, (outs GPR64:$Rd), - (ins GPR64:$src, GPR64:$Rn, bitfield64_imm:$ImmR, bfx64_width:$ImmS), - "bfxil\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - let Constraints = "$src = $Rd"; -} +multiclass VecROStorePat { + def : Pat<(store (VecTy FPR:$Rt), + (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)), + (STRW FPR:$Rt, GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)>; -// SBFX instructions can do a 1-instruction sign-extension of boolean values. -def : Pat<(sext_inreg GPR64:$Rn, i1), (SBFXxxii GPR64:$Rn, 0, 0)>; -def : Pat<(sext_inreg GPR32:$Rn, i1), (SBFXwwii GPR32:$Rn, 0, 0)>; -def : Pat<(i64 (sext_inreg (anyext GPR32:$Rn), i1)), - (SBFXxxii (SUBREG_TO_REG (i64 0), GPR32:$Rn, sub_32), 0, 0)>; - -// UBFX makes sense as an implementation of a 64-bit zero-extension too. Could -// use either 64-bit or 32-bit variant, but 32-bit might be more efficient. -def : Pat<(zext GPR32:$Rn), (SUBREG_TO_REG (i64 0), (UBFXwwii GPR32:$Rn, 0, 31), - sub_32)>; - -//===------------------------------- -// 6. Aliases for bitfield insert instructions -//===------------------------------- - -def bfi32_lsb_asmoperand : AsmOperandClass { - let Name = "BFI32LSB"; - let PredicateMethod = "isUImm<5>"; - let RenderMethod = "addBFILSBOperands<32>"; - let DiagnosticType = "UImm5"; + def : Pat<(store (VecTy FPR:$Rt), + (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)), + (STRX FPR:$Rt, GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)>; } -def bfi32_lsb : Operand, - ImmLeaf= 0 && Imm <= 31; }]> { - let PrintMethod = "printBFILSBOperand<32>"; - let ParserMatchClass = bfi32_lsb_asmoperand; -} +let AddedComplexity = 10 in { +// Match all store 64 bits width whose type is compatible with FPR64 +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; +} + +defm : VecROStorePat; +defm : VecROStorePat; + +// Match all store 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; + defm : VecROStorePat; +} +} // AddedComplexity = 10 + +// Match stores from lane 0 to the appropriate subreg's store. +multiclass VecROStoreLane0Pat { + + def : Pat<(storeop (STy (vector_extract (VecTy VecListOne128:$Vt), 0)), + (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)), + (STRW (EXTRACT_SUBREG VecListOne128:$Vt, SubRegIdx), + GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend)>; + + def : Pat<(storeop (STy (vector_extract (VecTy VecListOne128:$Vt), 0)), + (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)), + (STRX (EXTRACT_SUBREG VecListOne128:$Vt, SubRegIdx), + GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend)>; +} + +let AddedComplexity = 19 in { + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; + defm : VecROStoreLane0Pat; +} + +//--- +// (unsigned immediate) +defm STRX : StoreUI<0b11, 0, 0b00, GPR64, uimm12s8, "str", + [(store GPR64:$Rt, + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))]>; +defm STRW : StoreUI<0b10, 0, 0b00, GPR32, uimm12s4, "str", + [(store GPR32:$Rt, + (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))]>; +defm STRB : StoreUI<0b00, 1, 0b00, FPR8, uimm12s1, "str", + [(store FPR8:$Rt, + (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))]>; +defm STRH : StoreUI<0b01, 1, 0b00, FPR16, uimm12s2, "str", + [(store (f16 FPR16:$Rt), + (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))]>; +defm STRS : StoreUI<0b10, 1, 0b00, FPR32, uimm12s4, "str", + [(store (f32 FPR32:$Rt), + (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))]>; +defm STRD : StoreUI<0b11, 1, 0b00, FPR64, uimm12s8, "str", + [(store (f64 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset))]>; +defm STRQ : StoreUI<0b00, 1, 0b10, FPR128, uimm12s16, "str", []>; + +defm STRHH : StoreUI<0b01, 0, 0b00, GPR32, uimm12s2, "strh", + [(truncstorei16 GPR32:$Rt, + (am_indexed16 GPR64sp:$Rn, + uimm12s2:$offset))]>; +defm STRBB : StoreUI<0b00, 0, 0b00, GPR32, uimm12s1, "strb", + [(truncstorei8 GPR32:$Rt, + (am_indexed8 GPR64sp:$Rn, + uimm12s1:$offset))]>; + +// Match all store 64 bits width whose type is compatible with FPR64 +let AddedComplexity = 10 in { +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + def : Pat<(store (v2f32 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(store (v8i8 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(store (v4i16 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(store (v2i32 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; + def : Pat<(store (v4f16 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; +} +def : Pat<(store (v1f64 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; +def : Pat<(store (v1i64 FPR64:$Rt), + (am_indexed64 GPR64sp:$Rn, uimm12s8:$offset)), + (STRDui FPR64:$Rt, GPR64sp:$Rn, uimm12s8:$offset)>; + +// Match all store 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + def : Pat<(store (v4f32 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v2f64 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v16i8 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v8i16 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v4i32 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v2i64 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + def : Pat<(store (v8f16 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; +} +def : Pat<(store (f128 FPR128:$Rt), + (am_indexed128 GPR64sp:$Rn, uimm12s16:$offset)), + (STRQui FPR128:$Rt, GPR64sp:$Rn, uimm12s16:$offset)>; + +// truncstore i64 +def : Pat<(truncstorei32 GPR64:$Rt, + (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset)), + (STRWui (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, uimm12s4:$offset)>; +def : Pat<(truncstorei16 GPR64:$Rt, + (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset)), + (STRHHui (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, uimm12s2:$offset)>; +def : Pat<(truncstorei8 GPR64:$Rt, (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset)), + (STRBBui (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, uimm12s1:$offset)>; + +} // AddedComplexity = 10 + +//--- +// (unscaled immediate) +defm STURX : StoreUnscaled<0b11, 0, 0b00, GPR64, "stur", + [(store GPR64:$Rt, + (am_unscaled64 GPR64sp:$Rn, simm9:$offset))]>; +defm STURW : StoreUnscaled<0b10, 0, 0b00, GPR32, "stur", + [(store GPR32:$Rt, + (am_unscaled32 GPR64sp:$Rn, simm9:$offset))]>; +defm STURB : StoreUnscaled<0b00, 1, 0b00, FPR8, "stur", + [(store FPR8:$Rt, + (am_unscaled8 GPR64sp:$Rn, simm9:$offset))]>; +defm STURH : StoreUnscaled<0b01, 1, 0b00, FPR16, "stur", + [(store (f16 FPR16:$Rt), + (am_unscaled16 GPR64sp:$Rn, simm9:$offset))]>; +defm STURS : StoreUnscaled<0b10, 1, 0b00, FPR32, "stur", + [(store (f32 FPR32:$Rt), + (am_unscaled32 GPR64sp:$Rn, simm9:$offset))]>; +defm STURD : StoreUnscaled<0b11, 1, 0b00, FPR64, "stur", + [(store (f64 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset))]>; +defm STURQ : StoreUnscaled<0b00, 1, 0b10, FPR128, "stur", + [(store (f128 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset))]>; +defm STURHH : StoreUnscaled<0b01, 0, 0b00, GPR32, "sturh", + [(truncstorei16 GPR32:$Rt, + (am_unscaled16 GPR64sp:$Rn, simm9:$offset))]>; +defm STURBB : StoreUnscaled<0b00, 0, 0b00, GPR32, "sturb", + [(truncstorei8 GPR32:$Rt, + (am_unscaled8 GPR64sp:$Rn, simm9:$offset))]>; + +// Match all store 64 bits width whose type is compatible with FPR64 +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + def : Pat<(store (v2f32 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v8i8 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v4i16 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v2i32 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v4f16 FPR64:$Rt), + (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; +} +def : Pat<(store (v1f64 FPR64:$Rt), (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(store (v1i64 FPR64:$Rt), (am_unscaled64 GPR64sp:$Rn, simm9:$offset)), + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9:$offset)>; + +// Match all store 128 bits width whose type is compatible with FPR128 +let Predicates = [IsLE] in { + // We must use ST1 to store vectors in big-endian. + def : Pat<(store (v4f32 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v2f64 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v16i8 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v8i16 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v4i32 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v2i64 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v2f64 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; + def : Pat<(store (v8f16 FPR128:$Rt), + (am_unscaled128 GPR64sp:$Rn, simm9:$offset)), + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9:$offset)>; +} + +// unscaled i64 truncating stores +def : Pat<(truncstorei32 GPR64:$Rt, (am_unscaled32 GPR64sp:$Rn, simm9:$offset)), + (STURWi (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(truncstorei16 GPR64:$Rt, (am_unscaled16 GPR64sp:$Rn, simm9:$offset)), + (STURHHi (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, simm9:$offset)>; +def : Pat<(truncstorei8 GPR64:$Rt, (am_unscaled8 GPR64sp:$Rn, simm9:$offset)), + (STURBBi (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$Rn, simm9:$offset)>; + +//--- +// STR mnemonics fall back to STUR for negative or unaligned offsets. +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURXi GPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb64:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURWi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb32:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURBi FPR8:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURHi FPR16:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURSi FPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb32:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURDi FPR64:$Rt, GPR64sp:$Rn, simm9_offset_fb64:$offset), 0>; +def : InstAlias<"str $Rt, [$Rn, $offset]", + (STURQi FPR128:$Rt, GPR64sp:$Rn, simm9_offset_fb128:$offset), 0>; + +def : InstAlias<"strb $Rt, [$Rn, $offset]", + (STURBBi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb8:$offset), 0>; +def : InstAlias<"strh $Rt, [$Rn, $offset]", + (STURHHi GPR32:$Rt, GPR64sp:$Rn, simm9_offset_fb16:$offset), 0>; + +//--- +// (unscaled immediate, unprivileged) +defm STTRW : StoreUnprivileged<0b10, 0, 0b00, GPR32, "sttr">; +defm STTRX : StoreUnprivileged<0b11, 0, 0b00, GPR64, "sttr">; + +defm STTRH : StoreUnprivileged<0b01, 0, 0b00, GPR32, "sttrh">; +defm STTRB : StoreUnprivileged<0b00, 0, 0b00, GPR32, "sttrb">; + +//--- +// (immediate pre-indexed) +def STRWpre : StorePreIdx<0b10, 0, 0b00, GPR32, "str", pre_store, i32>; +def STRXpre : StorePreIdx<0b11, 0, 0b00, GPR64, "str", pre_store, i64>; +def STRBpre : StorePreIdx<0b00, 1, 0b00, FPR8, "str", pre_store, untyped>; +def STRHpre : StorePreIdx<0b01, 1, 0b00, FPR16, "str", pre_store, f16>; +def STRSpre : StorePreIdx<0b10, 1, 0b00, FPR32, "str", pre_store, f32>; +def STRDpre : StorePreIdx<0b11, 1, 0b00, FPR64, "str", pre_store, f64>; +def STRQpre : StorePreIdx<0b00, 1, 0b10, FPR128, "str", pre_store, f128>; + +def STRBBpre : StorePreIdx<0b00, 0, 0b00, GPR32, "strb", pre_truncsti8, i32>; +def STRHHpre : StorePreIdx<0b01, 0, 0b00, GPR32, "strh", pre_truncsti16, i32>; + +// truncstore i64 +def : Pat<(pre_truncsti32 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRWpre (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; +def : Pat<(pre_truncsti16 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRHHpre (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; +def : Pat<(pre_truncsti8 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRBBpre (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; + +def : Pat<(pre_store (v8i8 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v4i16 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v2i32 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v2f32 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v1i64 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v1f64 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v4f16 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpre FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; + +def : Pat<(pre_store (v16i8 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v8i16 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v4i32 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v4f32 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v2i64 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v2f64 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(pre_store (v8f16 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpre FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; + +//--- +// (immediate post-indexed) +def STRWpost : StorePostIdx<0b10, 0, 0b00, GPR32, "str", post_store, i32>; +def STRXpost : StorePostIdx<0b11, 0, 0b00, GPR64, "str", post_store, i64>; +def STRBpost : StorePostIdx<0b00, 1, 0b00, FPR8, "str", post_store, untyped>; +def STRHpost : StorePostIdx<0b01, 1, 0b00, FPR16, "str", post_store, f16>; +def STRSpost : StorePostIdx<0b10, 1, 0b00, FPR32, "str", post_store, f32>; +def STRDpost : StorePostIdx<0b11, 1, 0b00, FPR64, "str", post_store, f64>; +def STRQpost : StorePostIdx<0b00, 1, 0b10, FPR128, "str", post_store, f128>; + +def STRBBpost : StorePostIdx<0b00, 0, 0b00, GPR32, "strb", post_truncsti8, i32>; +def STRHHpost : StorePostIdx<0b01, 0, 0b00, GPR32, "strh", post_truncsti16, i32>; + +// truncstore i64 +def : Pat<(post_truncsti32 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRWpost (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; +def : Pat<(post_truncsti16 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRHHpost (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; +def : Pat<(post_truncsti8 GPR64:$Rt, GPR64sp:$addr, simm9:$off), + (STRBBpost (EXTRACT_SUBREG GPR64:$Rt, sub_32), GPR64sp:$addr, + simm9:$off)>; + +def : Pat<(post_store (v8i8 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v4i16 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v2i32 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v2f32 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v1i64 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v1f64 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v4f16 FPR64:$Rt), GPR64sp:$addr, simm9:$off), + (STRDpost FPR64:$Rt, GPR64sp:$addr, simm9:$off)>; + +def : Pat<(post_store (v16i8 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v8i16 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v4i32 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v4f32 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v2i64 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v2f64 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; +def : Pat<(post_store (v8f16 FPR128:$Rt), GPR64sp:$addr, simm9:$off), + (STRQpost FPR128:$Rt, GPR64sp:$addr, simm9:$off)>; -def bfi64_lsb_asmoperand : AsmOperandClass { - let Name = "BFI64LSB"; - let PredicateMethod = "isUImm<6>"; - let RenderMethod = "addBFILSBOperands<64>"; - let DiagnosticType = "UImm6"; -} +//===----------------------------------------------------------------------===// +// Load/store exclusive instructions. +//===----------------------------------------------------------------------===// -def bfi64_lsb : Operand, - ImmLeaf= 0 && Imm <= 63; }]> { - let PrintMethod = "printBFILSBOperand<64>"; - let ParserMatchClass = bfi64_lsb_asmoperand; -} +def LDARW : LoadAcquire <0b10, 1, 1, 0, 1, GPR32, "ldar">; +def LDARX : LoadAcquire <0b11, 1, 1, 0, 1, GPR64, "ldar">; +def LDARB : LoadAcquire <0b00, 1, 1, 0, 1, GPR32, "ldarb">; +def LDARH : LoadAcquire <0b01, 1, 1, 0, 1, GPR32, "ldarh">; -// Width verification is performed during conversion so width operand can be -// shared between 32/64-bit cases. Still needed for the print method though -// because ImmR encodes "width - 1". -def bfi32_width_asmoperand : AsmOperandClass { - let Name = "BFI32Width"; - let PredicateMethod = "isBitfieldWidth<32>"; - let RenderMethod = "addBFIWidthOperands"; - let DiagnosticType = "Width32"; -} +def LDAXRW : LoadExclusive <0b10, 0, 1, 0, 1, GPR32, "ldaxr">; +def LDAXRX : LoadExclusive <0b11, 0, 1, 0, 1, GPR64, "ldaxr">; +def LDAXRB : LoadExclusive <0b00, 0, 1, 0, 1, GPR32, "ldaxrb">; +def LDAXRH : LoadExclusive <0b01, 0, 1, 0, 1, GPR32, "ldaxrh">; -def bfi32_width : Operand, - ImmLeaf= 1 && Imm <= 32; }]> { - let PrintMethod = "printBFIWidthOperand"; - let ParserMatchClass = bfi32_width_asmoperand; -} +def LDXRW : LoadExclusive <0b10, 0, 1, 0, 0, GPR32, "ldxr">; +def LDXRX : LoadExclusive <0b11, 0, 1, 0, 0, GPR64, "ldxr">; +def LDXRB : LoadExclusive <0b00, 0, 1, 0, 0, GPR32, "ldxrb">; +def LDXRH : LoadExclusive <0b01, 0, 1, 0, 0, GPR32, "ldxrh">; -def bfi64_width_asmoperand : AsmOperandClass { - let Name = "BFI64Width"; - let PredicateMethod = "isBitfieldWidth<64>"; - let RenderMethod = "addBFIWidthOperands"; - let DiagnosticType = "Width64"; -} +def STLRW : StoreRelease <0b10, 1, 0, 0, 1, GPR32, "stlr">; +def STLRX : StoreRelease <0b11, 1, 0, 0, 1, GPR64, "stlr">; +def STLRB : StoreRelease <0b00, 1, 0, 0, 1, GPR32, "stlrb">; +def STLRH : StoreRelease <0b01, 1, 0, 0, 1, GPR32, "stlrh">; -def bfi64_width : Operand, - ImmLeaf= 1 && Imm <= 64; }]> { - let PrintMethod = "printBFIWidthOperand"; - let ParserMatchClass = bfi64_width_asmoperand; -} +def STLXRW : StoreExclusive<0b10, 0, 0, 0, 1, GPR32, "stlxr">; +def STLXRX : StoreExclusive<0b11, 0, 0, 0, 1, GPR64, "stlxr">; +def STLXRB : StoreExclusive<0b00, 0, 0, 0, 1, GPR32, "stlxrb">; +def STLXRH : StoreExclusive<0b01, 0, 0, 0, 1, GPR32, "stlxrh">; -multiclass A64I_bitfield_insert opc, string asmop> { - def wwii : A64I_bitfield<0b0, opc, 0b0, (outs GPR32:$Rd), - (ins GPR32:$Rn, bfi32_lsb:$ImmR, bfi32_width:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - } +def STXRW : StoreExclusive<0b10, 0, 0, 0, 0, GPR32, "stxr">; +def STXRX : StoreExclusive<0b11, 0, 0, 0, 0, GPR64, "stxr">; +def STXRB : StoreExclusive<0b00, 0, 0, 0, 0, GPR32, "stxrb">; +def STXRH : StoreExclusive<0b01, 0, 0, 0, 0, GPR32, "stxrh">; - def xxii : A64I_bitfield<0b1, opc, 0b1, (outs GPR64:$Rd), - (ins GPR64:$Rn, bfi64_lsb:$ImmR, bfi64_width:$ImmS), - !strconcat(asmop, "\t$Rd, $Rn, $ImmR, $ImmS"), - [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - } -} +def LDAXPW : LoadExclusivePair<0b10, 0, 1, 1, 1, GPR32, "ldaxp">; +def LDAXPX : LoadExclusivePair<0b11, 0, 1, 1, 1, GPR64, "ldaxp">; -defm SBFIZ : A64I_bitfield_insert<0b00, "sbfiz">; -defm UBFIZ : A64I_bitfield_insert<0b10, "ubfiz">; +def LDXPW : LoadExclusivePair<0b10, 0, 1, 1, 0, GPR32, "ldxp">; +def LDXPX : LoadExclusivePair<0b11, 0, 1, 1, 0, GPR64, "ldxp">; +def STLXPW : StoreExclusivePair<0b10, 0, 0, 1, 1, GPR32, "stlxp">; +def STLXPX : StoreExclusivePair<0b11, 0, 0, 1, 1, GPR64, "stlxp">; -def BFIwwii : A64I_bitfield<0b0, 0b01, 0b0, (outs GPR32:$Rd), - (ins GPR32:$src, GPR32:$Rn, bfi32_lsb:$ImmR, bfi32_width:$ImmS), - "bfi\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - let Constraints = "$src = $Rd"; -} +def STXPW : StoreExclusivePair<0b10, 0, 0, 1, 0, GPR32, "stxp">; +def STXPX : StoreExclusivePair<0b11, 0, 0, 1, 0, GPR64, "stxp">; + +let Predicates = [HasV8_1a] in { + // v8.1a "Limited Order Region" extension load-acquire instructions + def LDLARW : LoadAcquire <0b10, 1, 1, 0, 0, GPR32, "ldlar">; + def LDLARX : LoadAcquire <0b11, 1, 1, 0, 0, GPR64, "ldlar">; + def LDLARB : LoadAcquire <0b00, 1, 1, 0, 0, GPR32, "ldlarb">; + def LDLARH : LoadAcquire <0b01, 1, 1, 0, 0, GPR32, "ldlarh">; -def BFIxxii : A64I_bitfield<0b1, 0b01, 0b1, (outs GPR64:$Rd), - (ins GPR64:$src, GPR64:$Rn, bfi64_lsb:$ImmR, bfi64_width:$ImmS), - "bfi\t$Rd, $Rn, $ImmR, $ImmS", [], NoItinerary> { - // As above, no disassembler allowed. - let isAsmParserOnly = 1; - let Constraints = "$src = $Rd"; + // v8.1a "Limited Order Region" extension store-release instructions + def STLLRW : StoreRelease <0b10, 1, 0, 0, 0, GPR32, "stllr">; + def STLLRX : StoreRelease <0b11, 1, 0, 0, 0, GPR64, "stllr">; + def STLLRB : StoreRelease <0b00, 1, 0, 0, 0, GPR32, "stllrb">; + def STLLRH : StoreRelease <0b01, 1, 0, 0, 0, GPR32, "stllrh">; } //===----------------------------------------------------------------------===// -// Compare and branch (immediate) +// Scaled floating point to integer conversion instructions. //===----------------------------------------------------------------------===// -// Contains: CBZ, CBNZ -class label_asmoperand : AsmOperandClass { - let Name = "Label" # width # "_" # scale; - let PredicateMethod = "isLabel<" # width # "," # scale # ">"; - let RenderMethod = "addLabelOperands<" # width # ", " # scale # ">"; - let DiagnosticType = "Label"; -} - -def label_wid19_scal4_asmoperand : label_asmoperand<19, 4>; - -// All conditional immediate branches are the same really: 19 signed bits scaled -// by the instruction-size (4). -def bcc_target : Operand { - // This label is a 19-bit offset from PC, scaled by the instruction-width: 4. - let ParserMatchClass = label_wid19_scal4_asmoperand; - let PrintMethod = "printLabelOperand<19, 4>"; - let EncoderMethod = "getLabelOpValue"; - let OperandType = "OPERAND_PCREL"; -} +defm FCVTAS : FPToIntegerUnscaled<0b00, 0b100, "fcvtas", int_aarch64_neon_fcvtas>; +defm FCVTAU : FPToIntegerUnscaled<0b00, 0b101, "fcvtau", int_aarch64_neon_fcvtau>; +defm FCVTMS : FPToIntegerUnscaled<0b10, 0b000, "fcvtms", int_aarch64_neon_fcvtms>; +defm FCVTMU : FPToIntegerUnscaled<0b10, 0b001, "fcvtmu", int_aarch64_neon_fcvtmu>; +defm FCVTNS : FPToIntegerUnscaled<0b00, 0b000, "fcvtns", int_aarch64_neon_fcvtns>; +defm FCVTNU : FPToIntegerUnscaled<0b00, 0b001, "fcvtnu", int_aarch64_neon_fcvtnu>; +defm FCVTPS : FPToIntegerUnscaled<0b01, 0b000, "fcvtps", int_aarch64_neon_fcvtps>; +defm FCVTPU : FPToIntegerUnscaled<0b01, 0b001, "fcvtpu", int_aarch64_neon_fcvtpu>; +defm FCVTZS : FPToIntegerUnscaled<0b11, 0b000, "fcvtzs", fp_to_sint>; +defm FCVTZU : FPToIntegerUnscaled<0b11, 0b001, "fcvtzu", fp_to_uint>; +defm FCVTZS : FPToIntegerScaled<0b11, 0b000, "fcvtzs", fp_to_sint>; +defm FCVTZU : FPToIntegerScaled<0b11, 0b001, "fcvtzu", fp_to_uint>; +let isCodeGenOnly = 1 in { +defm FCVTZS_Int : FPToIntegerUnscaled<0b11, 0b000, "fcvtzs", int_aarch64_neon_fcvtzs>; +defm FCVTZU_Int : FPToIntegerUnscaled<0b11, 0b001, "fcvtzu", int_aarch64_neon_fcvtzu>; +defm FCVTZS_Int : FPToIntegerScaled<0b11, 0b000, "fcvtzs", int_aarch64_neon_fcvtzs>; +defm FCVTZU_Int : FPToIntegerScaled<0b11, 0b001, "fcvtzu", int_aarch64_neon_fcvtzu>; +} + +multiclass FPToIntegerPats { + def : Pat<(i32 (to_int (round f32:$Rn))), + (!cast(INST # UWSr) f32:$Rn)>; + def : Pat<(i64 (to_int (round f32:$Rn))), + (!cast(INST # UXSr) f32:$Rn)>; + def : Pat<(i32 (to_int (round f64:$Rn))), + (!cast(INST # UWDr) f64:$Rn)>; + def : Pat<(i64 (to_int (round f64:$Rn))), + (!cast(INST # UXDr) f64:$Rn)>; +} + +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; +defm : FPToIntegerPats; -multiclass cmpbr_sizes { - let isBranch = 1, isTerminator = 1 in { - def x : A64I_cmpbr<0b1, op, - (outs), - (ins GPR64:$Rt, bcc_target:$Label), - !strconcat(asmop,"\t$Rt, $Label"), - [(A64br_cc (A64cmp GPR64:$Rt, 0), SETOP, bb:$Label)], - NoItinerary>; - - def w : A64I_cmpbr<0b0, op, - (outs), - (ins GPR32:$Rt, bcc_target:$Label), - !strconcat(asmop,"\t$Rt, $Label"), - [(A64br_cc (A64cmp GPR32:$Rt, 0), SETOP, bb:$Label)], - NoItinerary>; - } -} +//===----------------------------------------------------------------------===// +// Scaled integer to floating point conversion instructions. +//===----------------------------------------------------------------------===// -defm CBZ : cmpbr_sizes<0b0, "cbz", ImmLeaf >; -defm CBNZ : cmpbr_sizes<0b1, "cbnz", ImmLeaf >; +defm SCVTF : IntegerToFP<0, "scvtf", sint_to_fp>; +defm UCVTF : IntegerToFP<1, "ucvtf", uint_to_fp>; //===----------------------------------------------------------------------===// -// Conditional branch (immediate) instructions +// Unscaled integer to floating point conversion instruction. //===----------------------------------------------------------------------===// -// Contains: B.cc -def cond_code_asmoperand : AsmOperandClass { - let Name = "CondCode"; - let DiagnosticType = "CondCode"; -} - -def cond_code : Operand, ImmLeaf= 0 && Imm <= 15; -}]> { - let PrintMethod = "printCondCodeOperand"; - let ParserMatchClass = cond_code_asmoperand; -} +defm FMOV : UnscaledConversion<"fmov">; -def Bcc : A64I_condbr<0b0, 0b0, (outs), - (ins cond_code:$Cond, bcc_target:$Label), - "b.$Cond $Label", [(A64br_cc NZCV, (i32 imm:$Cond), bb:$Label)], - NoItinerary> { - let Uses = [NZCV]; - let isBranch = 1; - let isTerminator = 1; +// Add pseudo ops for FMOV 0 so we can mark them as isReMaterializable +let isReMaterializable = 1, isCodeGenOnly = 1 in { +def FMOVS0 : Pseudo<(outs FPR32:$Rd), (ins), [(set f32:$Rd, (fpimm0))]>, + PseudoInstExpansion<(FMOVWSr FPR32:$Rd, WZR)>, + Requires<[NoZCZ]>; +def FMOVD0 : Pseudo<(outs FPR64:$Rd), (ins), [(set f64:$Rd, (fpimm0))]>, + PseudoInstExpansion<(FMOVXDr FPR64:$Rd, XZR)>, + Requires<[NoZCZ]>; } //===----------------------------------------------------------------------===// -// Conditional compare (immediate) instructions +// Floating point conversion instruction. //===----------------------------------------------------------------------===// -// Contains: CCMN, CCMP -def uimm4_asmoperand : AsmOperandClass { - let Name = "UImm4"; - let PredicateMethod = "isUImm<4>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm4"; -} +defm FCVT : FPConversion<"fcvt">; -def uimm4 : Operand { - let ParserMatchClass = uimm4_asmoperand; -} +//===----------------------------------------------------------------------===// +// Floating point single operand instructions. +//===----------------------------------------------------------------------===// -def uimm5 : Operand { - let ParserMatchClass = uimm5_asmoperand; -} +defm FABS : SingleOperandFPData<0b0001, "fabs", fabs>; +defm FMOV : SingleOperandFPData<0b0000, "fmov">; +defm FNEG : SingleOperandFPData<0b0010, "fneg", fneg>; +defm FRINTA : SingleOperandFPData<0b1100, "frinta", frnd>; +defm FRINTI : SingleOperandFPData<0b1111, "frinti", fnearbyint>; +defm FRINTM : SingleOperandFPData<0b1010, "frintm", ffloor>; +defm FRINTN : SingleOperandFPData<0b1000, "frintn", int_aarch64_neon_frintn>; +defm FRINTP : SingleOperandFPData<0b1001, "frintp", fceil>; -// The only difference between this operand and the one for instructions like -// B.cc is that it's parsed manually. The other get parsed implicitly as part of -// the mnemonic handling. -def cond_code_op_asmoperand : AsmOperandClass { - let Name = "CondCodeOp"; - let RenderMethod = "addCondCodeOperands"; - let PredicateMethod = "isCondCode"; - let ParserMethod = "ParseCondCodeOperand"; - let DiagnosticType = "CondCode"; -} +def : Pat<(v1f64 (int_aarch64_neon_frintn (v1f64 FPR64:$Rn))), + (FRINTNDr FPR64:$Rn)>; -def cond_code_op : Operand { - let PrintMethod = "printCondCodeOperand"; - let ParserMatchClass = cond_code_op_asmoperand; -} +defm FRINTX : SingleOperandFPData<0b1110, "frintx", frint>; +defm FRINTZ : SingleOperandFPData<0b1011, "frintz", ftrunc>; -class A64I_condcmpimmImpl - : A64I_condcmpimm { - let Defs = [NZCV]; +let SchedRW = [WriteFDiv] in { +defm FSQRT : SingleOperandFPData<0b0011, "fsqrt", fsqrt>; } -def CCMNwi : A64I_condcmpimmImpl<0b0, 0b0, GPR32, "ccmn">; -def CCMNxi : A64I_condcmpimmImpl<0b1, 0b0, GPR64, "ccmn">; -def CCMPwi : A64I_condcmpimmImpl<0b0, 0b1, GPR32, "ccmp">; -def CCMPxi : A64I_condcmpimmImpl<0b1, 0b1, GPR64, "ccmp">; - //===----------------------------------------------------------------------===// -// Conditional compare (register) instructions +// Floating point two operand instructions. //===----------------------------------------------------------------------===// -// Contains: CCMN, CCMP - -class A64I_condcmpregImpl - : A64I_condcmpreg { - let Defs = [NZCV]; -} -def CCMNww : A64I_condcmpregImpl<0b0, 0b0, GPR32, "ccmn">; -def CCMNxx : A64I_condcmpregImpl<0b1, 0b0, GPR64, "ccmn">; -def CCMPww : A64I_condcmpregImpl<0b0, 0b1, GPR32, "ccmp">; -def CCMPxx : A64I_condcmpregImpl<0b1, 0b1, GPR64, "ccmp">; +defm FADD : TwoOperandFPData<0b0010, "fadd", fadd>; +let SchedRW = [WriteFDiv] in { +defm FDIV : TwoOperandFPData<0b0001, "fdiv", fdiv>; +} +defm FMAXNM : TwoOperandFPData<0b0110, "fmaxnm", fmaxnum>; +defm FMAX : TwoOperandFPData<0b0100, "fmax", fmaxnan>; +defm FMINNM : TwoOperandFPData<0b0111, "fminnm", fminnum>; +defm FMIN : TwoOperandFPData<0b0101, "fmin", fminnan>; +let SchedRW = [WriteFMul] in { +defm FMUL : TwoOperandFPData<0b0000, "fmul", fmul>; +defm FNMUL : TwoOperandFPDataNeg<0b1000, "fnmul", fmul>; +} +defm FSUB : TwoOperandFPData<0b0011, "fsub", fsub>; + +def : Pat<(v1f64 (fmaxnan (v1f64 FPR64:$Rn), (v1f64 FPR64:$Rm))), + (FMAXDrr FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(v1f64 (fminnan (v1f64 FPR64:$Rn), (v1f64 FPR64:$Rm))), + (FMINDrr FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(v1f64 (fmaxnum (v1f64 FPR64:$Rn), (v1f64 FPR64:$Rm))), + (FMAXNMDrr FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(v1f64 (fminnum (v1f64 FPR64:$Rn), (v1f64 FPR64:$Rm))), + (FMINNMDrr FPR64:$Rn, FPR64:$Rm)>; //===----------------------------------------------------------------------===// -// Conditional select instructions +// Floating point three operand instructions. //===----------------------------------------------------------------------===// -// Contains: CSEL, CSINC, CSINV, CSNEG + aliases CSET, CSETM, CINC, CINV, CNEG - -// Condition code which is encoded as the inversion (semantically rather than -// bitwise) in the instruction. -def inv_cond_code_op_asmoperand : AsmOperandClass { - let Name = "InvCondCodeOp"; - let RenderMethod = "addInvCondCodeOperands"; - let PredicateMethod = "isCondCode"; - let ParserMethod = "ParseCondCodeOperand"; - let DiagnosticType = "CondCode"; -} -def inv_cond_code_op : Operand { - let ParserMatchClass = inv_cond_code_op_asmoperand; -} +defm FMADD : ThreeOperandFPData<0, 0, "fmadd", fma>; +defm FMSUB : ThreeOperandFPData<0, 1, "fmsub", + TriOpFrag<(fma node:$LHS, (fneg node:$MHS), node:$RHS)> >; +defm FNMADD : ThreeOperandFPData<1, 0, "fnmadd", + TriOpFrag<(fneg (fma node:$LHS, node:$MHS, node:$RHS))> >; +defm FNMSUB : ThreeOperandFPData<1, 1, "fnmsub", + TriOpFrag<(fma node:$LHS, node:$MHS, (fneg node:$RHS))> >; -// Having a separate operand for the selectable use-case is debatable, but gives -// consistency with cond_code. -def inv_cond_XFORM : SDNodeXForm(N->getZExtValue()); - return CurDAG->getTargetConstant(A64InvertCondCode(CC), MVT::i32); -}]>; +// The following def pats catch the case where the LHS of an FMA is negated. +// The TriOpFrag above catches the case where the middle operand is negated. -def inv_cond_code - : ImmLeaf= 0 && Imm <= 15; }], inv_cond_XFORM>; +// N.b. FMSUB etc have the accumulator at the *end* of (outs), unlike +// the NEON variant. +def : Pat<(f32 (fma (fneg FPR32:$Rn), FPR32:$Rm, FPR32:$Ra)), + (FMSUBSrrr FPR32:$Rn, FPR32:$Rm, FPR32:$Ra)>; +def : Pat<(f64 (fma (fneg FPR64:$Rn), FPR64:$Rm, FPR64:$Ra)), + (FMSUBDrrr FPR64:$Rn, FPR64:$Rm, FPR64:$Ra)>; -multiclass A64I_condselSizes op2, string asmop, - SDPatternOperator select> { - let Uses = [NZCV] in { - def wwwc : A64I_condsel<0b0, op, 0b0, op2, - (outs GPR32:$Rd), - (ins GPR32:$Rn, GPR32:$Rm, cond_code_op:$Cond), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Cond"), - [(set GPR32:$Rd, (select GPR32:$Rn, GPR32:$Rm))], - NoItinerary>; +// We handled -(a + b*c) for FNMADD above, now it's time for "(-a) + (-b)*c" and +// "(-a) + b*(-c)". +def : Pat<(f32 (fma (fneg FPR32:$Rn), FPR32:$Rm, (fneg FPR32:$Ra))), + (FNMADDSrrr FPR32:$Rn, FPR32:$Rm, FPR32:$Ra)>; +def : Pat<(f64 (fma (fneg FPR64:$Rn), FPR64:$Rm, (fneg FPR64:$Ra))), + (FNMADDDrrr FPR64:$Rn, FPR64:$Rm, FPR64:$Ra)>; - def xxxc : A64I_condsel<0b1, op, 0b0, op2, - (outs GPR64:$Rd), - (ins GPR64:$Rn, GPR64:$Rm, cond_code_op:$Cond), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Cond"), - [(set GPR64:$Rd, (select GPR64:$Rn, GPR64:$Rm))], - NoItinerary>; - } -} +def : Pat<(f32 (fma FPR32:$Rn, (fneg FPR32:$Rm), (fneg FPR32:$Ra))), + (FNMADDSrrr FPR32:$Rn, FPR32:$Rm, FPR32:$Ra)>; -def simple_select - : PatFrag<(ops node:$lhs, node:$rhs), - (A64select_cc NZCV, node:$lhs, node:$rhs, (i32 imm:$Cond))>; - -class complex_select - : PatFrag<(ops node:$lhs, node:$rhs), - (A64select_cc NZCV, node:$lhs, (opnode node:$rhs), (i32 imm:$Cond))>; - - -defm CSEL : A64I_condselSizes<0b0, 0b00, "csel", simple_select>; -defm CSINC : A64I_condselSizes<0b0, 0b01, "csinc", - complex_select>>; -defm CSINV : A64I_condselSizes<0b1, 0b00, "csinv", complex_select>; -defm CSNEG : A64I_condselSizes<0b1, 0b01, "csneg", complex_select>; - -// Now the instruction aliases, which fit nicely into LLVM's model: - -def : InstAlias<"cset $Rd, $Cond", - (CSINCwwwc GPR32:$Rd, WZR, WZR, inv_cond_code_op:$Cond)>; -def : InstAlias<"cset $Rd, $Cond", - (CSINCxxxc GPR64:$Rd, XZR, XZR, inv_cond_code_op:$Cond)>; -def : InstAlias<"csetm $Rd, $Cond", - (CSINVwwwc GPR32:$Rd, WZR, WZR, inv_cond_code_op:$Cond)>; -def : InstAlias<"csetm $Rd, $Cond", - (CSINVxxxc GPR64:$Rd, XZR, XZR, inv_cond_code_op:$Cond)>; -def : InstAlias<"cinc $Rd, $Rn, $Cond", - (CSINCwwwc GPR32:$Rd, GPR32:$Rn, GPR32:$Rn, inv_cond_code_op:$Cond)>; -def : InstAlias<"cinc $Rd, $Rn, $Cond", - (CSINCxxxc GPR64:$Rd, GPR64:$Rn, GPR64:$Rn, inv_cond_code_op:$Cond)>; -def : InstAlias<"cinv $Rd, $Rn, $Cond", - (CSINVwwwc GPR32:$Rd, GPR32:$Rn, GPR32:$Rn, inv_cond_code_op:$Cond)>; -def : InstAlias<"cinv $Rd, $Rn, $Cond", - (CSINVxxxc GPR64:$Rd, GPR64:$Rn, GPR64:$Rn, inv_cond_code_op:$Cond)>; -def : InstAlias<"cneg $Rd, $Rn, $Cond", - (CSNEGwwwc GPR32:$Rd, GPR32:$Rn, GPR32:$Rn, inv_cond_code_op:$Cond)>; -def : InstAlias<"cneg $Rd, $Rn, $Cond", - (CSNEGxxxc GPR64:$Rd, GPR64:$Rn, GPR64:$Rn, inv_cond_code_op:$Cond)>; - -// Finally some helper patterns. - -// For CSET (a.k.a. zero-extension of icmp) -def : Pat<(A64select_cc NZCV, 0, 1, cond_code:$Cond), - (CSINCwwwc WZR, WZR, cond_code:$Cond)>; -def : Pat<(A64select_cc NZCV, 1, 0, inv_cond_code:$Cond), - (CSINCwwwc WZR, WZR, inv_cond_code:$Cond)>; - -def : Pat<(A64select_cc NZCV, 0, 1, cond_code:$Cond), - (CSINCxxxc XZR, XZR, cond_code:$Cond)>; -def : Pat<(A64select_cc NZCV, 1, 0, inv_cond_code:$Cond), - (CSINCxxxc XZR, XZR, inv_cond_code:$Cond)>; - -// For CSETM (a.k.a. sign-extension of icmp) -def : Pat<(A64select_cc NZCV, 0, -1, cond_code:$Cond), - (CSINVwwwc WZR, WZR, cond_code:$Cond)>; -def : Pat<(A64select_cc NZCV, -1, 0, inv_cond_code:$Cond), - (CSINVwwwc WZR, WZR, inv_cond_code:$Cond)>; - -def : Pat<(A64select_cc NZCV, 0, -1, cond_code:$Cond), - (CSINVxxxc XZR, XZR, cond_code:$Cond)>; -def : Pat<(A64select_cc NZCV, -1, 0, inv_cond_code:$Cond), - (CSINVxxxc XZR, XZR, inv_cond_code:$Cond)>; - -// CINC, CINV and CNEG get dealt with automatically, which leaves the issue of -// commutativity. The instructions are to complex for isCommutable to be used, -// so we have to create the patterns manually: - -// No commutable pattern for CSEL since the commuted version is isomorphic. - -// CSINC -def :Pat<(A64select_cc NZCV, (add GPR32:$Rm, 1), GPR32:$Rn, - inv_cond_code:$Cond), - (CSINCwwwc GPR32:$Rn, GPR32:$Rm, inv_cond_code:$Cond)>; -def :Pat<(A64select_cc NZCV, (add GPR64:$Rm, 1), GPR64:$Rn, - inv_cond_code:$Cond), - (CSINCxxxc GPR64:$Rn, GPR64:$Rm, inv_cond_code:$Cond)>; - -// CSINV -def :Pat<(A64select_cc NZCV, (not GPR32:$Rm), GPR32:$Rn, inv_cond_code:$Cond), - (CSINVwwwc GPR32:$Rn, GPR32:$Rm, inv_cond_code:$Cond)>; -def :Pat<(A64select_cc NZCV, (not GPR64:$Rm), GPR64:$Rn, inv_cond_code:$Cond), - (CSINVxxxc GPR64:$Rn, GPR64:$Rm, inv_cond_code:$Cond)>; - -// CSNEG -def :Pat<(A64select_cc NZCV, (ineg GPR32:$Rm), GPR32:$Rn, inv_cond_code:$Cond), - (CSNEGwwwc GPR32:$Rn, GPR32:$Rm, inv_cond_code:$Cond)>; -def :Pat<(A64select_cc NZCV, (ineg GPR64:$Rm), GPR64:$Rn, inv_cond_code:$Cond), - (CSNEGxxxc GPR64:$Rn, GPR64:$Rm, inv_cond_code:$Cond)>; +def : Pat<(f64 (fma FPR64:$Rn, (fneg FPR64:$Rm), (fneg FPR64:$Ra))), + (FNMADDDrrr FPR64:$Rn, FPR64:$Rm, FPR64:$Ra)>; //===----------------------------------------------------------------------===// -// Data Processing (1 source) instructions +// Floating point comparison instructions. //===----------------------------------------------------------------------===// -// Contains: RBIT, REV16, REV, REV32, CLZ, CLS. - -// We define an unary operator which always fails. We will use this to -// define unary operators that cannot be matched. - -class A64I_dp_1src_impl opcode, string asmop, - list patterns, RegisterClass GPRrc, - InstrItinClass itin>: - A64I_dp_1src; - -multiclass A64I_dp_1src opcode, string asmop> { - let hasSideEffects = 0 in { - def ww : A64I_dp_1src_impl<0b0, opcode, asmop, [], GPR32, NoItinerary>; - def xx : A64I_dp_1src_impl<0b1, opcode, asmop, [], GPR64, NoItinerary>; - } -} -defm RBIT : A64I_dp_1src<0b000000, "rbit">; -defm CLS : A64I_dp_1src<0b000101, "cls">; -defm CLZ : A64I_dp_1src<0b000100, "clz">; - -def : Pat<(ctlz GPR32:$Rn), (CLZww GPR32:$Rn)>; -def : Pat<(ctlz GPR64:$Rn), (CLZxx GPR64:$Rn)>; -def : Pat<(ctlz_zero_undef GPR32:$Rn), (CLZww GPR32:$Rn)>; -def : Pat<(ctlz_zero_undef GPR64:$Rn), (CLZxx GPR64:$Rn)>; - -def : Pat<(cttz GPR32:$Rn), (CLZww (RBITww GPR32:$Rn))>; -def : Pat<(cttz GPR64:$Rn), (CLZxx (RBITxx GPR64:$Rn))>; -def : Pat<(cttz_zero_undef GPR32:$Rn), (CLZww (RBITww GPR32:$Rn))>; -def : Pat<(cttz_zero_undef GPR64:$Rn), (CLZxx (RBITxx GPR64:$Rn))>; - - -def REVww : A64I_dp_1src_impl<0b0, 0b000010, "rev", - [(set GPR32:$Rd, (bswap GPR32:$Rn))], - GPR32, NoItinerary>; -def REVxx : A64I_dp_1src_impl<0b1, 0b000011, "rev", - [(set GPR64:$Rd, (bswap GPR64:$Rn))], - GPR64, NoItinerary>; -def REV32xx : A64I_dp_1src_impl<0b1, 0b000010, "rev32", - [(set GPR64:$Rd, (bswap (rotr GPR64:$Rn, (i64 32))))], - GPR64, NoItinerary>; -def REV16ww : A64I_dp_1src_impl<0b0, 0b000001, "rev16", - [(set GPR32:$Rd, (bswap (rotr GPR32:$Rn, (i64 16))))], - GPR32, - NoItinerary>; -def REV16xx : A64I_dp_1src_impl<0b1, 0b000001, "rev16", [], GPR64, NoItinerary>; +defm FCMPE : FPComparison<1, "fcmpe">; +defm FCMP : FPComparison<0, "fcmp", AArch64fcmp>; //===----------------------------------------------------------------------===// -// Data Processing (2 sources) instructions +// Floating point conditional comparison instructions. //===----------------------------------------------------------------------===// -// Contains: CRC32C?[BHWX], UDIV, SDIV, LSLV, LSRV, ASRV, RORV + aliases LSL, -// LSR, ASR, ROR - - -class dp_2src_impl opcode, string asmop, list patterns, - RegisterClass GPRsp, - InstrItinClass itin>: - A64I_dp_2src; - -multiclass dp_2src_crc { - def B_www : dp_2src_impl<0b0, {0, 1, 0, c, 0, 0}, - !strconcat(asmop, "b"), [], GPR32, NoItinerary>; - def H_www : dp_2src_impl<0b0, {0, 1, 0, c, 0, 1}, - !strconcat(asmop, "h"), [], GPR32, NoItinerary>; - def W_www : dp_2src_impl<0b0, {0, 1, 0, c, 1, 0}, - !strconcat(asmop, "w"), [], GPR32, NoItinerary>; - def X_wwx : A64I_dp_2src<0b1, {0, 1, 0, c, 1, 1}, 0b0, - !strconcat(asmop, "x\t$Rd, $Rn, $Rm"), - (outs GPR32:$Rd), (ins GPR32:$Rn, GPR64:$Rm), [], - NoItinerary>; -} - -multiclass dp_2src_zext opcode, string asmop, SDPatternOperator op> { - def www : dp_2src_impl<0b0, - opcode, - asmop, - [(set GPR32:$Rd, - (op GPR32:$Rn, (i64 (zext GPR32:$Rm))))], - GPR32, - NoItinerary>; - def xxx : dp_2src_impl<0b1, - opcode, - asmop, - [(set GPR64:$Rd, (op GPR64:$Rn, GPR64:$Rm))], - GPR64, - NoItinerary>; -} - - -multiclass dp_2src opcode, string asmop, SDPatternOperator op> { - def www : dp_2src_impl<0b0, - opcode, - asmop, - [(set GPR32:$Rd, (op GPR32:$Rn, GPR32:$Rm))], - GPR32, - NoItinerary>; - def xxx : dp_2src_impl<0b1, - opcode, - asmop, - [(set GPR64:$Rd, (op GPR64:$Rn, GPR64:$Rm))], - GPR64, - NoItinerary>; -} -// Here we define the data processing 2 source instructions. -defm CRC32 : dp_2src_crc<0b0, "crc32">; -defm CRC32C : dp_2src_crc<0b1, "crc32c">; - -defm UDIV : dp_2src<0b000010, "udiv", udiv>; -defm SDIV : dp_2src<0b000011, "sdiv", sdiv>; - -defm LSLV : dp_2src_zext<0b001000, "lsl", shl>; -defm LSRV : dp_2src_zext<0b001001, "lsr", srl>; -defm ASRV : dp_2src_zext<0b001010, "asr", sra>; -defm RORV : dp_2src_zext<0b001011, "ror", rotr>; - -// Extra patterns for an incoming 64-bit value for a 32-bit -// operation. Since the LLVM operations are undefined (as in C) if the -// RHS is out of range, it's perfectly permissible to discard the high -// bits of the GPR64. -def : Pat<(shl GPR32:$Rn, GPR64:$Rm), - (LSLVwww GPR32:$Rn, (EXTRACT_SUBREG GPR64:$Rm, sub_32))>; -def : Pat<(srl GPR32:$Rn, GPR64:$Rm), - (LSRVwww GPR32:$Rn, (EXTRACT_SUBREG GPR64:$Rm, sub_32))>; -def : Pat<(sra GPR32:$Rn, GPR64:$Rm), - (ASRVwww GPR32:$Rn, (EXTRACT_SUBREG GPR64:$Rm, sub_32))>; -def : Pat<(rotr GPR32:$Rn, GPR64:$Rm), - (RORVwww GPR32:$Rn, (EXTRACT_SUBREG GPR64:$Rm, sub_32))>; - -// Here we define the aliases for the data processing 2 source instructions. -def LSL_mnemonic : MnemonicAlias<"lslv", "lsl">; -def LSR_mnemonic : MnemonicAlias<"lsrv", "lsr">; -def ASR_menmonic : MnemonicAlias<"asrv", "asr">; -def ROR_menmonic : MnemonicAlias<"rorv", "ror">; +defm FCCMPE : FPCondComparison<1, "fccmpe">; +defm FCCMP : FPCondComparison<0, "fccmp", AArch64fccmp>; //===----------------------------------------------------------------------===// -// Data Processing (3 sources) instructions +// Floating point conditional select instruction. //===----------------------------------------------------------------------===// -// Contains: MADD, MSUB, SMADDL, SMSUBL, SMULH, UMADDL, UMSUBL, UMULH -// + aliases MUL, MNEG, SMULL, SMNEGL, UMULL, UMNEGL - -class A64I_dp3_4operand opcode, RegisterClass AccReg, - RegisterClass SrcReg, string asmop, dag pattern> - : A64I_dp3 { - RegisterClass AccGPR = AccReg; - RegisterClass SrcGPR = SrcReg; -} - -def MADDwwww : A64I_dp3_4operand<0b0, 0b000000, GPR32, GPR32, "madd", - (add GPR32:$Ra, (mul GPR32:$Rn, GPR32:$Rm))>; -def MADDxxxx : A64I_dp3_4operand<0b1, 0b000000, GPR64, GPR64, "madd", - (add GPR64:$Ra, (mul GPR64:$Rn, GPR64:$Rm))>; - -def MSUBwwww : A64I_dp3_4operand<0b0, 0b000001, GPR32, GPR32, "msub", - (sub GPR32:$Ra, (mul GPR32:$Rn, GPR32:$Rm))>; -def MSUBxxxx : A64I_dp3_4operand<0b1, 0b000001, GPR64, GPR64, "msub", - (sub GPR64:$Ra, (mul GPR64:$Rn, GPR64:$Rm))>; - -def SMADDLxwwx : A64I_dp3_4operand<0b1, 0b000010, GPR64, GPR32, "smaddl", - (add GPR64:$Ra, (mul (i64 (sext GPR32:$Rn)), (sext GPR32:$Rm)))>; -def SMSUBLxwwx : A64I_dp3_4operand<0b1, 0b000011, GPR64, GPR32, "smsubl", - (sub GPR64:$Ra, (mul (i64 (sext GPR32:$Rn)), (sext GPR32:$Rm)))>; - -def UMADDLxwwx : A64I_dp3_4operand<0b1, 0b001010, GPR64, GPR32, "umaddl", - (add GPR64:$Ra, (mul (i64 (zext GPR32:$Rn)), (zext GPR32:$Rm)))>; -def UMSUBLxwwx : A64I_dp3_4operand<0b1, 0b001011, GPR64, GPR32, "umsubl", - (sub GPR64:$Ra, (mul (i64 (zext GPR32:$Rn)), (zext GPR32:$Rm)))>; - -let isCommutable = 1, PostEncoderMethod = "fixMulHigh" in { - def UMULHxxx : A64I_dp3<0b1, 0b001100, (outs GPR64:$Rd), - (ins GPR64:$Rn, GPR64:$Rm), - "umulh\t$Rd, $Rn, $Rm", - [(set GPR64:$Rd, (mulhu GPR64:$Rn, GPR64:$Rm))], - NoItinerary>; - - def SMULHxxx : A64I_dp3<0b1, 0b000100, (outs GPR64:$Rd), - (ins GPR64:$Rn, GPR64:$Rm), - "smulh\t$Rd, $Rn, $Rm", - [(set GPR64:$Rd, (mulhs GPR64:$Rn, GPR64:$Rm))], - NoItinerary>; -} -multiclass A64I_dp3_3operand { - def : InstAlias; +defm FCSEL : FPCondSelect<"fcsel">; - def : Pat; +// CSEL instructions providing f128 types need to be handled by a +// pseudo-instruction since the eventual code will need to introduce basic +// blocks and control flow. +def F128CSEL : Pseudo<(outs FPR128:$Rd), + (ins FPR128:$Rn, FPR128:$Rm, ccode:$cond), + [(set (f128 FPR128:$Rd), + (AArch64csel FPR128:$Rn, FPR128:$Rm, + (i32 imm:$cond), NZCV))]> { + let Uses = [NZCV]; + let usesCustomInserter = 1; } -defm : A64I_dp3_3operand<"mul", MADDwwww, WZR, (mul GPR32:$Rn, GPR32:$Rm)>; -defm : A64I_dp3_3operand<"mul", MADDxxxx, XZR, (mul GPR64:$Rn, GPR64:$Rm)>; - -defm : A64I_dp3_3operand<"mneg", MSUBwwww, WZR, - (sub 0, (mul GPR32:$Rn, GPR32:$Rm))>; -defm : A64I_dp3_3operand<"mneg", MSUBxxxx, XZR, - (sub 0, (mul GPR64:$Rn, GPR64:$Rm))>; - -defm : A64I_dp3_3operand<"smull", SMADDLxwwx, XZR, - (mul (i64 (sext GPR32:$Rn)), (sext GPR32:$Rm))>; -defm : A64I_dp3_3operand<"smnegl", SMSUBLxwwx, XZR, - (sub 0, (mul (i64 (sext GPR32:$Rn)), (sext GPR32:$Rm)))>; - -defm : A64I_dp3_3operand<"umull", UMADDLxwwx, XZR, - (mul (i64 (zext GPR32:$Rn)), (zext GPR32:$Rm))>; -defm : A64I_dp3_3operand<"umnegl", UMSUBLxwwx, XZR, - (sub 0, (mul (i64 (zext GPR32:$Rn)), (zext GPR32:$Rm)))>; - //===----------------------------------------------------------------------===// -// Exception generation +// Floating point immediate move. //===----------------------------------------------------------------------===// -// Contains: SVC, HVC, SMC, BRK, HLT, DCPS1, DCPS2, DCPS3 - -def uimm16_asmoperand : AsmOperandClass { - let Name = "UImm16"; - let PredicateMethod = "isUImm<16>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm16"; -} - -def uimm16 : Operand { - let ParserMatchClass = uimm16_asmoperand; -} -class A64I_exceptImpl opc, bits<2> ll, string asmop> - : A64I_exception { - let isBranch = 1; - let isTerminator = 1; +let isReMaterializable = 1 in { +defm FMOV : FPMoveImmediate<"fmov">; } -def SVCi : A64I_exceptImpl<0b000, 0b01, "svc">; -def HVCi : A64I_exceptImpl<0b000, 0b10, "hvc">; -def SMCi : A64I_exceptImpl<0b000, 0b11, "smc">; -def BRKi : A64I_exceptImpl<0b001, 0b00, "brk">; -def HLTi : A64I_exceptImpl<0b010, 0b00, "hlt">; - -def DCPS1i : A64I_exceptImpl<0b101, 0b01, "dcps1">; -def DCPS2i : A64I_exceptImpl<0b101, 0b10, "dcps2">; -def DCPS3i : A64I_exceptImpl<0b101, 0b11, "dcps3">; - -// The immediate is optional for the DCPS instructions, defaulting to 0. -def : InstAlias<"dcps1", (DCPS1i 0)>; -def : InstAlias<"dcps2", (DCPS2i 0)>; -def : InstAlias<"dcps3", (DCPS3i 0)>; - //===----------------------------------------------------------------------===// -// Extract (immediate) +// Advanced SIMD two vector instructions. //===----------------------------------------------------------------------===// -// Contains: EXTR + alias ROR - -def EXTRwwwi : A64I_extract<0b0, 0b000, 0b0, - (outs GPR32:$Rd), - (ins GPR32:$Rn, GPR32:$Rm, bitfield32_imm:$LSB), - "extr\t$Rd, $Rn, $Rm, $LSB", - [(set GPR32:$Rd, - (A64Extr GPR32:$Rn, GPR32:$Rm, imm:$LSB))], - NoItinerary>; -def EXTRxxxi : A64I_extract<0b1, 0b000, 0b1, - (outs GPR64:$Rd), - (ins GPR64:$Rn, GPR64:$Rm, bitfield64_imm:$LSB), - "extr\t$Rd, $Rn, $Rm, $LSB", - [(set GPR64:$Rd, - (A64Extr GPR64:$Rn, GPR64:$Rm, imm:$LSB))], - NoItinerary>; - -def : InstAlias<"ror $Rd, $Rs, $LSB", - (EXTRwwwi GPR32:$Rd, GPR32:$Rs, GPR32:$Rs, bitfield32_imm:$LSB)>; -def : InstAlias<"ror $Rd, $Rs, $LSB", - (EXTRxxxi GPR64:$Rd, GPR64:$Rs, GPR64:$Rs, bitfield64_imm:$LSB)>; - -def : Pat<(rotr GPR32:$Rn, bitfield32_imm:$LSB), - (EXTRwwwi GPR32:$Rn, GPR32:$Rn, bitfield32_imm:$LSB)>; -def : Pat<(rotr GPR64:$Rn, bitfield64_imm:$LSB), - (EXTRxxxi GPR64:$Rn, GPR64:$Rn, bitfield64_imm:$LSB)>; + +defm UABDL : SIMDLongThreeVectorBHSabdl<1, 0b0111, "uabdl", + uabsdiff>; +// Match UABDL in log2-shuffle patterns. +def : Pat<(xor (v8i16 (AArch64vashr v8i16:$src, (i32 15))), + (v8i16 (add (sub (zext (v8i8 V64:$opA)), + (zext (v8i8 V64:$opB))), + (AArch64vashr v8i16:$src, (i32 15))))), + (UABDLv8i8_v8i16 V64:$opA, V64:$opB)>; +def : Pat<(xor (v8i16 (AArch64vashr v8i16:$src, (i32 15))), + (v8i16 (add (sub (zext (extract_high_v16i8 V128:$opA)), + (zext (extract_high_v16i8 V128:$opB))), + (AArch64vashr v8i16:$src, (i32 15))))), + (UABDLv16i8_v8i16 V128:$opA, V128:$opB)>; +def : Pat<(xor (v4i32 (AArch64vashr v4i32:$src, (i32 31))), + (v4i32 (add (sub (zext (v4i16 V64:$opA)), + (zext (v4i16 V64:$opB))), + (AArch64vashr v4i32:$src, (i32 31))))), + (UABDLv4i16_v4i32 V64:$opA, V64:$opB)>; +def : Pat<(xor (v4i32 (AArch64vashr v4i32:$src, (i32 31))), + (v4i32 (add (sub (zext (extract_high_v8i16 V128:$opA)), + (zext (extract_high_v8i16 V128:$opB))), + (AArch64vashr v4i32:$src, (i32 31))))), + (UABDLv8i16_v4i32 V128:$opA, V128:$opB)>; +def : Pat<(xor (v2i64 (AArch64vashr v2i64:$src, (i32 63))), + (v2i64 (add (sub (zext (v2i32 V64:$opA)), + (zext (v2i32 V64:$opB))), + (AArch64vashr v2i64:$src, (i32 63))))), + (UABDLv2i32_v2i64 V64:$opA, V64:$opB)>; +def : Pat<(xor (v2i64 (AArch64vashr v2i64:$src, (i32 63))), + (v2i64 (add (sub (zext (extract_high_v4i32 V128:$opA)), + (zext (extract_high_v4i32 V128:$opB))), + (AArch64vashr v2i64:$src, (i32 63))))), + (UABDLv4i32_v2i64 V128:$opA, V128:$opB)>; + +defm ABS : SIMDTwoVectorBHSD<0, 0b01011, "abs", int_aarch64_neon_abs>; +def : Pat<(xor (v8i8 (AArch64vashr V64:$src, (i32 7))), + (v8i8 (add V64:$src, (AArch64vashr V64:$src, (i32 7))))), + (ABSv8i8 V64:$src)>; +def : Pat<(xor (v4i16 (AArch64vashr V64:$src, (i32 15))), + (v4i16 (add V64:$src, (AArch64vashr V64:$src, (i32 15))))), + (ABSv4i16 V64:$src)>; +def : Pat<(xor (v2i32 (AArch64vashr V64:$src, (i32 31))), + (v2i32 (add V64:$src, (AArch64vashr V64:$src, (i32 31))))), + (ABSv2i32 V64:$src)>; +def : Pat<(xor (v16i8 (AArch64vashr V128:$src, (i32 7))), + (v16i8 (add V128:$src, (AArch64vashr V128:$src, (i32 7))))), + (ABSv16i8 V128:$src)>; +def : Pat<(xor (v8i16 (AArch64vashr V128:$src, (i32 15))), + (v8i16 (add V128:$src, (AArch64vashr V128:$src, (i32 15))))), + (ABSv8i16 V128:$src)>; +def : Pat<(xor (v4i32 (AArch64vashr V128:$src, (i32 31))), + (v4i32 (add V128:$src, (AArch64vashr V128:$src, (i32 31))))), + (ABSv4i32 V128:$src)>; +def : Pat<(xor (v2i64 (AArch64vashr V128:$src, (i32 63))), + (v2i64 (add V128:$src, (AArch64vashr V128:$src, (i32 63))))), + (ABSv2i64 V128:$src)>; + +defm CLS : SIMDTwoVectorBHS<0, 0b00100, "cls", int_aarch64_neon_cls>; +defm CLZ : SIMDTwoVectorBHS<1, 0b00100, "clz", ctlz>; +defm CMEQ : SIMDCmpTwoVector<0, 0b01001, "cmeq", AArch64cmeqz>; +defm CMGE : SIMDCmpTwoVector<1, 0b01000, "cmge", AArch64cmgez>; +defm CMGT : SIMDCmpTwoVector<0, 0b01000, "cmgt", AArch64cmgtz>; +defm CMLE : SIMDCmpTwoVector<1, 0b01001, "cmle", AArch64cmlez>; +defm CMLT : SIMDCmpTwoVector<0, 0b01010, "cmlt", AArch64cmltz>; +defm CNT : SIMDTwoVectorB<0, 0b00, 0b00101, "cnt", ctpop>; +defm FABS : SIMDTwoVectorFP<0, 1, 0b01111, "fabs", fabs>; + +defm FCMEQ : SIMDFPCmpTwoVector<0, 1, 0b01101, "fcmeq", AArch64fcmeqz>; +defm FCMGE : SIMDFPCmpTwoVector<1, 1, 0b01100, "fcmge", AArch64fcmgez>; +defm FCMGT : SIMDFPCmpTwoVector<0, 1, 0b01100, "fcmgt", AArch64fcmgtz>; +defm FCMLE : SIMDFPCmpTwoVector<1, 1, 0b01101, "fcmle", AArch64fcmlez>; +defm FCMLT : SIMDFPCmpTwoVector<0, 1, 0b01110, "fcmlt", AArch64fcmltz>; +defm FCVTAS : SIMDTwoVectorFPToInt<0,0,0b11100, "fcvtas",int_aarch64_neon_fcvtas>; +defm FCVTAU : SIMDTwoVectorFPToInt<1,0,0b11100, "fcvtau",int_aarch64_neon_fcvtau>; +defm FCVTL : SIMDFPWidenTwoVector<0, 0, 0b10111, "fcvtl">; +def : Pat<(v4f32 (int_aarch64_neon_vcvthf2fp (v4i16 V64:$Rn))), + (FCVTLv4i16 V64:$Rn)>; +def : Pat<(v4f32 (int_aarch64_neon_vcvthf2fp (extract_subvector (v8i16 V128:$Rn), + (i64 4)))), + (FCVTLv8i16 V128:$Rn)>; +def : Pat<(v2f64 (fextend (v2f32 V64:$Rn))), (FCVTLv2i32 V64:$Rn)>; +def : Pat<(v2f64 (fextend (v2f32 (extract_subvector (v4f32 V128:$Rn), + (i64 2))))), + (FCVTLv4i32 V128:$Rn)>; + +def : Pat<(v4f32 (fextend (v4f16 V64:$Rn))), (FCVTLv4i16 V64:$Rn)>; +def : Pat<(v4f32 (fextend (v4f16 (extract_subvector (v8f16 V128:$Rn), + (i64 4))))), + (FCVTLv8i16 V128:$Rn)>; + +defm FCVTMS : SIMDTwoVectorFPToInt<0,0,0b11011, "fcvtms",int_aarch64_neon_fcvtms>; +defm FCVTMU : SIMDTwoVectorFPToInt<1,0,0b11011, "fcvtmu",int_aarch64_neon_fcvtmu>; +defm FCVTNS : SIMDTwoVectorFPToInt<0,0,0b11010, "fcvtns",int_aarch64_neon_fcvtns>; +defm FCVTNU : SIMDTwoVectorFPToInt<1,0,0b11010, "fcvtnu",int_aarch64_neon_fcvtnu>; +defm FCVTN : SIMDFPNarrowTwoVector<0, 0, 0b10110, "fcvtn">; +def : Pat<(v4i16 (int_aarch64_neon_vcvtfp2hf (v4f32 V128:$Rn))), + (FCVTNv4i16 V128:$Rn)>; +def : Pat<(concat_vectors V64:$Rd, + (v4i16 (int_aarch64_neon_vcvtfp2hf (v4f32 V128:$Rn)))), + (FCVTNv8i16 (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), V128:$Rn)>; +def : Pat<(v2f32 (fround (v2f64 V128:$Rn))), (FCVTNv2i32 V128:$Rn)>; +def : Pat<(v4f16 (fround (v4f32 V128:$Rn))), (FCVTNv4i16 V128:$Rn)>; +def : Pat<(concat_vectors V64:$Rd, (v2f32 (fround (v2f64 V128:$Rn)))), + (FCVTNv4i32 (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), V128:$Rn)>; +defm FCVTPS : SIMDTwoVectorFPToInt<0,1,0b11010, "fcvtps",int_aarch64_neon_fcvtps>; +defm FCVTPU : SIMDTwoVectorFPToInt<1,1,0b11010, "fcvtpu",int_aarch64_neon_fcvtpu>; +defm FCVTXN : SIMDFPInexactCvtTwoVector<1, 0, 0b10110, "fcvtxn", + int_aarch64_neon_fcvtxn>; +defm FCVTZS : SIMDTwoVectorFPToInt<0, 1, 0b11011, "fcvtzs", fp_to_sint>; +defm FCVTZU : SIMDTwoVectorFPToInt<1, 1, 0b11011, "fcvtzu", fp_to_uint>; +let isCodeGenOnly = 1 in { +defm FCVTZS_Int : SIMDTwoVectorFPToInt<0, 1, 0b11011, "fcvtzs", + int_aarch64_neon_fcvtzs>; +defm FCVTZU_Int : SIMDTwoVectorFPToInt<1, 1, 0b11011, "fcvtzu", + int_aarch64_neon_fcvtzu>; +} +defm FNEG : SIMDTwoVectorFP<1, 1, 0b01111, "fneg", fneg>; +defm FRECPE : SIMDTwoVectorFP<0, 1, 0b11101, "frecpe", int_aarch64_neon_frecpe>; +defm FRINTA : SIMDTwoVectorFP<1, 0, 0b11000, "frinta", frnd>; +defm FRINTI : SIMDTwoVectorFP<1, 1, 0b11001, "frinti", fnearbyint>; +defm FRINTM : SIMDTwoVectorFP<0, 0, 0b11001, "frintm", ffloor>; +defm FRINTN : SIMDTwoVectorFP<0, 0, 0b11000, "frintn", int_aarch64_neon_frintn>; +defm FRINTP : SIMDTwoVectorFP<0, 1, 0b11000, "frintp", fceil>; +defm FRINTX : SIMDTwoVectorFP<1, 0, 0b11001, "frintx", frint>; +defm FRINTZ : SIMDTwoVectorFP<0, 1, 0b11001, "frintz", ftrunc>; +defm FRSQRTE: SIMDTwoVectorFP<1, 1, 0b11101, "frsqrte", int_aarch64_neon_frsqrte>; +defm FSQRT : SIMDTwoVectorFP<1, 1, 0b11111, "fsqrt", fsqrt>; +defm NEG : SIMDTwoVectorBHSD<1, 0b01011, "neg", + UnOpFrag<(sub immAllZerosV, node:$LHS)> >; +defm NOT : SIMDTwoVectorB<1, 0b00, 0b00101, "not", vnot>; +// Aliases for MVN -> NOT. +def : InstAlias<"mvn{ $Vd.8b, $Vn.8b|.8b $Vd, $Vn}", + (NOTv8i8 V64:$Vd, V64:$Vn)>; +def : InstAlias<"mvn{ $Vd.16b, $Vn.16b|.16b $Vd, $Vn}", + (NOTv16i8 V128:$Vd, V128:$Vn)>; + +def : Pat<(AArch64neg (v8i8 V64:$Rn)), (NEGv8i8 V64:$Rn)>; +def : Pat<(AArch64neg (v16i8 V128:$Rn)), (NEGv16i8 V128:$Rn)>; +def : Pat<(AArch64neg (v4i16 V64:$Rn)), (NEGv4i16 V64:$Rn)>; +def : Pat<(AArch64neg (v8i16 V128:$Rn)), (NEGv8i16 V128:$Rn)>; +def : Pat<(AArch64neg (v2i32 V64:$Rn)), (NEGv2i32 V64:$Rn)>; +def : Pat<(AArch64neg (v4i32 V128:$Rn)), (NEGv4i32 V128:$Rn)>; +def : Pat<(AArch64neg (v2i64 V128:$Rn)), (NEGv2i64 V128:$Rn)>; + +def : Pat<(AArch64not (v8i8 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(AArch64not (v16i8 V128:$Rn)), (NOTv16i8 V128:$Rn)>; +def : Pat<(AArch64not (v4i16 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(AArch64not (v8i16 V128:$Rn)), (NOTv16i8 V128:$Rn)>; +def : Pat<(AArch64not (v2i32 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(AArch64not (v1i64 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(AArch64not (v4i32 V128:$Rn)), (NOTv16i8 V128:$Rn)>; +def : Pat<(AArch64not (v2i64 V128:$Rn)), (NOTv16i8 V128:$Rn)>; + +def : Pat<(vnot (v4i16 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(vnot (v8i16 V128:$Rn)), (NOTv16i8 V128:$Rn)>; +def : Pat<(vnot (v2i32 V64:$Rn)), (NOTv8i8 V64:$Rn)>; +def : Pat<(vnot (v4i32 V128:$Rn)), (NOTv16i8 V128:$Rn)>; +def : Pat<(vnot (v2i64 V128:$Rn)), (NOTv16i8 V128:$Rn)>; + +defm RBIT : SIMDTwoVectorB<1, 0b01, 0b00101, "rbit", int_aarch64_neon_rbit>; +defm REV16 : SIMDTwoVectorB<0, 0b00, 0b00001, "rev16", AArch64rev16>; +defm REV32 : SIMDTwoVectorBH<1, 0b00000, "rev32", AArch64rev32>; +defm REV64 : SIMDTwoVectorBHS<0, 0b00000, "rev64", AArch64rev64>; +defm SADALP : SIMDLongTwoVectorTied<0, 0b00110, "sadalp", + BinOpFrag<(add node:$LHS, (int_aarch64_neon_saddlp node:$RHS))> >; +defm SADDLP : SIMDLongTwoVector<0, 0b00010, "saddlp", int_aarch64_neon_saddlp>; +defm SCVTF : SIMDTwoVectorIntToFP<0, 0, 0b11101, "scvtf", sint_to_fp>; +defm SHLL : SIMDVectorLShiftLongBySizeBHS; +defm SQABS : SIMDTwoVectorBHSD<0, 0b00111, "sqabs", int_aarch64_neon_sqabs>; +defm SQNEG : SIMDTwoVectorBHSD<1, 0b00111, "sqneg", int_aarch64_neon_sqneg>; +defm SQXTN : SIMDMixedTwoVector<0, 0b10100, "sqxtn", int_aarch64_neon_sqxtn>; +defm SQXTUN : SIMDMixedTwoVector<1, 0b10010, "sqxtun", int_aarch64_neon_sqxtun>; +defm SUQADD : SIMDTwoVectorBHSDTied<0, 0b00011, "suqadd",int_aarch64_neon_suqadd>; +defm UADALP : SIMDLongTwoVectorTied<1, 0b00110, "uadalp", + BinOpFrag<(add node:$LHS, (int_aarch64_neon_uaddlp node:$RHS))> >; +defm UADDLP : SIMDLongTwoVector<1, 0b00010, "uaddlp", + int_aarch64_neon_uaddlp>; +defm UCVTF : SIMDTwoVectorIntToFP<1, 0, 0b11101, "ucvtf", uint_to_fp>; +defm UQXTN : SIMDMixedTwoVector<1, 0b10100, "uqxtn", int_aarch64_neon_uqxtn>; +defm URECPE : SIMDTwoVectorS<0, 1, 0b11100, "urecpe", int_aarch64_neon_urecpe>; +defm URSQRTE: SIMDTwoVectorS<1, 1, 0b11100, "ursqrte", int_aarch64_neon_ursqrte>; +defm USQADD : SIMDTwoVectorBHSDTied<1, 0b00011, "usqadd",int_aarch64_neon_usqadd>; +defm XTN : SIMDMixedTwoVector<0, 0b10010, "xtn", trunc>; + +def : Pat<(v4f16 (AArch64rev32 V64:$Rn)), (REV32v4i16 V64:$Rn)>; +def : Pat<(v4f16 (AArch64rev64 V64:$Rn)), (REV64v4i16 V64:$Rn)>; +def : Pat<(v8f16 (AArch64rev32 V128:$Rn)), (REV32v8i16 V128:$Rn)>; +def : Pat<(v8f16 (AArch64rev64 V128:$Rn)), (REV64v8i16 V128:$Rn)>; +def : Pat<(v2f32 (AArch64rev64 V64:$Rn)), (REV64v2i32 V64:$Rn)>; +def : Pat<(v4f32 (AArch64rev64 V128:$Rn)), (REV64v4i32 V128:$Rn)>; + +// Patterns for vector long shift (by element width). These need to match all +// three of zext, sext and anyext so it's easier to pull the patterns out of the +// definition. +multiclass SIMDVectorLShiftLongBySizeBHSPats { + def : Pat<(AArch64vshl (v8i16 (ext (v8i8 V64:$Rn))), (i32 8)), + (SHLLv8i8 V64:$Rn)>; + def : Pat<(AArch64vshl (v8i16 (ext (extract_high_v16i8 V128:$Rn))), (i32 8)), + (SHLLv16i8 V128:$Rn)>; + def : Pat<(AArch64vshl (v4i32 (ext (v4i16 V64:$Rn))), (i32 16)), + (SHLLv4i16 V64:$Rn)>; + def : Pat<(AArch64vshl (v4i32 (ext (extract_high_v8i16 V128:$Rn))), (i32 16)), + (SHLLv8i16 V128:$Rn)>; + def : Pat<(AArch64vshl (v2i64 (ext (v2i32 V64:$Rn))), (i32 32)), + (SHLLv2i32 V64:$Rn)>; + def : Pat<(AArch64vshl (v2i64 (ext (extract_high_v4i32 V128:$Rn))), (i32 32)), + (SHLLv4i32 V128:$Rn)>; +} + +defm : SIMDVectorLShiftLongBySizeBHSPats; +defm : SIMDVectorLShiftLongBySizeBHSPats; +defm : SIMDVectorLShiftLongBySizeBHSPats; //===----------------------------------------------------------------------===// -// Floating-point compare instructions +// Advanced SIMD three vector instructions. //===----------------------------------------------------------------------===// -// Contains: FCMP, FCMPE - -def fpzero_asmoperand : AsmOperandClass { - let Name = "FPZero"; - let ParserMethod = "ParseFPImmOperand"; - let DiagnosticType = "FPZero"; -} - -def fpz32 : Operand, - ComplexPattern { - let ParserMatchClass = fpzero_asmoperand; - let PrintMethod = "printFPZeroOperand"; -} - -def fpz64 : Operand, - ComplexPattern { - let ParserMatchClass = fpzero_asmoperand; - let PrintMethod = "printFPZeroOperand"; -} - -multiclass A64I_fpcmpSignal type, bit imm, dag ins, string asmop2, - dag pattern> { - def _quiet : A64I_fpcmp<0b0, 0b0, type, 0b00, {0b0, imm, 0b0, 0b0, 0b0}, - (outs), ins, !strconcat("fcmp\t$Rn, ", asmop2), - [pattern], NoItinerary> { - let Defs = [NZCV]; - } - - def _sig : A64I_fpcmp<0b0, 0b0, type, 0b00, {0b1, imm, 0b0, 0b0, 0b0}, - (outs), ins, !strconcat("fcmpe\t$Rn, ", asmop2), - [], NoItinerary> { - let Defs = [NZCV]; - } -} - -defm FCMPss : A64I_fpcmpSignal<0b00, 0b0, (ins FPR32:$Rn, FPR32:$Rm), "$Rm", - (set NZCV, (A64cmp (f32 FPR32:$Rn), FPR32:$Rm))>; -defm FCMPdd : A64I_fpcmpSignal<0b01, 0b0, (ins FPR64:$Rn, FPR64:$Rm), "$Rm", - (set NZCV, (A64cmp (f64 FPR64:$Rn), FPR64:$Rm))>; - -// What would be Rm should be written as 0, but anything is valid for -// disassembly so we can't set the bits -let PostEncoderMethod = "fixFCMPImm" in { - defm FCMPsi : A64I_fpcmpSignal<0b00, 0b1, (ins FPR32:$Rn, fpz32:$Imm), "$Imm", - (set NZCV, (A64cmp (f32 FPR32:$Rn), fpz32:$Imm))>; - - defm FCMPdi : A64I_fpcmpSignal<0b01, 0b1, (ins FPR64:$Rn, fpz64:$Imm), "$Imm", - (set NZCV, (A64cmp (f64 FPR64:$Rn), fpz64:$Imm))>; -} +defm ADD : SIMDThreeSameVector<0, 0b10000, "add", add>; +defm ADDP : SIMDThreeSameVector<0, 0b10111, "addp", int_aarch64_neon_addp>; +defm CMEQ : SIMDThreeSameVector<1, 0b10001, "cmeq", AArch64cmeq>; +defm CMGE : SIMDThreeSameVector<0, 0b00111, "cmge", AArch64cmge>; +defm CMGT : SIMDThreeSameVector<0, 0b00110, "cmgt", AArch64cmgt>; +defm CMHI : SIMDThreeSameVector<1, 0b00110, "cmhi", AArch64cmhi>; +defm CMHS : SIMDThreeSameVector<1, 0b00111, "cmhs", AArch64cmhs>; +defm CMTST : SIMDThreeSameVector<0, 0b10001, "cmtst", AArch64cmtst>; +defm FABD : SIMDThreeSameVectorFP<1,1,0b010,"fabd", int_aarch64_neon_fabd>; +defm FACGE : SIMDThreeSameVectorFPCmp<1,0,0b101,"facge",int_aarch64_neon_facge>; +defm FACGT : SIMDThreeSameVectorFPCmp<1,1,0b101,"facgt",int_aarch64_neon_facgt>; +defm FADDP : SIMDThreeSameVectorFP<1,0,0b010,"faddp",int_aarch64_neon_addp>; +defm FADD : SIMDThreeSameVectorFP<0,0,0b010,"fadd", fadd>; +defm FCMEQ : SIMDThreeSameVectorFPCmp<0, 0, 0b100, "fcmeq", AArch64fcmeq>; +defm FCMGE : SIMDThreeSameVectorFPCmp<1, 0, 0b100, "fcmge", AArch64fcmge>; +defm FCMGT : SIMDThreeSameVectorFPCmp<1, 1, 0b100, "fcmgt", AArch64fcmgt>; +defm FDIV : SIMDThreeSameVectorFP<1,0,0b111,"fdiv", fdiv>; +defm FMAXNMP : SIMDThreeSameVectorFP<1,0,0b000,"fmaxnmp", int_aarch64_neon_fmaxnmp>; +defm FMAXNM : SIMDThreeSameVectorFP<0,0,0b000,"fmaxnm", fmaxnum>; +defm FMAXP : SIMDThreeSameVectorFP<1,0,0b110,"fmaxp", int_aarch64_neon_fmaxp>; +defm FMAX : SIMDThreeSameVectorFP<0,0,0b110,"fmax", fmaxnan>; +defm FMINNMP : SIMDThreeSameVectorFP<1,1,0b000,"fminnmp", int_aarch64_neon_fminnmp>; +defm FMINNM : SIMDThreeSameVectorFP<0,1,0b000,"fminnm", fminnum>; +defm FMINP : SIMDThreeSameVectorFP<1,1,0b110,"fminp", int_aarch64_neon_fminp>; +defm FMIN : SIMDThreeSameVectorFP<0,1,0b110,"fmin", fminnan>; + +// NOTE: The operands of the PatFrag are reordered on FMLA/FMLS because the +// instruction expects the addend first, while the fma intrinsic puts it last. +defm FMLA : SIMDThreeSameVectorFPTied<0, 0, 0b001, "fmla", + TriOpFrag<(fma node:$RHS, node:$MHS, node:$LHS)> >; +defm FMLS : SIMDThreeSameVectorFPTied<0, 1, 0b001, "fmls", + TriOpFrag<(fma node:$MHS, (fneg node:$RHS), node:$LHS)> >; + +// The following def pats catch the case where the LHS of an FMA is negated. +// The TriOpFrag above catches the case where the middle operand is negated. +def : Pat<(v2f32 (fma (fneg V64:$Rn), V64:$Rm, V64:$Rd)), + (FMLSv2f32 V64:$Rd, V64:$Rn, V64:$Rm)>; + +def : Pat<(v4f32 (fma (fneg V128:$Rn), V128:$Rm, V128:$Rd)), + (FMLSv4f32 V128:$Rd, V128:$Rn, V128:$Rm)>; + +def : Pat<(v2f64 (fma (fneg V128:$Rn), V128:$Rm, V128:$Rd)), + (FMLSv2f64 V128:$Rd, V128:$Rn, V128:$Rm)>; + +defm FMULX : SIMDThreeSameVectorFP<0,0,0b011,"fmulx", int_aarch64_neon_fmulx>; +defm FMUL : SIMDThreeSameVectorFP<1,0,0b011,"fmul", fmul>; +defm FRECPS : SIMDThreeSameVectorFP<0,0,0b111,"frecps", int_aarch64_neon_frecps>; +defm FRSQRTS : SIMDThreeSameVectorFP<0,1,0b111,"frsqrts", int_aarch64_neon_frsqrts>; +defm FSUB : SIMDThreeSameVectorFP<0,1,0b010,"fsub", fsub>; +defm MLA : SIMDThreeSameVectorBHSTied<0, 0b10010, "mla", + TriOpFrag<(add node:$LHS, (mul node:$MHS, node:$RHS))> >; +defm MLS : SIMDThreeSameVectorBHSTied<1, 0b10010, "mls", + TriOpFrag<(sub node:$LHS, (mul node:$MHS, node:$RHS))> >; +defm MUL : SIMDThreeSameVectorBHS<0, 0b10011, "mul", mul>; +defm PMUL : SIMDThreeSameVectorB<1, 0b10011, "pmul", int_aarch64_neon_pmul>; +defm SABA : SIMDThreeSameVectorBHSTied<0, 0b01111, "saba", + TriOpFrag<(add node:$LHS, (sabsdiff node:$MHS, node:$RHS))> >; +defm SABD : SIMDThreeSameVectorBHS<0,0b01110,"sabd", sabsdiff>; +defm SHADD : SIMDThreeSameVectorBHS<0,0b00000,"shadd", int_aarch64_neon_shadd>; +defm SHSUB : SIMDThreeSameVectorBHS<0,0b00100,"shsub", int_aarch64_neon_shsub>; +defm SMAXP : SIMDThreeSameVectorBHS<0,0b10100,"smaxp", int_aarch64_neon_smaxp>; +defm SMAX : SIMDThreeSameVectorBHS<0,0b01100,"smax", smax>; +defm SMINP : SIMDThreeSameVectorBHS<0,0b10101,"sminp", int_aarch64_neon_sminp>; +defm SMIN : SIMDThreeSameVectorBHS<0,0b01101,"smin", smin>; +defm SQADD : SIMDThreeSameVector<0,0b00001,"sqadd", int_aarch64_neon_sqadd>; +defm SQDMULH : SIMDThreeSameVectorHS<0,0b10110,"sqdmulh",int_aarch64_neon_sqdmulh>; +defm SQRDMULH : SIMDThreeSameVectorHS<1,0b10110,"sqrdmulh",int_aarch64_neon_sqrdmulh>; +defm SQRSHL : SIMDThreeSameVector<0,0b01011,"sqrshl", int_aarch64_neon_sqrshl>; +defm SQSHL : SIMDThreeSameVector<0,0b01001,"sqshl", int_aarch64_neon_sqshl>; +defm SQSUB : SIMDThreeSameVector<0,0b00101,"sqsub", int_aarch64_neon_sqsub>; +defm SRHADD : SIMDThreeSameVectorBHS<0,0b00010,"srhadd",int_aarch64_neon_srhadd>; +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:$LHS, (uabsdiff node:$MHS, node:$RHS))> >; +defm UABD : SIMDThreeSameVectorBHS<1,0b01110,"uabd", uabsdiff>; +defm UHADD : SIMDThreeSameVectorBHS<1,0b00000,"uhadd", int_aarch64_neon_uhadd>; +defm UHSUB : SIMDThreeSameVectorBHS<1,0b00100,"uhsub", int_aarch64_neon_uhsub>; +defm UMAXP : SIMDThreeSameVectorBHS<1,0b10100,"umaxp", int_aarch64_neon_umaxp>; +defm UMAX : SIMDThreeSameVectorBHS<1,0b01100,"umax", umax>; +defm UMINP : SIMDThreeSameVectorBHS<1,0b10101,"uminp", int_aarch64_neon_uminp>; +defm UMIN : SIMDThreeSameVectorBHS<1,0b01101,"umin", umin>; +defm UQADD : SIMDThreeSameVector<1,0b00001,"uqadd", int_aarch64_neon_uqadd>; +defm UQRSHL : SIMDThreeSameVector<1,0b01011,"uqrshl", int_aarch64_neon_uqrshl>; +defm UQSHL : SIMDThreeSameVector<1,0b01001,"uqshl", int_aarch64_neon_uqshl>; +defm UQSUB : SIMDThreeSameVector<1,0b00101,"uqsub", int_aarch64_neon_uqsub>; +defm URHADD : SIMDThreeSameVectorBHS<1,0b00010,"urhadd", int_aarch64_neon_urhadd>; +defm URSHL : SIMDThreeSameVector<1,0b01010,"urshl", int_aarch64_neon_urshl>; +defm USHL : SIMDThreeSameVector<1,0b01000,"ushl", int_aarch64_neon_ushl>; +defm SQRDMLAH : SIMDThreeSameVectorSQRDMLxHTiedHS<1,0b10000,"sqrdmlah", + int_aarch64_neon_sqadd>; +defm SQRDMLSH : SIMDThreeSameVectorSQRDMLxHTiedHS<1,0b10001,"sqrdmlsh", + int_aarch64_neon_sqsub>; + +defm AND : SIMDLogicalThreeVector<0, 0b00, "and", and>; +defm BIC : SIMDLogicalThreeVector<0, 0b01, "bic", + BinOpFrag<(and node:$LHS, (vnot node:$RHS))> >; +defm BIF : SIMDLogicalThreeVector<1, 0b11, "bif">; +defm BIT : SIMDLogicalThreeVectorTied<1, 0b10, "bit", AArch64bit>; +defm BSL : SIMDLogicalThreeVectorTied<1, 0b01, "bsl", + TriOpFrag<(or (and node:$LHS, node:$MHS), (and (vnot node:$LHS), node:$RHS))>>; +defm EOR : SIMDLogicalThreeVector<1, 0b00, "eor", xor>; +defm ORN : SIMDLogicalThreeVector<0, 0b11, "orn", + BinOpFrag<(or node:$LHS, (vnot node:$RHS))> >; +defm ORR : SIMDLogicalThreeVector<0, 0b10, "orr", or>; + + +def : Pat<(AArch64bsl (v8i8 V64:$Rd), V64:$Rn, V64:$Rm), + (BSLv8i8 V64:$Rd, V64:$Rn, V64:$Rm)>; +def : Pat<(AArch64bsl (v4i16 V64:$Rd), V64:$Rn, V64:$Rm), + (BSLv8i8 V64:$Rd, V64:$Rn, V64:$Rm)>; +def : Pat<(AArch64bsl (v2i32 V64:$Rd), V64:$Rn, V64:$Rm), + (BSLv8i8 V64:$Rd, V64:$Rn, V64:$Rm)>; +def : Pat<(AArch64bsl (v1i64 V64:$Rd), V64:$Rn, V64:$Rm), + (BSLv8i8 V64:$Rd, V64:$Rn, V64:$Rm)>; + +def : Pat<(AArch64bsl (v16i8 V128:$Rd), V128:$Rn, V128:$Rm), + (BSLv16i8 V128:$Rd, V128:$Rn, V128:$Rm)>; +def : Pat<(AArch64bsl (v8i16 V128:$Rd), V128:$Rn, V128:$Rm), + (BSLv16i8 V128:$Rd, V128:$Rn, V128:$Rm)>; +def : Pat<(AArch64bsl (v4i32 V128:$Rd), V128:$Rn, V128:$Rm), + (BSLv16i8 V128:$Rd, V128:$Rn, V128:$Rm)>; +def : Pat<(AArch64bsl (v2i64 V128:$Rd), V128:$Rn, V128:$Rm), + (BSLv16i8 V128:$Rd, V128:$Rn, V128:$Rm)>; + +def : InstAlias<"mov{\t$dst.16b, $src.16b|.16b\t$dst, $src}", + (ORRv16i8 V128:$dst, V128:$src, V128:$src), 1>; +def : InstAlias<"mov{\t$dst.8h, $src.8h|.8h\t$dst, $src}", + (ORRv16i8 V128:$dst, V128:$src, V128:$src), 0>; +def : InstAlias<"mov{\t$dst.4s, $src.4s|.4s\t$dst, $src}", + (ORRv16i8 V128:$dst, V128:$src, V128:$src), 0>; +def : InstAlias<"mov{\t$dst.2d, $src.2d|.2d\t$dst, $src}", + (ORRv16i8 V128:$dst, V128:$src, V128:$src), 0>; + +def : InstAlias<"mov{\t$dst.8b, $src.8b|.8b\t$dst, $src}", + (ORRv8i8 V64:$dst, V64:$src, V64:$src), 1>; +def : InstAlias<"mov{\t$dst.4h, $src.4h|.4h\t$dst, $src}", + (ORRv8i8 V64:$dst, V64:$src, V64:$src), 0>; +def : InstAlias<"mov{\t$dst.2s, $src.2s|.2s\t$dst, $src}", + (ORRv8i8 V64:$dst, V64:$src, V64:$src), 0>; +def : InstAlias<"mov{\t$dst.1d, $src.1d|.1d\t$dst, $src}", + (ORRv8i8 V64:$dst, V64:$src, V64:$src), 0>; + +def : InstAlias<"{cmls\t$dst.8b, $src1.8b, $src2.8b" # + "|cmls.8b\t$dst, $src1, $src2}", + (CMHSv8i8 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmls\t$dst.16b, $src1.16b, $src2.16b" # + "|cmls.16b\t$dst, $src1, $src2}", + (CMHSv16i8 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmls\t$dst.4h, $src1.4h, $src2.4h" # + "|cmls.4h\t$dst, $src1, $src2}", + (CMHSv4i16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmls\t$dst.8h, $src1.8h, $src2.8h" # + "|cmls.8h\t$dst, $src1, $src2}", + (CMHSv8i16 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmls\t$dst.2s, $src1.2s, $src2.2s" # + "|cmls.2s\t$dst, $src1, $src2}", + (CMHSv2i32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmls\t$dst.4s, $src1.4s, $src2.4s" # + "|cmls.4s\t$dst, $src1, $src2}", + (CMHSv4i32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmls\t$dst.2d, $src1.2d, $src2.2d" # + "|cmls.2d\t$dst, $src1, $src2}", + (CMHSv2i64 V128:$dst, V128:$src2, V128:$src1), 0>; + +def : InstAlias<"{cmlo\t$dst.8b, $src1.8b, $src2.8b" # + "|cmlo.8b\t$dst, $src1, $src2}", + (CMHIv8i8 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.16b, $src1.16b, $src2.16b" # + "|cmlo.16b\t$dst, $src1, $src2}", + (CMHIv16i8 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.4h, $src1.4h, $src2.4h" # + "|cmlo.4h\t$dst, $src1, $src2}", + (CMHIv4i16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.8h, $src1.8h, $src2.8h" # + "|cmlo.8h\t$dst, $src1, $src2}", + (CMHIv8i16 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.2s, $src1.2s, $src2.2s" # + "|cmlo.2s\t$dst, $src1, $src2}", + (CMHIv2i32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.4s, $src1.4s, $src2.4s" # + "|cmlo.4s\t$dst, $src1, $src2}", + (CMHIv4i32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlo\t$dst.2d, $src1.2d, $src2.2d" # + "|cmlo.2d\t$dst, $src1, $src2}", + (CMHIv2i64 V128:$dst, V128:$src2, V128:$src1), 0>; + +def : InstAlias<"{cmle\t$dst.8b, $src1.8b, $src2.8b" # + "|cmle.8b\t$dst, $src1, $src2}", + (CMGEv8i8 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmle\t$dst.16b, $src1.16b, $src2.16b" # + "|cmle.16b\t$dst, $src1, $src2}", + (CMGEv16i8 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmle\t$dst.4h, $src1.4h, $src2.4h" # + "|cmle.4h\t$dst, $src1, $src2}", + (CMGEv4i16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmle\t$dst.8h, $src1.8h, $src2.8h" # + "|cmle.8h\t$dst, $src1, $src2}", + (CMGEv8i16 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmle\t$dst.2s, $src1.2s, $src2.2s" # + "|cmle.2s\t$dst, $src1, $src2}", + (CMGEv2i32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmle\t$dst.4s, $src1.4s, $src2.4s" # + "|cmle.4s\t$dst, $src1, $src2}", + (CMGEv4i32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmle\t$dst.2d, $src1.2d, $src2.2d" # + "|cmle.2d\t$dst, $src1, $src2}", + (CMGEv2i64 V128:$dst, V128:$src2, V128:$src1), 0>; + +def : InstAlias<"{cmlt\t$dst.8b, $src1.8b, $src2.8b" # + "|cmlt.8b\t$dst, $src1, $src2}", + (CMGTv8i8 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.16b, $src1.16b, $src2.16b" # + "|cmlt.16b\t$dst, $src1, $src2}", + (CMGTv16i8 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.4h, $src1.4h, $src2.4h" # + "|cmlt.4h\t$dst, $src1, $src2}", + (CMGTv4i16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.8h, $src1.8h, $src2.8h" # + "|cmlt.8h\t$dst, $src1, $src2}", + (CMGTv8i16 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.2s, $src1.2s, $src2.2s" # + "|cmlt.2s\t$dst, $src1, $src2}", + (CMGTv2i32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.4s, $src1.4s, $src2.4s" # + "|cmlt.4s\t$dst, $src1, $src2}", + (CMGTv4i32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{cmlt\t$dst.2d, $src1.2d, $src2.2d" # + "|cmlt.2d\t$dst, $src1, $src2}", + (CMGTv2i64 V128:$dst, V128:$src2, V128:$src1), 0>; + +let Predicates = [HasNEON, HasFullFP16] in { +def : InstAlias<"{fcmle\t$dst.4h, $src1.4h, $src2.4h" # + "|fcmle.4h\t$dst, $src1, $src2}", + (FCMGEv4f16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{fcmle\t$dst.8h, $src1.8h, $src2.8h" # + "|fcmle.8h\t$dst, $src1, $src2}", + (FCMGEv8f16 V128:$dst, V128:$src2, V128:$src1), 0>; +} +def : InstAlias<"{fcmle\t$dst.2s, $src1.2s, $src2.2s" # + "|fcmle.2s\t$dst, $src1, $src2}", + (FCMGEv2f32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{fcmle\t$dst.4s, $src1.4s, $src2.4s" # + "|fcmle.4s\t$dst, $src1, $src2}", + (FCMGEv4f32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{fcmle\t$dst.2d, $src1.2d, $src2.2d" # + "|fcmle.2d\t$dst, $src1, $src2}", + (FCMGEv2f64 V128:$dst, V128:$src2, V128:$src1), 0>; + +let Predicates = [HasNEON, HasFullFP16] in { +def : InstAlias<"{fcmlt\t$dst.4h, $src1.4h, $src2.4h" # + "|fcmlt.4h\t$dst, $src1, $src2}", + (FCMGTv4f16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{fcmlt\t$dst.8h, $src1.8h, $src2.8h" # + "|fcmlt.8h\t$dst, $src1, $src2}", + (FCMGTv8f16 V128:$dst, V128:$src2, V128:$src1), 0>; +} +def : InstAlias<"{fcmlt\t$dst.2s, $src1.2s, $src2.2s" # + "|fcmlt.2s\t$dst, $src1, $src2}", + (FCMGTv2f32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{fcmlt\t$dst.4s, $src1.4s, $src2.4s" # + "|fcmlt.4s\t$dst, $src1, $src2}", + (FCMGTv4f32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{fcmlt\t$dst.2d, $src1.2d, $src2.2d" # + "|fcmlt.2d\t$dst, $src1, $src2}", + (FCMGTv2f64 V128:$dst, V128:$src2, V128:$src1), 0>; + +let Predicates = [HasNEON, HasFullFP16] in { +def : InstAlias<"{facle\t$dst.4h, $src1.4h, $src2.4h" # + "|facle.4h\t$dst, $src1, $src2}", + (FACGEv4f16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{facle\t$dst.8h, $src1.8h, $src2.8h" # + "|facle.8h\t$dst, $src1, $src2}", + (FACGEv8f16 V128:$dst, V128:$src2, V128:$src1), 0>; +} +def : InstAlias<"{facle\t$dst.2s, $src1.2s, $src2.2s" # + "|facle.2s\t$dst, $src1, $src2}", + (FACGEv2f32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{facle\t$dst.4s, $src1.4s, $src2.4s" # + "|facle.4s\t$dst, $src1, $src2}", + (FACGEv4f32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{facle\t$dst.2d, $src1.2d, $src2.2d" # + "|facle.2d\t$dst, $src1, $src2}", + (FACGEv2f64 V128:$dst, V128:$src2, V128:$src1), 0>; + +let Predicates = [HasNEON, HasFullFP16] in { +def : InstAlias<"{faclt\t$dst.4h, $src1.4h, $src2.4h" # + "|faclt.4h\t$dst, $src1, $src2}", + (FACGTv4f16 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{faclt\t$dst.8h, $src1.8h, $src2.8h" # + "|faclt.8h\t$dst, $src1, $src2}", + (FACGTv8f16 V128:$dst, V128:$src2, V128:$src1), 0>; +} +def : InstAlias<"{faclt\t$dst.2s, $src1.2s, $src2.2s" # + "|faclt.2s\t$dst, $src1, $src2}", + (FACGTv2f32 V64:$dst, V64:$src2, V64:$src1), 0>; +def : InstAlias<"{faclt\t$dst.4s, $src1.4s, $src2.4s" # + "|faclt.4s\t$dst, $src1, $src2}", + (FACGTv4f32 V128:$dst, V128:$src2, V128:$src1), 0>; +def : InstAlias<"{faclt\t$dst.2d, $src1.2d, $src2.2d" # + "|faclt.2d\t$dst, $src1, $src2}", + (FACGTv2f64 V128:$dst, V128:$src2, V128:$src1), 0>; //===----------------------------------------------------------------------===// -// Floating-point conditional compare instructions +// Advanced SIMD three scalar instructions. //===----------------------------------------------------------------------===// -// Contains: FCCMP, FCCMPE - -class A64I_fpccmpImpl type, bit op, RegisterClass FPR, string asmop> - : A64I_fpccmp<0b0, 0b0, type, op, - (outs), - (ins FPR:$Rn, FPR:$Rm, uimm4:$NZCVImm, cond_code_op:$Cond), - !strconcat(asmop, "\t$Rn, $Rm, $NZCVImm, $Cond"), - [], NoItinerary> { - let Defs = [NZCV]; -} -def FCCMPss : A64I_fpccmpImpl<0b00, 0b0, FPR32, "fccmp">; -def FCCMPEss : A64I_fpccmpImpl<0b00, 0b1, FPR32, "fccmpe">; -def FCCMPdd : A64I_fpccmpImpl<0b01, 0b0, FPR64, "fccmp">; -def FCCMPEdd : A64I_fpccmpImpl<0b01, 0b1, FPR64, "fccmpe">; +defm ADD : SIMDThreeScalarD<0, 0b10000, "add", add>; +defm CMEQ : SIMDThreeScalarD<1, 0b10001, "cmeq", AArch64cmeq>; +defm CMGE : SIMDThreeScalarD<0, 0b00111, "cmge", AArch64cmge>; +defm CMGT : SIMDThreeScalarD<0, 0b00110, "cmgt", AArch64cmgt>; +defm CMHI : SIMDThreeScalarD<1, 0b00110, "cmhi", AArch64cmhi>; +defm CMHS : SIMDThreeScalarD<1, 0b00111, "cmhs", AArch64cmhs>; +defm CMTST : SIMDThreeScalarD<0, 0b10001, "cmtst", AArch64cmtst>; +defm FABD : SIMDFPThreeScalar<1, 1, 0b010, "fabd", int_aarch64_sisd_fabd>; +def : Pat<(v1f64 (int_aarch64_neon_fabd (v1f64 FPR64:$Rn), (v1f64 FPR64:$Rm))), + (FABD64 FPR64:$Rn, FPR64:$Rm)>; +defm FACGE : SIMDThreeScalarFPCmp<1, 0, 0b101, "facge", + int_aarch64_neon_facge>; +defm FACGT : SIMDThreeScalarFPCmp<1, 1, 0b101, "facgt", + int_aarch64_neon_facgt>; +defm FCMEQ : SIMDThreeScalarFPCmp<0, 0, 0b100, "fcmeq", AArch64fcmeq>; +defm FCMGE : SIMDThreeScalarFPCmp<1, 0, 0b100, "fcmge", AArch64fcmge>; +defm FCMGT : SIMDThreeScalarFPCmp<1, 1, 0b100, "fcmgt", AArch64fcmgt>; +defm FMULX : SIMDFPThreeScalar<0, 0, 0b011, "fmulx", int_aarch64_neon_fmulx>; +defm FRECPS : SIMDFPThreeScalar<0, 0, 0b111, "frecps", int_aarch64_neon_frecps>; +defm FRSQRTS : SIMDFPThreeScalar<0, 1, 0b111, "frsqrts", int_aarch64_neon_frsqrts>; +defm SQADD : SIMDThreeScalarBHSD<0, 0b00001, "sqadd", int_aarch64_neon_sqadd>; +defm SQDMULH : SIMDThreeScalarHS< 0, 0b10110, "sqdmulh", int_aarch64_neon_sqdmulh>; +defm SQRDMULH : SIMDThreeScalarHS< 1, 0b10110, "sqrdmulh", int_aarch64_neon_sqrdmulh>; +defm SQRSHL : SIMDThreeScalarBHSD<0, 0b01011, "sqrshl",int_aarch64_neon_sqrshl>; +defm SQSHL : SIMDThreeScalarBHSD<0, 0b01001, "sqshl", int_aarch64_neon_sqshl>; +defm SQSUB : SIMDThreeScalarBHSD<0, 0b00101, "sqsub", int_aarch64_neon_sqsub>; +defm SRSHL : SIMDThreeScalarD< 0, 0b01010, "srshl", int_aarch64_neon_srshl>; +defm SSHL : SIMDThreeScalarD< 0, 0b01000, "sshl", int_aarch64_neon_sshl>; +defm SUB : SIMDThreeScalarD< 1, 0b10000, "sub", sub>; +defm UQADD : SIMDThreeScalarBHSD<1, 0b00001, "uqadd", int_aarch64_neon_uqadd>; +defm UQRSHL : SIMDThreeScalarBHSD<1, 0b01011, "uqrshl",int_aarch64_neon_uqrshl>; +defm UQSHL : SIMDThreeScalarBHSD<1, 0b01001, "uqshl", int_aarch64_neon_uqshl>; +defm UQSUB : SIMDThreeScalarBHSD<1, 0b00101, "uqsub", int_aarch64_neon_uqsub>; +defm URSHL : SIMDThreeScalarD< 1, 0b01010, "urshl", int_aarch64_neon_urshl>; +defm USHL : SIMDThreeScalarD< 1, 0b01000, "ushl", int_aarch64_neon_ushl>; +let Predicates = [HasV8_1a] in { + defm SQRDMLAH : SIMDThreeScalarHSTied<1, 0, 0b10000, "sqrdmlah">; + defm SQRDMLSH : SIMDThreeScalarHSTied<1, 0, 0b10001, "sqrdmlsh">; + def : Pat<(i32 (int_aarch64_neon_sqadd + (i32 FPR32:$Rd), + (i32 (int_aarch64_neon_sqrdmulh (i32 FPR32:$Rn), + (i32 FPR32:$Rm))))), + (SQRDMLAHv1i32 FPR32:$Rd, FPR32:$Rn, FPR32:$Rm)>; + def : Pat<(i32 (int_aarch64_neon_sqsub + (i32 FPR32:$Rd), + (i32 (int_aarch64_neon_sqrdmulh (i32 FPR32:$Rn), + (i32 FPR32:$Rm))))), + (SQRDMLSHv1i32 FPR32:$Rd, FPR32:$Rn, FPR32:$Rm)>; +} + +def : InstAlias<"cmls $dst, $src1, $src2", + (CMHSv1i64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"cmle $dst, $src1, $src2", + (CMGEv1i64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"cmlo $dst, $src1, $src2", + (CMHIv1i64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"cmlt $dst, $src1, $src2", + (CMGTv1i64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"fcmle $dst, $src1, $src2", + (FCMGE32 FPR32:$dst, FPR32:$src2, FPR32:$src1), 0>; +def : InstAlias<"fcmle $dst, $src1, $src2", + (FCMGE64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"fcmlt $dst, $src1, $src2", + (FCMGT32 FPR32:$dst, FPR32:$src2, FPR32:$src1), 0>; +def : InstAlias<"fcmlt $dst, $src1, $src2", + (FCMGT64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"facle $dst, $src1, $src2", + (FACGE32 FPR32:$dst, FPR32:$src2, FPR32:$src1), 0>; +def : InstAlias<"facle $dst, $src1, $src2", + (FACGE64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; +def : InstAlias<"faclt $dst, $src1, $src2", + (FACGT32 FPR32:$dst, FPR32:$src2, FPR32:$src1), 0>; +def : InstAlias<"faclt $dst, $src1, $src2", + (FACGT64 FPR64:$dst, FPR64:$src2, FPR64:$src1), 0>; //===----------------------------------------------------------------------===// -// Floating-point conditional select instructions +// Advanced SIMD three scalar instructions (mixed operands). //===----------------------------------------------------------------------===// -// Contains: FCSEL - -let Uses = [NZCV] in { - def FCSELsssc : A64I_fpcondsel<0b0, 0b0, 0b00, (outs FPR32:$Rd), - (ins FPR32:$Rn, FPR32:$Rm, cond_code_op:$Cond), - "fcsel\t$Rd, $Rn, $Rm, $Cond", - [(set FPR32:$Rd, - (simple_select (f32 FPR32:$Rn), - FPR32:$Rm))], - NoItinerary>; - - - def FCSELdddc : A64I_fpcondsel<0b0, 0b0, 0b01, (outs FPR64:$Rd), - (ins FPR64:$Rn, FPR64:$Rm, cond_code_op:$Cond), - "fcsel\t$Rd, $Rn, $Rm, $Cond", - [(set FPR64:$Rd, - (simple_select (f64 FPR64:$Rn), - FPR64:$Rm))], - NoItinerary>; -} +defm SQDMULL : SIMDThreeScalarMixedHS<0, 0b11010, "sqdmull", + int_aarch64_neon_sqdmulls_scalar>; +defm SQDMLAL : SIMDThreeScalarMixedTiedHS<0, 0b10010, "sqdmlal">; +defm SQDMLSL : SIMDThreeScalarMixedTiedHS<0, 0b10110, "sqdmlsl">; + +def : Pat<(i64 (int_aarch64_neon_sqadd (i64 FPR64:$Rd), + (i64 (int_aarch64_neon_sqdmulls_scalar (i32 FPR32:$Rn), + (i32 FPR32:$Rm))))), + (SQDMLALi32 FPR64:$Rd, FPR32:$Rn, FPR32:$Rm)>; +def : Pat<(i64 (int_aarch64_neon_sqsub (i64 FPR64:$Rd), + (i64 (int_aarch64_neon_sqdmulls_scalar (i32 FPR32:$Rn), + (i32 FPR32:$Rm))))), + (SQDMLSLi32 FPR64:$Rd, FPR32:$Rn, FPR32:$Rm)>; //===----------------------------------------------------------------------===// -// Floating-point data-processing (1 source) +// Advanced SIMD two scalar instructions. //===----------------------------------------------------------------------===// -// Contains: FMOV, FABS, FNEG, FSQRT, FCVT, FRINT[NPMZAXI]. - -def FPNoUnop : PatFrag<(ops node:$val), (fneg node:$val), - [{ (void)N; return false; }]>; - -// First we do the fairly trivial bunch with uniform "OP s, s" and "OP d, d" -// syntax. Default to no pattern because most are odd enough not to have one. -multiclass A64I_fpdp1sizes opcode, string asmstr, - SDPatternOperator opnode = FPNoUnop> { - def ss : A64I_fpdp1<0b0, 0b0, 0b00, opcode, (outs FPR32:$Rd), (ins FPR32:$Rn), - !strconcat(asmstr, "\t$Rd, $Rn"), - [(set (f32 FPR32:$Rd), (opnode FPR32:$Rn))], - NoItinerary>; - - def dd : A64I_fpdp1<0b0, 0b0, 0b01, opcode, (outs FPR64:$Rd), (ins FPR64:$Rn), - !strconcat(asmstr, "\t$Rd, $Rn"), - [(set (f64 FPR64:$Rd), (opnode FPR64:$Rn))], - NoItinerary>; -} -defm FMOV : A64I_fpdp1sizes<0b000000, "fmov">; -defm FABS : A64I_fpdp1sizes<0b000001, "fabs", fabs>; -defm FNEG : A64I_fpdp1sizes<0b000010, "fneg", fneg>; -defm FSQRT : A64I_fpdp1sizes<0b000011, "fsqrt", fsqrt>; - -defm FRINTN : A64I_fpdp1sizes<0b001000, "frintn">; -defm FRINTP : A64I_fpdp1sizes<0b001001, "frintp", fceil>; -defm FRINTM : A64I_fpdp1sizes<0b001010, "frintm", ffloor>; -defm FRINTZ : A64I_fpdp1sizes<0b001011, "frintz", ftrunc>; -defm FRINTA : A64I_fpdp1sizes<0b001100, "frinta">; -defm FRINTX : A64I_fpdp1sizes<0b001110, "frintx", frint>; -defm FRINTI : A64I_fpdp1sizes<0b001111, "frinti", fnearbyint>; - -// The FCVT instrucitons have different source and destination register-types, -// but the fields are uniform everywhere a D-register (say) crops up. Package -// this information in a Record. -class FCVTRegType fld, ValueType vt> { - RegisterClass Class = rc; - ValueType VT = vt; - bit t1 = fld{1}; - bit t0 = fld{0}; -} - -def FCVT16 : FCVTRegType; -def FCVT32 : FCVTRegType; -def FCVT64 : FCVTRegType; - -class A64I_fpdp1_fcvt - : A64I_fpdp1<0b0, 0b0, {SrcReg.t1, SrcReg.t0}, - {0,0,0,1, DestReg.t1, DestReg.t0}, - (outs DestReg.Class:$Rd), (ins SrcReg.Class:$Rn), - "fcvt\t$Rd, $Rn", - [(set (DestReg.VT DestReg.Class:$Rd), - (opnode (SrcReg.VT SrcReg.Class:$Rn)))], NoItinerary>; - -def FCVTds : A64I_fpdp1_fcvt; -def FCVThs : A64I_fpdp1_fcvt; -def FCVTsd : A64I_fpdp1_fcvt; -def FCVThd : A64I_fpdp1_fcvt; -def FCVTsh : A64I_fpdp1_fcvt; -def FCVTdh : A64I_fpdp1_fcvt; - - -//===----------------------------------------------------------------------===// -// Floating-point data-processing (2 sources) instructions -//===----------------------------------------------------------------------===// -// Contains: FMUL, FDIV, FADD, FSUB, FMAX, FMIN, FMAXNM, FMINNM, FNMUL - -def FPNoBinop : PatFrag<(ops node:$lhs, node:$rhs), (fadd node:$lhs, node:$rhs), - [{ (void)N; return false; }]>; - -multiclass A64I_fpdp2sizes opcode, string asmstr, - SDPatternOperator opnode> { - def sss : A64I_fpdp2<0b0, 0b0, 0b00, opcode, - (outs FPR32:$Rd), - (ins FPR32:$Rn, FPR32:$Rm), - !strconcat(asmstr, "\t$Rd, $Rn, $Rm"), - [(set (f32 FPR32:$Rd), (opnode FPR32:$Rn, FPR32:$Rm))], - NoItinerary>; - - def ddd : A64I_fpdp2<0b0, 0b0, 0b01, opcode, - (outs FPR64:$Rd), - (ins FPR64:$Rn, FPR64:$Rm), - !strconcat(asmstr, "\t$Rd, $Rn, $Rm"), - [(set (f64 FPR64:$Rd), (opnode FPR64:$Rn, FPR64:$Rm))], - NoItinerary>; -} - -let isCommutable = 1 in { - defm FMUL : A64I_fpdp2sizes<0b0000, "fmul", fmul>; - defm FADD : A64I_fpdp2sizes<0b0010, "fadd", fadd>; - - // No patterns for these. - defm FMAX : A64I_fpdp2sizes<0b0100, "fmax", FPNoBinop>; - defm FMIN : A64I_fpdp2sizes<0b0101, "fmin", FPNoBinop>; - defm FMAXNM : A64I_fpdp2sizes<0b0110, "fmaxnm", FPNoBinop>; - defm FMINNM : A64I_fpdp2sizes<0b0111, "fminnm", FPNoBinop>; - - defm FNMUL : A64I_fpdp2sizes<0b1000, "fnmul", - PatFrag<(ops node:$lhs, node:$rhs), - (fneg (fmul node:$lhs, node:$rhs))> >; -} - -defm FDIV : A64I_fpdp2sizes<0b0001, "fdiv", fdiv>; -defm FSUB : A64I_fpdp2sizes<0b0011, "fsub", fsub>; - -//===----------------------------------------------------------------------===// -// Floating-point data-processing (3 sources) instructions -//===----------------------------------------------------------------------===// -// Contains: FMADD, FMSUB, FNMADD, FNMSUB - -def fmsub : PatFrag<(ops node:$Rn, node:$Rm, node:$Ra), - (fma (fneg node:$Rn), node:$Rm, node:$Ra)>; -def fnmadd : PatFrag<(ops node:$Rn, node:$Rm, node:$Ra), - (fma node:$Rn, node:$Rm, (fneg node:$Ra))>; -def fnmsub : PatFrag<(ops node:$Rn, node:$Rm, node:$Ra), - (fma (fneg node:$Rn), node:$Rm, (fneg node:$Ra))>; - -class A64I_fpdp3Impl type, bit o1, bit o0, SDPatternOperator fmakind> - : A64I_fpdp3<0b0, 0b0, type, o1, o0, (outs FPR:$Rd), - (ins FPR:$Rn, FPR:$Rm, FPR:$Ra), - !strconcat(asmop,"\t$Rd, $Rn, $Rm, $Ra"), - [(set FPR:$Rd, (fmakind (VT FPR:$Rn), FPR:$Rm, FPR:$Ra))], - NoItinerary>; - -def FMADDssss : A64I_fpdp3Impl<"fmadd", FPR32, f32, 0b00, 0b0, 0b0, fma>; -def FMSUBssss : A64I_fpdp3Impl<"fmsub", FPR32, f32, 0b00, 0b0, 0b1, fmsub>; -def FNMADDssss : A64I_fpdp3Impl<"fnmadd", FPR32, f32, 0b00, 0b1, 0b0, fnmadd>; -def FNMSUBssss : A64I_fpdp3Impl<"fnmsub", FPR32, f32, 0b00, 0b1, 0b1, fnmsub>; - -def FMADDdddd : A64I_fpdp3Impl<"fmadd", FPR64, f64, 0b01, 0b0, 0b0, fma>; -def FMSUBdddd : A64I_fpdp3Impl<"fmsub", FPR64, f64, 0b01, 0b0, 0b1, fmsub>; -def FNMADDdddd : A64I_fpdp3Impl<"fnmadd", FPR64, f64, 0b01, 0b1, 0b0, fnmadd>; -def FNMSUBdddd : A64I_fpdp3Impl<"fnmsub", FPR64, f64, 0b01, 0b1, 0b1, fnmsub>; - -//===----------------------------------------------------------------------===// -// Floating-point <-> fixed-point conversion instructions -//===----------------------------------------------------------------------===// -// Contains: FCVTZS, FCVTZU, SCVTF, UCVTF - -// #1-#32 allowed, encoded as "64 - -def fixedpos_asmoperand_i32 : AsmOperandClass { - let Name = "CVTFixedPos32"; - let RenderMethod = "addCVTFixedPosOperands"; - let PredicateMethod = "isCVTFixedPos<32>"; - let DiagnosticType = "CVTFixedPos32"; -} - -// Also encoded as "64 - " but #1-#64 allowed. -def fixedpos_asmoperand_i64 : AsmOperandClass { - let Name = "CVTFixedPos64"; - let RenderMethod = "addCVTFixedPosOperands"; - let PredicateMethod = "isCVTFixedPos<64>"; - let DiagnosticType = "CVTFixedPos64"; -} - -// We need the cartesian product of f32/f64 i32/i64 operands for -// conversions: -// + Selection needs to use operands of correct floating type -// + Assembly parsing and decoding depend on integer width -class cvtfix_i32_op - : Operand, - ComplexPattern", [fpimm]> { - let ParserMatchClass = fixedpos_asmoperand_i32; - let DecoderMethod = "DecodeCVT32FixedPosOperand"; - let PrintMethod = "printCVTFixedPosOperand"; -} - -class cvtfix_i64_op - : Operand, - ComplexPattern", [fpimm]> { - let ParserMatchClass = fixedpos_asmoperand_i64; - let PrintMethod = "printCVTFixedPosOperand"; -} - -// Because of the proliferation of weird operands, it's not really -// worth going for a multiclass here. Oh well. - -class A64I_fptofix type, bits<3> opcode, - RegisterClass GPR, RegisterClass FPR, Operand scale_op, - string asmop, SDNode cvtop> - : A64I_fpfixed; - -def FCVTZSwsi : A64I_fptofix<0b0, 0b00, 0b000, GPR32, FPR32, - cvtfix_i32_op, "fcvtzs", fp_to_sint>; -def FCVTZSxsi : A64I_fptofix<0b1, 0b00, 0b000, GPR64, FPR32, - cvtfix_i64_op, "fcvtzs", fp_to_sint>; -def FCVTZUwsi : A64I_fptofix<0b0, 0b00, 0b001, GPR32, FPR32, - cvtfix_i32_op, "fcvtzu", fp_to_uint>; -def FCVTZUxsi : A64I_fptofix<0b1, 0b00, 0b001, GPR64, FPR32, - cvtfix_i64_op, "fcvtzu", fp_to_uint>; - -def FCVTZSwdi : A64I_fptofix<0b0, 0b01, 0b000, GPR32, FPR64, - cvtfix_i32_op, "fcvtzs", fp_to_sint>; -def FCVTZSxdi : A64I_fptofix<0b1, 0b01, 0b000, GPR64, FPR64, - cvtfix_i64_op, "fcvtzs", fp_to_sint>; -def FCVTZUwdi : A64I_fptofix<0b0, 0b01, 0b001, GPR32, FPR64, - cvtfix_i32_op, "fcvtzu", fp_to_uint>; -def FCVTZUxdi : A64I_fptofix<0b1, 0b01, 0b001, GPR64, FPR64, - cvtfix_i64_op, "fcvtzu", fp_to_uint>; - - -class A64I_fixtofp type, bits<3> opcode, - RegisterClass FPR, RegisterClass GPR, Operand scale_op, - string asmop, SDNode cvtop> - : A64I_fpfixed; - -def SCVTFswi : A64I_fixtofp<0b0, 0b00, 0b010, FPR32, GPR32, - cvtfix_i32_op, "scvtf", sint_to_fp>; -def SCVTFsxi : A64I_fixtofp<0b1, 0b00, 0b010, FPR32, GPR64, - cvtfix_i64_op, "scvtf", sint_to_fp>; -def UCVTFswi : A64I_fixtofp<0b0, 0b00, 0b011, FPR32, GPR32, - cvtfix_i32_op, "ucvtf", uint_to_fp>; -def UCVTFsxi : A64I_fixtofp<0b1, 0b00, 0b011, FPR32, GPR64, - cvtfix_i64_op, "ucvtf", uint_to_fp>; -def SCVTFdwi : A64I_fixtofp<0b0, 0b01, 0b010, FPR64, GPR32, - cvtfix_i32_op, "scvtf", sint_to_fp>; -def SCVTFdxi : A64I_fixtofp<0b1, 0b01, 0b010, FPR64, GPR64, - cvtfix_i64_op, "scvtf", sint_to_fp>; -def UCVTFdwi : A64I_fixtofp<0b0, 0b01, 0b011, FPR64, GPR32, - cvtfix_i32_op, "ucvtf", uint_to_fp>; -def UCVTFdxi : A64I_fixtofp<0b1, 0b01, 0b011, FPR64, GPR64, - cvtfix_i64_op, "ucvtf", uint_to_fp>; - -//===----------------------------------------------------------------------===// -// Floating-point <-> integer conversion instructions -//===----------------------------------------------------------------------===// -// Contains: FCVTZS, FCVTZU, SCVTF, UCVTF - -class A64I_fpintI type, bits<2> rmode, bits<3> opcode, - RegisterClass DestPR, RegisterClass SrcPR, string asmop> - : A64I_fpint; - -multiclass A64I_fptointRM rmode, bit o2, string asmop> { - def Sws : A64I_fpintI<0b0, 0b00, rmode, {o2, 0, 0}, - GPR32, FPR32, asmop # "s">; - def Sxs : A64I_fpintI<0b1, 0b00, rmode, {o2, 0, 0}, - GPR64, FPR32, asmop # "s">; - def Uws : A64I_fpintI<0b0, 0b00, rmode, {o2, 0, 1}, - GPR32, FPR32, asmop # "u">; - def Uxs : A64I_fpintI<0b1, 0b00, rmode, {o2, 0, 1}, - GPR64, FPR32, asmop # "u">; - - def Swd : A64I_fpintI<0b0, 0b01, rmode, {o2, 0, 0}, - GPR32, FPR64, asmop # "s">; - def Sxd : A64I_fpintI<0b1, 0b01, rmode, {o2, 0, 0}, - GPR64, FPR64, asmop # "s">; - def Uwd : A64I_fpintI<0b0, 0b01, rmode, {o2, 0, 1}, - GPR32, FPR64, asmop # "u">; - def Uxd : A64I_fpintI<0b1, 0b01, rmode, {o2, 0, 1}, - GPR64, FPR64, asmop # "u">; -} - -defm FCVTN : A64I_fptointRM<0b00, 0b0, "fcvtn">; -defm FCVTP : A64I_fptointRM<0b01, 0b0, "fcvtp">; -defm FCVTM : A64I_fptointRM<0b10, 0b0, "fcvtm">; -defm FCVTZ : A64I_fptointRM<0b11, 0b0, "fcvtz">; -defm FCVTA : A64I_fptointRM<0b00, 0b1, "fcvta">; - -def : Pat<(i32 (fp_to_sint FPR32:$Rn)), (FCVTZSws FPR32:$Rn)>; -def : Pat<(i64 (fp_to_sint FPR32:$Rn)), (FCVTZSxs FPR32:$Rn)>; -def : Pat<(i32 (fp_to_uint FPR32:$Rn)), (FCVTZUws FPR32:$Rn)>; -def : Pat<(i64 (fp_to_uint FPR32:$Rn)), (FCVTZUxs FPR32:$Rn)>; -def : Pat<(i32 (fp_to_sint (f64 FPR64:$Rn))), (FCVTZSwd FPR64:$Rn)>; -def : Pat<(i64 (fp_to_sint (f64 FPR64:$Rn))), (FCVTZSxd FPR64:$Rn)>; -def : Pat<(i32 (fp_to_uint (f64 FPR64:$Rn))), (FCVTZUwd FPR64:$Rn)>; -def : Pat<(i64 (fp_to_uint (f64 FPR64:$Rn))), (FCVTZUxd FPR64:$Rn)>; - -multiclass A64I_inttofp { - def CVTFsw : A64I_fpintI<0b0, 0b00, 0b00, {0, 1, o0}, FPR32, GPR32, asmop>; - def CVTFsx : A64I_fpintI<0b1, 0b00, 0b00, {0, 1, o0}, FPR32, GPR64, asmop>; - def CVTFdw : A64I_fpintI<0b0, 0b01, 0b00, {0, 1, o0}, FPR64, GPR32, asmop>; - def CVTFdx : A64I_fpintI<0b1, 0b01, 0b00, {0, 1, o0}, FPR64, GPR64, asmop>; -} - -defm S : A64I_inttofp<0b0, "scvtf">; -defm U : A64I_inttofp<0b1, "ucvtf">; - -def : Pat<(f32 (sint_to_fp GPR32:$Rn)), (SCVTFsw GPR32:$Rn)>; -def : Pat<(f32 (sint_to_fp GPR64:$Rn)), (SCVTFsx GPR64:$Rn)>; -def : Pat<(f64 (sint_to_fp GPR32:$Rn)), (SCVTFdw GPR32:$Rn)>; -def : Pat<(f64 (sint_to_fp GPR64:$Rn)), (SCVTFdx GPR64:$Rn)>; -def : Pat<(f32 (uint_to_fp GPR32:$Rn)), (UCVTFsw GPR32:$Rn)>; -def : Pat<(f32 (uint_to_fp GPR64:$Rn)), (UCVTFsx GPR64:$Rn)>; -def : Pat<(f64 (uint_to_fp GPR32:$Rn)), (UCVTFdw GPR32:$Rn)>; -def : Pat<(f64 (uint_to_fp GPR64:$Rn)), (UCVTFdx GPR64:$Rn)>; - -def FMOVws : A64I_fpintI<0b0, 0b00, 0b00, 0b110, GPR32, FPR32, "fmov">; -def FMOVsw : A64I_fpintI<0b0, 0b00, 0b00, 0b111, FPR32, GPR32, "fmov">; -def FMOVxd : A64I_fpintI<0b1, 0b01, 0b00, 0b110, GPR64, FPR64, "fmov">; -def FMOVdx : A64I_fpintI<0b1, 0b01, 0b00, 0b111, FPR64, GPR64, "fmov">; - -def : Pat<(i32 (bitconvert (f32 FPR32:$Rn))), (FMOVws FPR32:$Rn)>; -def : Pat<(f32 (bitconvert (i32 GPR32:$Rn))), (FMOVsw GPR32:$Rn)>; -def : Pat<(i64 (bitconvert (f64 FPR64:$Rn))), (FMOVxd FPR64:$Rn)>; -def : Pat<(f64 (bitconvert (i64 GPR64:$Rn))), (FMOVdx GPR64:$Rn)>; - -def lane1_asmoperand : AsmOperandClass { - let Name = "Lane1"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "Lane1"; -} - -def lane1 : Operand { - let ParserMatchClass = lane1_asmoperand; - let PrintMethod = "printBareImmOperand"; -} - -let DecoderMethod = "DecodeFMOVLaneInstruction" in { - def FMOVxv : A64I_fpint<0b1, 0b0, 0b10, 0b01, 0b110, - (outs GPR64:$Rd), (ins VPR128:$Rn, lane1:$Lane), - "fmov\t$Rd, $Rn.d[$Lane]", [], NoItinerary>; - - def FMOVvx : A64I_fpint<0b1, 0b0, 0b10, 0b01, 0b111, - (outs VPR128:$Rd), (ins GPR64:$Rn, lane1:$Lane), - "fmov\t$Rd.d[$Lane], $Rn", [], NoItinerary>; -} - -def : InstAlias<"fmov $Rd, $Rn.2d[$Lane]", - (FMOVxv GPR64:$Rd, VPR128:$Rn, lane1:$Lane), 0b0>; - -def : InstAlias<"fmov $Rd.2d[$Lane], $Rn", - (FMOVvx VPR128:$Rd, GPR64:$Rn, lane1:$Lane), 0b0>; +defm ABS : SIMDTwoScalarD< 0, 0b01011, "abs", int_aarch64_neon_abs>; +defm CMEQ : SIMDCmpTwoScalarD< 0, 0b01001, "cmeq", AArch64cmeqz>; +defm CMGE : SIMDCmpTwoScalarD< 1, 0b01000, "cmge", AArch64cmgez>; +defm CMGT : SIMDCmpTwoScalarD< 0, 0b01000, "cmgt", AArch64cmgtz>; +defm CMLE : SIMDCmpTwoScalarD< 1, 0b01001, "cmle", AArch64cmlez>; +defm CMLT : SIMDCmpTwoScalarD< 0, 0b01010, "cmlt", AArch64cmltz>; +defm FCMEQ : SIMDFPCmpTwoScalar<0, 1, 0b01101, "fcmeq", AArch64fcmeqz>; +defm FCMGE : SIMDFPCmpTwoScalar<1, 1, 0b01100, "fcmge", AArch64fcmgez>; +defm FCMGT : SIMDFPCmpTwoScalar<0, 1, 0b01100, "fcmgt", AArch64fcmgtz>; +defm FCMLE : SIMDFPCmpTwoScalar<1, 1, 0b01101, "fcmle", AArch64fcmlez>; +defm FCMLT : SIMDFPCmpTwoScalar<0, 1, 0b01110, "fcmlt", AArch64fcmltz>; +defm FCVTAS : SIMDFPTwoScalar< 0, 0, 0b11100, "fcvtas">; +defm FCVTAU : SIMDFPTwoScalar< 1, 0, 0b11100, "fcvtau">; +defm FCVTMS : SIMDFPTwoScalar< 0, 0, 0b11011, "fcvtms">; +defm FCVTMU : SIMDFPTwoScalar< 1, 0, 0b11011, "fcvtmu">; +defm FCVTNS : SIMDFPTwoScalar< 0, 0, 0b11010, "fcvtns">; +defm FCVTNU : SIMDFPTwoScalar< 1, 0, 0b11010, "fcvtnu">; +defm FCVTPS : SIMDFPTwoScalar< 0, 1, 0b11010, "fcvtps">; +defm FCVTPU : SIMDFPTwoScalar< 1, 1, 0b11010, "fcvtpu">; +def FCVTXNv1i64 : SIMDInexactCvtTwoScalar<0b10110, "fcvtxn">; +defm FCVTZS : SIMDFPTwoScalar< 0, 1, 0b11011, "fcvtzs">; +defm FCVTZU : SIMDFPTwoScalar< 1, 1, 0b11011, "fcvtzu">; +defm FRECPE : SIMDFPTwoScalar< 0, 1, 0b11101, "frecpe">; +defm FRECPX : SIMDFPTwoScalar< 0, 1, 0b11111, "frecpx">; +defm FRSQRTE : SIMDFPTwoScalar< 1, 1, 0b11101, "frsqrte">; +defm NEG : SIMDTwoScalarD< 1, 0b01011, "neg", + UnOpFrag<(sub immAllZerosV, node:$LHS)> >; +defm SCVTF : SIMDFPTwoScalarCVT< 0, 0, 0b11101, "scvtf", AArch64sitof>; +defm SQABS : SIMDTwoScalarBHSD< 0, 0b00111, "sqabs", int_aarch64_neon_sqabs>; +defm SQNEG : SIMDTwoScalarBHSD< 1, 0b00111, "sqneg", int_aarch64_neon_sqneg>; +defm SQXTN : SIMDTwoScalarMixedBHS< 0, 0b10100, "sqxtn", int_aarch64_neon_scalar_sqxtn>; +defm SQXTUN : SIMDTwoScalarMixedBHS< 1, 0b10010, "sqxtun", int_aarch64_neon_scalar_sqxtun>; +defm SUQADD : SIMDTwoScalarBHSDTied< 0, 0b00011, "suqadd", + int_aarch64_neon_suqadd>; +defm UCVTF : SIMDFPTwoScalarCVT< 1, 0, 0b11101, "ucvtf", AArch64uitof>; +defm UQXTN : SIMDTwoScalarMixedBHS<1, 0b10100, "uqxtn", int_aarch64_neon_scalar_uqxtn>; +defm USQADD : SIMDTwoScalarBHSDTied< 1, 0b00011, "usqadd", + int_aarch64_neon_usqadd>; + +def : Pat<(AArch64neg (v1i64 V64:$Rn)), (NEGv1i64 V64:$Rn)>; + +def : Pat<(v1i64 (int_aarch64_neon_fcvtas (v1f64 FPR64:$Rn))), + (FCVTASv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtau (v1f64 FPR64:$Rn))), + (FCVTAUv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtms (v1f64 FPR64:$Rn))), + (FCVTMSv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtmu (v1f64 FPR64:$Rn))), + (FCVTMUv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtns (v1f64 FPR64:$Rn))), + (FCVTNSv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtnu (v1f64 FPR64:$Rn))), + (FCVTNUv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtps (v1f64 FPR64:$Rn))), + (FCVTPSv1i64 FPR64:$Rn)>; +def : Pat<(v1i64 (int_aarch64_neon_fcvtpu (v1f64 FPR64:$Rn))), + (FCVTPUv1i64 FPR64:$Rn)>; + +def : Pat<(f32 (int_aarch64_neon_frecpe (f32 FPR32:$Rn))), + (FRECPEv1i32 FPR32:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_frecpe (f64 FPR64:$Rn))), + (FRECPEv1i64 FPR64:$Rn)>; +def : Pat<(v1f64 (int_aarch64_neon_frecpe (v1f64 FPR64:$Rn))), + (FRECPEv1i64 FPR64:$Rn)>; + +def : Pat<(f32 (int_aarch64_neon_frecpx (f32 FPR32:$Rn))), + (FRECPXv1i32 FPR32:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_frecpx (f64 FPR64:$Rn))), + (FRECPXv1i64 FPR64:$Rn)>; + +def : Pat<(f32 (int_aarch64_neon_frsqrte (f32 FPR32:$Rn))), + (FRSQRTEv1i32 FPR32:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_frsqrte (f64 FPR64:$Rn))), + (FRSQRTEv1i64 FPR64:$Rn)>; +def : Pat<(v1f64 (int_aarch64_neon_frsqrte (v1f64 FPR64:$Rn))), + (FRSQRTEv1i64 FPR64:$Rn)>; + +// If an integer is about to be converted to a floating point value, +// just load it on the floating point unit. +// Here are the patterns for 8 and 16-bits to float. +// 8-bits -> float. +multiclass UIntToFPROLoadPat { + def : Pat<(DstTy (uint_to_fp (SrcTy + (loadop (ro.Wpat GPR64sp:$Rn, GPR32:$Rm, + ro.Wext:$extend))))), + (UCVTF (INSERT_SUBREG (DstTy (IMPLICIT_DEF)), + (LDRW GPR64sp:$Rn, GPR32:$Rm, ro.Wext:$extend), + sub))>; + + def : Pat<(DstTy (uint_to_fp (SrcTy + (loadop (ro.Xpat GPR64sp:$Rn, GPR64:$Rm, + ro.Wext:$extend))))), + (UCVTF (INSERT_SUBREG (DstTy (IMPLICIT_DEF)), + (LDRX GPR64sp:$Rn, GPR64:$Rm, ro.Xext:$extend), + sub))>; +} + +defm : UIntToFPROLoadPat; +def : Pat <(f32 (uint_to_fp (i32 + (zextloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))))), + (UCVTFv1i32 (INSERT_SUBREG (f32 (IMPLICIT_DEF)), + (LDRBui GPR64sp:$Rn, uimm12s1:$offset), bsub))>; +def : Pat <(f32 (uint_to_fp (i32 + (zextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))))), + (UCVTFv1i32 (INSERT_SUBREG (f32 (IMPLICIT_DEF)), + (LDURBi GPR64sp:$Rn, simm9:$offset), bsub))>; +// 16-bits -> float. +defm : UIntToFPROLoadPat; +def : Pat <(f32 (uint_to_fp (i32 + (zextloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))))), + (UCVTFv1i32 (INSERT_SUBREG (f32 (IMPLICIT_DEF)), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset), hsub))>; +def : Pat <(f32 (uint_to_fp (i32 + (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))))), + (UCVTFv1i32 (INSERT_SUBREG (f32 (IMPLICIT_DEF)), + (LDURHi GPR64sp:$Rn, simm9:$offset), hsub))>; +// 32-bits are handled in target specific dag combine: +// performIntToFpCombine. +// 64-bits integer to 32-bits floating point, not possible with +// UCVTF on floating point registers (both source and destination +// must have the same size). + +// Here are the patterns for 8, 16, 32, and 64-bits to double. +// 8-bits -> double. +defm : UIntToFPROLoadPat; +def : Pat <(f64 (uint_to_fp (i32 + (zextloadi8 (am_indexed8 GPR64sp:$Rn, uimm12s1:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDRBui GPR64sp:$Rn, uimm12s1:$offset), bsub))>; +def : Pat <(f64 (uint_to_fp (i32 + (zextloadi8 (am_unscaled8 GPR64sp:$Rn, simm9:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDURBi GPR64sp:$Rn, simm9:$offset), bsub))>; +// 16-bits -> double. +defm : UIntToFPROLoadPat; +def : Pat <(f64 (uint_to_fp (i32 + (zextloadi16 (am_indexed16 GPR64sp:$Rn, uimm12s2:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset), hsub))>; +def : Pat <(f64 (uint_to_fp (i32 + (zextloadi16 (am_unscaled16 GPR64sp:$Rn, simm9:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDURHi GPR64sp:$Rn, simm9:$offset), hsub))>; +// 32-bits -> double. +defm : UIntToFPROLoadPat; +def : Pat <(f64 (uint_to_fp (i32 + (load (am_indexed32 GPR64sp:$Rn, uimm12s4:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDRSui GPR64sp:$Rn, uimm12s4:$offset), ssub))>; +def : Pat <(f64 (uint_to_fp (i32 + (load (am_unscaled32 GPR64sp:$Rn, simm9:$offset))))), + (UCVTFv1i64 (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + (LDURSi GPR64sp:$Rn, simm9:$offset), ssub))>; +// 64-bits -> double are handled in target specific dag combine: +// performIntToFpCombine. //===----------------------------------------------------------------------===// -// Floating-point immediate instructions +// Advanced SIMD three different-sized vector instructions. //===----------------------------------------------------------------------===// -// Contains: FMOV - -def fpimm_asmoperand : AsmOperandClass { - let Name = "FMOVImm"; - let ParserMethod = "ParseFPImmOperand"; - let DiagnosticType = "FPImm"; -} -// The MCOperand for these instructions are the encoded 8-bit values. -def SDXF_fpimm : SDNodeXFormgetValueAPF(), Imm8); - return CurDAG->getTargetConstant(Imm8, MVT::i32); +defm ADDHN : SIMDNarrowThreeVectorBHS<0,0b0100,"addhn", int_aarch64_neon_addhn>; +defm SUBHN : SIMDNarrowThreeVectorBHS<0,0b0110,"subhn", int_aarch64_neon_subhn>; +defm RADDHN : SIMDNarrowThreeVectorBHS<1,0b0100,"raddhn",int_aarch64_neon_raddhn>; +defm RSUBHN : SIMDNarrowThreeVectorBHS<1,0b0110,"rsubhn",int_aarch64_neon_rsubhn>; +defm PMULL : SIMDDifferentThreeVectorBD<0,0b1110,"pmull",int_aarch64_neon_pmull>; +defm SABAL : SIMDLongThreeVectorTiedBHSabal<0,0b0101,"sabal", + sabsdiff>; +defm SABDL : SIMDLongThreeVectorBHSabdl<0, 0b0111, "sabdl", + sabsdiff>; +defm SADDL : SIMDLongThreeVectorBHS< 0, 0b0000, "saddl", + BinOpFrag<(add (sext node:$LHS), (sext node:$RHS))>>; +defm SADDW : SIMDWideThreeVectorBHS< 0, 0b0001, "saddw", + BinOpFrag<(add node:$LHS, (sext node:$RHS))>>; +defm SMLAL : SIMDLongThreeVectorTiedBHS<0, 0b1000, "smlal", + TriOpFrag<(add node:$LHS, (int_aarch64_neon_smull node:$MHS, node:$RHS))>>; +defm SMLSL : SIMDLongThreeVectorTiedBHS<0, 0b1010, "smlsl", + TriOpFrag<(sub node:$LHS, (int_aarch64_neon_smull node:$MHS, node:$RHS))>>; +defm SMULL : SIMDLongThreeVectorBHS<0, 0b1100, "smull", int_aarch64_neon_smull>; +defm SQDMLAL : SIMDLongThreeVectorSQDMLXTiedHS<0, 0b1001, "sqdmlal", + int_aarch64_neon_sqadd>; +defm SQDMLSL : SIMDLongThreeVectorSQDMLXTiedHS<0, 0b1011, "sqdmlsl", + int_aarch64_neon_sqsub>; +defm SQDMULL : SIMDLongThreeVectorHS<0, 0b1101, "sqdmull", + int_aarch64_neon_sqdmull>; +defm SSUBL : SIMDLongThreeVectorBHS<0, 0b0010, "ssubl", + BinOpFrag<(sub (sext node:$LHS), (sext node:$RHS))>>; +defm SSUBW : SIMDWideThreeVectorBHS<0, 0b0011, "ssubw", + BinOpFrag<(sub node:$LHS, (sext node:$RHS))>>; +defm UABAL : SIMDLongThreeVectorTiedBHSabal<1, 0b0101, "uabal", + uabsdiff>; +defm UADDL : SIMDLongThreeVectorBHS<1, 0b0000, "uaddl", + BinOpFrag<(add (zext node:$LHS), (zext node:$RHS))>>; +defm UADDW : SIMDWideThreeVectorBHS<1, 0b0001, "uaddw", + BinOpFrag<(add node:$LHS, (zext node:$RHS))>>; +defm UMLAL : SIMDLongThreeVectorTiedBHS<1, 0b1000, "umlal", + TriOpFrag<(add node:$LHS, (int_aarch64_neon_umull node:$MHS, node:$RHS))>>; +defm UMLSL : SIMDLongThreeVectorTiedBHS<1, 0b1010, "umlsl", + TriOpFrag<(sub node:$LHS, (int_aarch64_neon_umull node:$MHS, node:$RHS))>>; +defm UMULL : SIMDLongThreeVectorBHS<1, 0b1100, "umull", int_aarch64_neon_umull>; +defm USUBL : SIMDLongThreeVectorBHS<1, 0b0010, "usubl", + BinOpFrag<(sub (zext node:$LHS), (zext node:$RHS))>>; +defm USUBW : SIMDWideThreeVectorBHS< 1, 0b0011, "usubw", + BinOpFrag<(sub node:$LHS, (zext node:$RHS))>>; + +// Additional patterns for SMULL and UMULL +multiclass Neon_mul_widen_patterns { + def : Pat<(v8i16 (opnode (v8i8 V64:$Rn), (v8i8 V64:$Rm))), + (INST8B V64:$Rn, V64:$Rm)>; + def : Pat<(v4i32 (opnode (v4i16 V64:$Rn), (v4i16 V64:$Rm))), + (INST4H V64:$Rn, V64:$Rm)>; + def : Pat<(v2i64 (opnode (v2i32 V64:$Rn), (v2i32 V64:$Rm))), + (INST2S V64:$Rn, V64:$Rm)>; +} + +defm : Neon_mul_widen_patterns; +defm : Neon_mul_widen_patterns; + +// Additional patterns for SMLAL/SMLSL and UMLAL/UMLSL +multiclass Neon_mulacc_widen_patterns { + def : Pat<(v8i16 (opnode (v8i16 V128:$Rd), (v8i8 V64:$Rn), (v8i8 V64:$Rm))), + (INST8B V128:$Rd, V64:$Rn, V64:$Rm)>; + def : Pat<(v4i32 (opnode (v4i32 V128:$Rd), (v4i16 V64:$Rn), (v4i16 V64:$Rm))), + (INST4H V128:$Rd, V64:$Rn, V64:$Rm)>; + def : Pat<(v2i64 (opnode (v2i64 V128:$Rd), (v2i32 V64:$Rn), (v2i32 V64:$Rm))), + (INST2S V128:$Rd, V64:$Rn, V64:$Rm)>; +} + +defm : Neon_mulacc_widen_patterns< + TriOpFrag<(add node:$LHS, (AArch64smull node:$MHS, node:$RHS))>, + SMLALv8i8_v8i16, SMLALv4i16_v4i32, SMLALv2i32_v2i64>; +defm : Neon_mulacc_widen_patterns< + TriOpFrag<(add node:$LHS, (AArch64umull node:$MHS, node:$RHS))>, + UMLALv8i8_v8i16, UMLALv4i16_v4i32, UMLALv2i32_v2i64>; +defm : Neon_mulacc_widen_patterns< + TriOpFrag<(sub node:$LHS, (AArch64smull node:$MHS, node:$RHS))>, + SMLSLv8i8_v8i16, SMLSLv4i16_v4i32, SMLSLv2i32_v2i64>; +defm : Neon_mulacc_widen_patterns< + TriOpFrag<(sub node:$LHS, (AArch64umull node:$MHS, node:$RHS))>, + UMLSLv8i8_v8i16, UMLSLv4i16_v4i32, UMLSLv2i32_v2i64>; + +// Patterns for 64-bit pmull +def : Pat<(int_aarch64_neon_pmull64 V64:$Rn, V64:$Rm), + (PMULLv1i64 V64:$Rn, V64:$Rm)>; +def : Pat<(int_aarch64_neon_pmull64 (extractelt (v2i64 V128:$Rn), (i64 1)), + (extractelt (v2i64 V128:$Rm), (i64 1))), + (PMULLv2i64 V128:$Rn, V128:$Rm)>; + +// CodeGen patterns for addhn and subhn instructions, which can actually be +// written in LLVM IR without too much difficulty. + +// ADDHN +def : Pat<(v8i8 (trunc (v8i16 (AArch64vlshr (add V128:$Rn, V128:$Rm), (i32 8))))), + (ADDHNv8i16_v8i8 V128:$Rn, V128:$Rm)>; +def : Pat<(v4i16 (trunc (v4i32 (AArch64vlshr (add V128:$Rn, V128:$Rm), + (i32 16))))), + (ADDHNv4i32_v4i16 V128:$Rn, V128:$Rm)>; +def : Pat<(v2i32 (trunc (v2i64 (AArch64vlshr (add V128:$Rn, V128:$Rm), + (i32 32))))), + (ADDHNv2i64_v2i32 V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v8i8 V64:$Rd), + (trunc (v8i16 (AArch64vlshr (add V128:$Rn, V128:$Rm), + (i32 8))))), + (ADDHNv8i16_v16i8 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v4i16 V64:$Rd), + (trunc (v4i32 (AArch64vlshr (add V128:$Rn, V128:$Rm), + (i32 16))))), + (ADDHNv4i32_v8i16 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v2i32 V64:$Rd), + (trunc (v2i64 (AArch64vlshr (add V128:$Rn, V128:$Rm), + (i32 32))))), + (ADDHNv2i64_v4i32 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; + +// SUBHN +def : Pat<(v8i8 (trunc (v8i16 (AArch64vlshr (sub V128:$Rn, V128:$Rm), (i32 8))))), + (SUBHNv8i16_v8i8 V128:$Rn, V128:$Rm)>; +def : Pat<(v4i16 (trunc (v4i32 (AArch64vlshr (sub V128:$Rn, V128:$Rm), + (i32 16))))), + (SUBHNv4i32_v4i16 V128:$Rn, V128:$Rm)>; +def : Pat<(v2i32 (trunc (v2i64 (AArch64vlshr (sub V128:$Rn, V128:$Rm), + (i32 32))))), + (SUBHNv2i64_v2i32 V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v8i8 V64:$Rd), + (trunc (v8i16 (AArch64vlshr (sub V128:$Rn, V128:$Rm), + (i32 8))))), + (SUBHNv8i16_v16i8 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v4i16 V64:$Rd), + (trunc (v4i32 (AArch64vlshr (sub V128:$Rn, V128:$Rm), + (i32 16))))), + (SUBHNv4i32_v8i16 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; +def : Pat<(concat_vectors (v2i32 V64:$Rd), + (trunc (v2i64 (AArch64vlshr (sub V128:$Rn, V128:$Rm), + (i32 32))))), + (SUBHNv2i64_v4i32 (SUBREG_TO_REG (i32 0), V64:$Rd, dsub), + V128:$Rn, V128:$Rm)>; + +//---------------------------------------------------------------------------- +// AdvSIMD bitwise extract from vector instruction. +//---------------------------------------------------------------------------- + +defm EXT : SIMDBitwiseExtract<"ext">; + +def : Pat<(v4i16 (AArch64ext V64:$Rn, V64:$Rm, (i32 imm:$imm))), + (EXTv8i8 V64:$Rn, V64:$Rm, imm:$imm)>; +def : Pat<(v8i16 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; +def : Pat<(v2i32 (AArch64ext V64:$Rn, V64:$Rm, (i32 imm:$imm))), + (EXTv8i8 V64:$Rn, V64:$Rm, imm:$imm)>; +def : Pat<(v2f32 (AArch64ext V64:$Rn, V64:$Rm, (i32 imm:$imm))), + (EXTv8i8 V64:$Rn, V64:$Rm, imm:$imm)>; +def : Pat<(v4i32 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; +def : Pat<(v4f32 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; +def : Pat<(v2i64 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; +def : Pat<(v2f64 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; +def : Pat<(v4f16 (AArch64ext V64:$Rn, V64:$Rm, (i32 imm:$imm))), + (EXTv8i8 V64:$Rn, V64:$Rm, imm:$imm)>; +def : Pat<(v8f16 (AArch64ext V128:$Rn, V128:$Rm, (i32 imm:$imm))), + (EXTv16i8 V128:$Rn, V128:$Rm, imm:$imm)>; + +// We use EXT to handle extract_subvector to copy the upper 64-bits of a +// 128-bit vector. +def : Pat<(v8i8 (extract_subvector V128:$Rn, (i64 8))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v4i16 (extract_subvector V128:$Rn, (i64 4))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v2i32 (extract_subvector V128:$Rn, (i64 2))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v1i64 (extract_subvector V128:$Rn, (i64 1))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v4f16 (extract_subvector V128:$Rn, (i64 4))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v2f32 (extract_subvector V128:$Rn, (i64 2))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; +def : Pat<(v1f64 (extract_subvector V128:$Rn, (i64 1))), + (EXTRACT_SUBREG (EXTv16i8 V128:$Rn, V128:$Rn, 8), dsub)>; + + +//---------------------------------------------------------------------------- +// AdvSIMD zip vector +//---------------------------------------------------------------------------- + +defm TRN1 : SIMDZipVector<0b010, "trn1", AArch64trn1>; +defm TRN2 : SIMDZipVector<0b110, "trn2", AArch64trn2>; +defm UZP1 : SIMDZipVector<0b001, "uzp1", AArch64uzp1>; +defm UZP2 : SIMDZipVector<0b101, "uzp2", AArch64uzp2>; +defm ZIP1 : SIMDZipVector<0b011, "zip1", AArch64zip1>; +defm ZIP2 : SIMDZipVector<0b111, "zip2", AArch64zip2>; + +//---------------------------------------------------------------------------- +// AdvSIMD TBL/TBX instructions +//---------------------------------------------------------------------------- + +defm TBL : SIMDTableLookup< 0, "tbl">; +defm TBX : SIMDTableLookupTied<1, "tbx">; + +def : Pat<(v8i8 (int_aarch64_neon_tbl1 (v16i8 VecListOne128:$Rn), (v8i8 V64:$Ri))), + (TBLv8i8One VecListOne128:$Rn, V64:$Ri)>; +def : Pat<(v16i8 (int_aarch64_neon_tbl1 (v16i8 V128:$Ri), (v16i8 V128:$Rn))), + (TBLv16i8One V128:$Ri, V128:$Rn)>; + +def : Pat<(v8i8 (int_aarch64_neon_tbx1 (v8i8 V64:$Rd), + (v16i8 VecListOne128:$Rn), (v8i8 V64:$Ri))), + (TBXv8i8One V64:$Rd, VecListOne128:$Rn, V64:$Ri)>; +def : Pat<(v16i8 (int_aarch64_neon_tbx1 (v16i8 V128:$Rd), + (v16i8 V128:$Ri), (v16i8 V128:$Rn))), + (TBXv16i8One V128:$Rd, V128:$Ri, V128:$Rn)>; + + +//---------------------------------------------------------------------------- +// AdvSIMD scalar CPY instruction +//---------------------------------------------------------------------------- + +defm CPY : SIMDScalarCPY<"cpy">; + +//---------------------------------------------------------------------------- +// AdvSIMD scalar pairwise instructions +//---------------------------------------------------------------------------- + +defm ADDP : SIMDPairwiseScalarD<0, 0b11011, "addp">; +defm FADDP : SIMDFPPairwiseScalar<0, 0b01101, "faddp">; +defm FMAXNMP : SIMDFPPairwiseScalar<0, 0b01100, "fmaxnmp">; +defm FMAXP : SIMDFPPairwiseScalar<0, 0b01111, "fmaxp">; +defm FMINNMP : SIMDFPPairwiseScalar<1, 0b01100, "fminnmp">; +defm FMINP : SIMDFPPairwiseScalar<1, 0b01111, "fminp">; +def : Pat<(v2i64 (AArch64saddv V128:$Rn)), + (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)), (ADDPv2i64p V128:$Rn), dsub)>; +def : Pat<(v2i64 (AArch64uaddv V128:$Rn)), + (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)), (ADDPv2i64p V128:$Rn), dsub)>; +def : Pat<(f32 (int_aarch64_neon_faddv (v2f32 V64:$Rn))), + (FADDPv2i32p V64:$Rn)>; +def : Pat<(f32 (int_aarch64_neon_faddv (v4f32 V128:$Rn))), + (FADDPv2i32p (EXTRACT_SUBREG (FADDPv4f32 V128:$Rn, V128:$Rn), dsub))>; +def : Pat<(f64 (int_aarch64_neon_faddv (v2f64 V128:$Rn))), + (FADDPv2i64p V128:$Rn)>; +def : Pat<(f32 (int_aarch64_neon_fmaxnmv (v2f32 V64:$Rn))), + (FMAXNMPv2i32p V64:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_fmaxnmv (v2f64 V128:$Rn))), + (FMAXNMPv2i64p V128:$Rn)>; +def : Pat<(f32 (int_aarch64_neon_fmaxv (v2f32 V64:$Rn))), + (FMAXPv2i32p V64:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_fmaxv (v2f64 V128:$Rn))), + (FMAXPv2i64p V128:$Rn)>; +def : Pat<(f32 (int_aarch64_neon_fminnmv (v2f32 V64:$Rn))), + (FMINNMPv2i32p V64:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_fminnmv (v2f64 V128:$Rn))), + (FMINNMPv2i64p V128:$Rn)>; +def : Pat<(f32 (int_aarch64_neon_fminv (v2f32 V64:$Rn))), + (FMINPv2i32p V64:$Rn)>; +def : Pat<(f64 (int_aarch64_neon_fminv (v2f64 V128:$Rn))), + (FMINPv2i64p V128:$Rn)>; + +//---------------------------------------------------------------------------- +// AdvSIMD INS/DUP instructions +//---------------------------------------------------------------------------- + +def DUPv8i8gpr : SIMDDupFromMain<0, {?,?,?,?,1}, ".8b", v8i8, V64, GPR32>; +def DUPv16i8gpr : SIMDDupFromMain<1, {?,?,?,?,1}, ".16b", v16i8, V128, GPR32>; +def DUPv4i16gpr : SIMDDupFromMain<0, {?,?,?,1,0}, ".4h", v4i16, V64, GPR32>; +def DUPv8i16gpr : SIMDDupFromMain<1, {?,?,?,1,0}, ".8h", v8i16, V128, GPR32>; +def DUPv2i32gpr : SIMDDupFromMain<0, {?,?,1,0,0}, ".2s", v2i32, V64, GPR32>; +def DUPv4i32gpr : SIMDDupFromMain<1, {?,?,1,0,0}, ".4s", v4i32, V128, GPR32>; +def DUPv2i64gpr : SIMDDupFromMain<1, {?,1,0,0,0}, ".2d", v2i64, V128, GPR64>; + +def DUPv2i64lane : SIMDDup64FromElement; +def DUPv2i32lane : SIMDDup32FromElement<0, ".2s", v2i32, V64>; +def DUPv4i32lane : SIMDDup32FromElement<1, ".4s", v4i32, V128>; +def DUPv4i16lane : SIMDDup16FromElement<0, ".4h", v4i16, V64>; +def DUPv8i16lane : SIMDDup16FromElement<1, ".8h", v8i16, V128>; +def DUPv8i8lane : SIMDDup8FromElement <0, ".8b", v8i8, V64>; +def DUPv16i8lane : SIMDDup8FromElement <1, ".16b", v16i8, V128>; + +def : Pat<(v2f32 (AArch64dup (f32 FPR32:$Rn))), + (v2f32 (DUPv2i32lane + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR32:$Rn, ssub), + (i64 0)))>; +def : Pat<(v4f32 (AArch64dup (f32 FPR32:$Rn))), + (v4f32 (DUPv4i32lane + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR32:$Rn, ssub), + (i64 0)))>; +def : Pat<(v2f64 (AArch64dup (f64 FPR64:$Rn))), + (v2f64 (DUPv2i64lane + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR64:$Rn, dsub), + (i64 0)))>; +def : Pat<(v4f16 (AArch64dup (f16 FPR16:$Rn))), + (v4f16 (DUPv4i16lane + (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), FPR16:$Rn, hsub), + (i64 0)))>; +def : Pat<(v8f16 (AArch64dup (f16 FPR16:$Rn))), + (v8f16 (DUPv8i16lane + (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), FPR16:$Rn, hsub), + (i64 0)))>; + +def : Pat<(v4f16 (AArch64duplane16 (v8f16 V128:$Rn), VectorIndexH:$imm)), + (DUPv4i16lane V128:$Rn, VectorIndexH:$imm)>; +def : Pat<(v8f16 (AArch64duplane16 (v8f16 V128:$Rn), VectorIndexH:$imm)), + (DUPv8i16lane V128:$Rn, VectorIndexH:$imm)>; + +def : Pat<(v2f32 (AArch64duplane32 (v4f32 V128:$Rn), VectorIndexS:$imm)), + (DUPv2i32lane V128:$Rn, VectorIndexS:$imm)>; +def : Pat<(v4f32 (AArch64duplane32 (v4f32 V128:$Rn), VectorIndexS:$imm)), + (DUPv4i32lane V128:$Rn, VectorIndexS:$imm)>; +def : Pat<(v2f64 (AArch64duplane64 (v2f64 V128:$Rn), VectorIndexD:$imm)), + (DUPv2i64lane V128:$Rn, VectorIndexD:$imm)>; + +// If there's an (AArch64dup (vector_extract ...) ...), we can use a duplane +// instruction even if the types don't match: we just have to remap the lane +// carefully. N.b. this trick only applies to truncations. +def VecIndex_x2 : SDNodeXFormgetTargetConstant(2 * N->getZExtValue(), SDLoc(N), MVT::i64); }]>; - -class fmov_operand - : Operand, - PatLeaf<(FT fpimm), [{ return A64Imms::isFPImm(N->getValueAPF()); }], - SDXF_fpimm> { - let PrintMethod = "printFPImmOperand"; - let ParserMatchClass = fpimm_asmoperand; -} - -def fmov32_operand : fmov_operand; -def fmov64_operand : fmov_operand; - -class A64I_fpimm_impl type, RegisterClass Reg, ValueType VT, - Operand fmov_operand> - : A64I_fpimm<0b0, 0b0, type, 0b00000, - (outs Reg:$Rd), - (ins fmov_operand:$Imm8), - "fmov\t$Rd, $Imm8", - [(set (VT Reg:$Rd), fmov_operand:$Imm8)], - NoItinerary>; - -def FMOVsi : A64I_fpimm_impl<0b00, FPR32, f32, fmov32_operand>; -def FMOVdi : A64I_fpimm_impl<0b01, FPR64, f64, fmov64_operand>; - -//===----------------------------------------------------------------------===// -// Load-register (literal) instructions -//===----------------------------------------------------------------------===// -// Contains: LDR, LDRSW, PRFM - -def ldrlit_label_asmoperand : AsmOperandClass { - let Name = "LoadLitLabel"; - let RenderMethod = "addLabelOperands<19, 4>"; - let DiagnosticType = "Label"; -} - -def ldrlit_label : Operand { - let EncoderMethod = "getLoadLitLabelOpValue"; - - // This label is a 19-bit offset from PC, scaled by the instruction-width: 4. - let PrintMethod = "printLabelOperand<19, 4>"; - let ParserMatchClass = ldrlit_label_asmoperand; - let OperandType = "OPERAND_PCREL"; -} - -// Various instructions take an immediate value (which can always be used), -// where some numbers have a symbolic name to make things easier. These operands -// and the associated functions abstract away the differences. -multiclass namedimm { - def _asmoperand : AsmOperandClass { - let Name = "NamedImm" # prefix; - let PredicateMethod = "isUImm"; - let RenderMethod = "addImmOperands"; - let ParserMethod = "ParseNamedImmOperand<" # mapper # ">"; - let DiagnosticType = "NamedImm_" # prefix; - } - - def _op : Operand { - let ParserMatchClass = !cast(prefix # "_asmoperand"); - let PrintMethod = "printNamedImmOperand<" # mapper # ">"; - let DecoderMethod = "DecodeNamedImmOperand<" # mapper # ">"; - } -} - -defm prefetch : namedimm<"prefetch", "A64PRFM::PRFMMapper">; - -class A64I_LDRlitSimple opc, bit v, RegisterClass OutReg, - list patterns = []> - : A64I_LDRlit; - -let mayLoad = 1 in { - def LDRw_lit : A64I_LDRlitSimple<0b00, 0b0, GPR32, - [(set (i32 GPR32:$Rt), (load constpool:$Imm19))]>; - def LDRx_lit : A64I_LDRlitSimple<0b01, 0b0, GPR64, - [(set (i64 GPR64:$Rt), (load constpool:$Imm19))]>; -} - -def LDRs_lit : A64I_LDRlitSimple<0b00, 0b1, FPR32, - [(set (f32 FPR32:$Rt), (load constpool:$Imm19))]>; -def LDRd_lit : A64I_LDRlitSimple<0b01, 0b1, FPR64, - [(set (f64 FPR64:$Rt), (load constpool:$Imm19))]>; - -let mayLoad = 1 in { - def LDRq_lit : A64I_LDRlitSimple<0b10, 0b1, FPR128>; - - - def LDRSWx_lit : A64I_LDRlit<0b10, 0b0, - (outs GPR64:$Rt), - (ins ldrlit_label:$Imm19), - "ldrsw\t$Rt, $Imm19", - [], NoItinerary>; - - def PRFM_lit : A64I_LDRlit<0b11, 0b0, - (outs), (ins prefetch_op:$Rt, ldrlit_label:$Imm19), - "prfm\t$Rt, $Imm19", - [], NoItinerary>; -} - -//===----------------------------------------------------------------------===// -// Load-store exclusive instructions -//===----------------------------------------------------------------------===// -// Contains: STXRB, STXRH, STXR, LDXRB, LDXRH, LDXR. STXP, LDXP, STLXRB, -// STLXRH, STLXR, LDAXRB, LDAXRH, LDAXR, STLXP, LDAXP, STLRB, -// STLRH, STLR, LDARB, LDARH, LDAR - -// Since these instructions have the undefined register bits set to 1 in -// their canonical form, we need a post encoder method to set those bits -// to 1 when encoding these instructions. We do this using the -// fixLoadStoreExclusive function. This function has template parameters: -// -// fixLoadStoreExclusive -// -// hasRs indicates that the instruction uses the Rs field, so we won't set -// it to 1 (and the same for Rt2). We don't need template parameters for -// the other register fiels since Rt and Rn are always used. - -// This operand parses a GPR64xsp register, followed by an optional immediate -// #0. -def GPR64xsp0_asmoperand : AsmOperandClass { - let Name = "GPR64xsp0"; - let PredicateMethod = "isWrappedReg"; - let RenderMethod = "addRegOperands"; - let ParserMethod = "ParseLSXAddressOperand"; - // Diagnostics are provided by ParserMethod -} - -def GPR64xsp0 : RegisterOperand { - let ParserMatchClass = GPR64xsp0_asmoperand; -} - -//===---------------------------------- -// Store-exclusive (releasing & normal) -//===---------------------------------- - -class A64I_SRexs_impl size, bits<3> opcode, string asm, dag outs, - dag ins, list pat, - InstrItinClass itin> : - A64I_LDSTex_stn { - let mayStore = 1; - let PostEncoderMethod = "fixLoadStoreExclusive<1,0>"; -} - -multiclass A64I_SRex opcode, string prefix> { - def _byte: A64I_SRexs_impl<0b00, opcode, !strconcat(asmstr, "b"), - (outs GPR32:$Rs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [], NoItinerary>; - - def _hword: A64I_SRexs_impl<0b01, opcode, !strconcat(asmstr, "h"), - (outs GPR32:$Rs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [],NoItinerary>; - - def _word: A64I_SRexs_impl<0b10, opcode, asmstr, - (outs GPR32:$Rs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [], NoItinerary>; - - def _dword: A64I_SRexs_impl<0b11, opcode, asmstr, - (outs GPR32:$Rs), (ins GPR64:$Rt, GPR64xsp0:$Rn), - [], NoItinerary>; -} - -defm STXR : A64I_SRex<"stxr", 0b000, "STXR">; -defm STLXR : A64I_SRex<"stlxr", 0b001, "STLXR">; - -//===---------------------------------- -// Loads -//===---------------------------------- - -class A64I_LRexs_impl size, bits<3> opcode, string asm, dag outs, - dag ins, list pat, - InstrItinClass itin> : - A64I_LDSTex_tn { - let mayLoad = 1; - let PostEncoderMethod = "fixLoadStoreExclusive<0,0>"; -} - -multiclass A64I_LRex opcode> { - def _byte: A64I_LRexs_impl<0b00, opcode, !strconcat(asmstr, "b"), - (outs GPR32:$Rt), (ins GPR64xsp0:$Rn), - [], NoItinerary>; - - def _hword: A64I_LRexs_impl<0b01, opcode, !strconcat(asmstr, "h"), - (outs GPR32:$Rt), (ins GPR64xsp0:$Rn), - [], NoItinerary>; - - def _word: A64I_LRexs_impl<0b10, opcode, asmstr, - (outs GPR32:$Rt), (ins GPR64xsp0:$Rn), - [], NoItinerary>; - - def _dword: A64I_LRexs_impl<0b11, opcode, asmstr, - (outs GPR64:$Rt), (ins GPR64xsp0:$Rn), - [], NoItinerary>; -} - -defm LDXR : A64I_LRex<"ldxr", 0b000>; -defm LDAXR : A64I_LRex<"ldaxr", 0b001>; -defm LDAR : A64I_LRex<"ldar", 0b101>; - -class acquiring_load - : PatFrag<(ops node:$ptr), (base node:$ptr), [{ - return cast(N)->getOrdering() == Acquire; +def VecIndex_x4 : SDNodeXFormgetTargetConstant(4 * N->getZExtValue(), SDLoc(N), MVT::i64); }]>; - -def atomic_load_acquire_8 : acquiring_load; -def atomic_load_acquire_16 : acquiring_load; -def atomic_load_acquire_32 : acquiring_load; -def atomic_load_acquire_64 : acquiring_load; - -def : Pat<(atomic_load_acquire_8 GPR64xsp:$Rn), (LDAR_byte GPR64xsp0:$Rn)>; -def : Pat<(atomic_load_acquire_16 GPR64xsp:$Rn), (LDAR_hword GPR64xsp0:$Rn)>; -def : Pat<(atomic_load_acquire_32 GPR64xsp:$Rn), (LDAR_word GPR64xsp0:$Rn)>; -def : Pat<(atomic_load_acquire_64 GPR64xsp:$Rn), (LDAR_dword GPR64xsp0:$Rn)>; - -//===---------------------------------- -// Store-release (no exclusivity) -//===---------------------------------- - -class A64I_SLexs_impl size, bits<3> opcode, string asm, dag outs, - dag ins, list pat, - InstrItinClass itin> : - A64I_LDSTex_tn { - let mayStore = 1; - let PostEncoderMethod = "fixLoadStoreExclusive<0,0>"; -} - -class releasing_store - : PatFrag<(ops node:$ptr, node:$val), (base node:$ptr, node:$val), [{ - return cast(N)->getOrdering() == Release; +def VecIndex_x8 : SDNodeXFormgetTargetConstant(8 * N->getZExtValue(), SDLoc(N), MVT::i64); }]>; -def atomic_store_release_8 : releasing_store; -def atomic_store_release_16 : releasing_store; -def atomic_store_release_32 : releasing_store; -def atomic_store_release_64 : releasing_store; - -multiclass A64I_SLex opcode, string prefix> { - def _byte: A64I_SLexs_impl<0b00, opcode, !strconcat(asmstr, "b"), - (outs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [(atomic_store_release_8 GPR64xsp0:$Rn, GPR32:$Rt)], - NoItinerary>; - - def _hword: A64I_SLexs_impl<0b01, opcode, !strconcat(asmstr, "h"), - (outs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [(atomic_store_release_16 GPR64xsp0:$Rn, GPR32:$Rt)], - NoItinerary>; - - def _word: A64I_SLexs_impl<0b10, opcode, asmstr, - (outs), (ins GPR32:$Rt, GPR64xsp0:$Rn), - [(atomic_store_release_32 GPR64xsp0:$Rn, GPR32:$Rt)], - NoItinerary>; - - def _dword: A64I_SLexs_impl<0b11, opcode, asmstr, - (outs), (ins GPR64:$Rt, GPR64xsp0:$Rn), - [(atomic_store_release_64 GPR64xsp0:$Rn, GPR64:$Rt)], - NoItinerary>; -} - -defm STLR : A64I_SLex<"stlr", 0b101, "STLR">; - -//===---------------------------------- -// Store-exclusive pair (releasing & normal) -//===---------------------------------- - -class A64I_SPexs_impl size, bits<3> opcode, string asm, dag outs, - dag ins, list pat, - InstrItinClass itin> : - A64I_LDSTex_stt2n { - let mayStore = 1; -} - - -multiclass A64I_SPex opcode> { - def _word: A64I_SPexs_impl<0b10, opcode, asmstr, (outs), - (ins GPR32:$Rs, GPR32:$Rt, GPR32:$Rt2, - GPR64xsp0:$Rn), - [], NoItinerary>; - - def _dword: A64I_SPexs_impl<0b11, opcode, asmstr, (outs), - (ins GPR32:$Rs, GPR64:$Rt, GPR64:$Rt2, - GPR64xsp0:$Rn), - [], NoItinerary>; -} - -defm STXP : A64I_SPex<"stxp", 0b010>; -defm STLXP : A64I_SPex<"stlxp", 0b011>; - -//===---------------------------------- -// Load-exclusive pair (acquiring & normal) -//===---------------------------------- - -class A64I_LPexs_impl size, bits<3> opcode, string asm, dag outs, - dag ins, list pat, - InstrItinClass itin> : - A64I_LDSTex_tt2n { - let mayLoad = 1; - let DecoderMethod = "DecodeLoadPairExclusiveInstruction"; - let PostEncoderMethod = "fixLoadStoreExclusive<0,1>"; -} - -multiclass A64I_LPex opcode> { - def _word: A64I_LPexs_impl<0b10, opcode, asmstr, - (outs GPR32:$Rt, GPR32:$Rt2), - (ins GPR64xsp0:$Rn), - [], NoItinerary>; - - def _dword: A64I_LPexs_impl<0b11, opcode, asmstr, - (outs GPR64:$Rt, GPR64:$Rt2), - (ins GPR64xsp0:$Rn), - [], NoItinerary>; -} - -defm LDXP : A64I_LPex<"ldxp", 0b010>; -defm LDAXP : A64I_LPex<"ldaxp", 0b011>; - -//===----------------------------------------------------------------------===// -// Load-store register (unscaled immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: LDURB, LDURH, LDRUSB, LDRUSH, LDRUSW, STUR, STURB, STURH and PRFUM -// -// and -// -//===----------------------------------------------------------------------===// -// Load-store register (register offset) instructions -//===----------------------------------------------------------------------===// -// Contains: LDRB, LDRH, LDRSB, LDRSH, LDRSW, STR, STRB, STRH and PRFM -// -// and -// -//===----------------------------------------------------------------------===// -// Load-store register (unsigned immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: LDRB, LDRH, LDRSB, LDRSH, LDRSW, STR, STRB, STRH and PRFM -// -// and -// -//===----------------------------------------------------------------------===// -// Load-store register (immediate post-indexed) instructions -//===----------------------------------------------------------------------===// -// Contains: STRB, STRH, STR, LDRB, LDRH, LDR, LDRSB, LDRSH, LDRSW -// -// and -// -//===----------------------------------------------------------------------===// -// Load-store register (immediate pre-indexed) instructions -//===----------------------------------------------------------------------===// -// Contains: STRB, STRH, STR, LDRB, LDRH, LDR, LDRSB, LDRSH, LDRSW - -// Note that patterns are much later on in a completely separate section (they -// need ADRPxi to be defined). - -//===------------------------------- -// 1. Various operands needed -//===------------------------------- - -//===------------------------------- -// 1.1 Unsigned 12-bit immediate operands -//===------------------------------- -// The addressing mode for these instructions consists of an unsigned 12-bit -// immediate which is scaled by the size of the memory access. -// -// We represent this in the MC layer by two operands: -// 1. A base register. -// 2. A 12-bit immediate: not multiplied by access size, so "LDR x0,[x0,#8]" -// would have '1' in this field. -// This means that separate functions are needed for converting representations -// which *are* aware of the intended access size. - -// Anything that creates an MCInst (Decoding, selection and AsmParsing) has to -// know the access size via some means. An isolated operand does not have this -// information unless told from here, which means we need separate tablegen -// Operands for each access size. This multiclass takes care of instantiating -// the correct template functions in the rest of the backend. - -//===------------------------------- -// 1.1 Unsigned 12-bit immediate operands -//===------------------------------- - -multiclass offsets_uimm12 { - def uimm12_asmoperand : AsmOperandClass { - let Name = "OffsetUImm12_" # MemSize; - let PredicateMethod = "isOffsetUImm12<" # MemSize # ">"; - let RenderMethod = "addOffsetUImm12Operands<" # MemSize # ">"; - let DiagnosticType = "LoadStoreUImm12_" # MemSize; - } - - // Pattern is really no more than an ImmLeaf, but predicated on MemSize which - // complicates things beyond TableGen's ken. - def uimm12 : Operand, - ComplexPattern"> { - let ParserMatchClass - = !cast(prefix # uimm12_asmoperand); - - let PrintMethod = "printOffsetUImm12Operand<" # MemSize # ">"; - let EncoderMethod = "getOffsetUImm12OpValue<" # MemSize # ">"; - } -} - -defm byte_ : offsets_uimm12<1, "byte_">; -defm hword_ : offsets_uimm12<2, "hword_">; -defm word_ : offsets_uimm12<4, "word_">; -defm dword_ : offsets_uimm12<8, "dword_">; -defm qword_ : offsets_uimm12<16, "qword_">; - -//===------------------------------- -// 1.1 Signed 9-bit immediate operands -//===------------------------------- +multiclass DUPWithTruncPats { + def : Pat<(ResVT (AArch64dup (ScalVT (vector_extract (Src128VT V128:$Rn), + imm:$idx)))), + (DUP V128:$Rn, (IdxXFORM imm:$idx))>; + + def : Pat<(ResVT (AArch64dup (ScalVT (vector_extract (Src64VT V64:$Rn), + imm:$idx)))), + (DUP (SUBREG_TO_REG (i64 0), V64:$Rn, dsub), (IdxXFORM imm:$idx))>; +} + +defm : DUPWithTruncPats; +defm : DUPWithTruncPats; +defm : DUPWithTruncPats; + +defm : DUPWithTruncPats; +defm : DUPWithTruncPats; +defm : DUPWithTruncPats; + +multiclass DUPWithTrunci64Pats { + def : Pat<(ResVT (AArch64dup (i32 (trunc (extractelt (v2i64 V128:$Rn), + imm:$idx))))), + (DUP V128:$Rn, (IdxXFORM imm:$idx))>; + + def : Pat<(ResVT (AArch64dup (i32 (trunc (extractelt (v1i64 V64:$Rn), + imm:$idx))))), + (DUP (SUBREG_TO_REG (i64 0), V64:$Rn, dsub), (IdxXFORM imm:$idx))>; +} + +defm : DUPWithTrunci64Pats; +defm : DUPWithTrunci64Pats; +defm : DUPWithTrunci64Pats; + +defm : DUPWithTrunci64Pats; +defm : DUPWithTrunci64Pats; +defm : DUPWithTrunci64Pats; + +// SMOV and UMOV definitions, with some extra patterns for convenience +defm SMOV : SMov; +defm UMOV : UMov; + +def : Pat<(sext_inreg (vector_extract (v16i8 V128:$Rn), VectorIndexB:$idx), i8), + (i32 (SMOVvi8to32 V128:$Rn, VectorIndexB:$idx))>; +def : Pat<(sext_inreg (vector_extract (v16i8 V128:$Rn), VectorIndexB:$idx), i8), + (i64 (SMOVvi8to64 V128:$Rn, VectorIndexB:$idx))>; +def : Pat<(sext_inreg (vector_extract (v8i16 V128:$Rn), VectorIndexH:$idx),i16), + (i32 (SMOVvi16to32 V128:$Rn, VectorIndexH:$idx))>; +def : Pat<(sext_inreg (vector_extract (v8i16 V128:$Rn), VectorIndexH:$idx),i16), + (i64 (SMOVvi16to64 V128:$Rn, VectorIndexH:$idx))>; +def : Pat<(sext_inreg (vector_extract (v8i16 V128:$Rn), VectorIndexH:$idx),i16), + (i32 (SMOVvi16to32 V128:$Rn, VectorIndexH:$idx))>; +def : Pat<(sext (i32 (vector_extract (v4i32 V128:$Rn), VectorIndexS:$idx))), + (i64 (SMOVvi32to64 V128:$Rn, VectorIndexS:$idx))>; + +// Extracting i8 or i16 elements will have the zero-extend transformed to +// an 'and' mask by type legalization since neither i8 nor i16 are legal types +// for AArch64. Match these patterns here since UMOV already zeroes out the high +// bits of the destination register. +def : Pat<(and (vector_extract (v16i8 V128:$Rn), VectorIndexB:$idx), + (i32 0xff)), + (i32 (UMOVvi8 V128:$Rn, VectorIndexB:$idx))>; +def : Pat<(and (vector_extract (v8i16 V128:$Rn), VectorIndexH:$idx), + (i32 0xffff)), + (i32 (UMOVvi16 V128:$Rn, VectorIndexH:$idx))>; + +defm INS : SIMDIns; + +def : Pat<(v16i8 (scalar_to_vector GPR32:$Rn)), + (SUBREG_TO_REG (i32 0), + (f32 (COPY_TO_REGCLASS GPR32:$Rn, FPR32)), ssub)>; +def : Pat<(v8i8 (scalar_to_vector GPR32:$Rn)), + (SUBREG_TO_REG (i32 0), + (f32 (COPY_TO_REGCLASS GPR32:$Rn, FPR32)), ssub)>; + +def : Pat<(v8i16 (scalar_to_vector GPR32:$Rn)), + (SUBREG_TO_REG (i32 0), + (f32 (COPY_TO_REGCLASS GPR32:$Rn, FPR32)), ssub)>; +def : Pat<(v4i16 (scalar_to_vector GPR32:$Rn)), + (SUBREG_TO_REG (i32 0), + (f32 (COPY_TO_REGCLASS GPR32:$Rn, FPR32)), ssub)>; + +def : Pat<(v2i32 (scalar_to_vector (i32 FPR32:$Rn))), + (v2i32 (INSERT_SUBREG (v2i32 (IMPLICIT_DEF)), + (i32 FPR32:$Rn), ssub))>; +def : Pat<(v4i32 (scalar_to_vector (i32 FPR32:$Rn))), + (v4i32 (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), + (i32 FPR32:$Rn), ssub))>; +def : Pat<(v2i64 (scalar_to_vector (i64 FPR64:$Rn))), + (v2i64 (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)), + (i64 FPR64:$Rn), dsub))>; + +def : Pat<(v4f16 (scalar_to_vector (f16 FPR16:$Rn))), + (INSERT_SUBREG (v4f16 (IMPLICIT_DEF)), FPR16:$Rn, hsub)>; +def : Pat<(v8f16 (scalar_to_vector (f16 FPR16:$Rn))), + (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), FPR16:$Rn, hsub)>; + +def : Pat<(v4f32 (scalar_to_vector (f32 FPR32:$Rn))), + (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FPR32:$Rn, ssub)>; +def : Pat<(v2f32 (scalar_to_vector (f32 FPR32:$Rn))), + (INSERT_SUBREG (v2f32 (IMPLICIT_DEF)), FPR32:$Rn, ssub)>; +def : Pat<(v2f64 (scalar_to_vector (f64 FPR64:$Rn))), + (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FPR64:$Rn, dsub)>; + +def : Pat<(v4f16 (vector_insert (v4f16 V64:$Rn), + (f16 FPR16:$Rm), (i64 VectorIndexS:$imm))), + (EXTRACT_SUBREG + (INSvi16lane + (v8f16 (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), V64:$Rn, dsub)), + VectorIndexS:$imm, + (v8f16 (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), FPR16:$Rm, hsub)), + (i64 0)), + dsub)>; + +def : Pat<(v8f16 (vector_insert (v8f16 V128:$Rn), + (f16 FPR16:$Rm), (i64 VectorIndexH:$imm))), + (INSvi16lane + V128:$Rn, VectorIndexH:$imm, + (v8f16 (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), FPR16:$Rm, hsub)), + (i64 0))>; + +def : Pat<(v2f32 (vector_insert (v2f32 V64:$Rn), + (f32 FPR32:$Rm), (i64 VectorIndexS:$imm))), + (EXTRACT_SUBREG + (INSvi32lane + (v4f32 (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), V64:$Rn, dsub)), + VectorIndexS:$imm, + (v4f32 (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FPR32:$Rm, ssub)), + (i64 0)), + dsub)>; +def : Pat<(v4f32 (vector_insert (v4f32 V128:$Rn), + (f32 FPR32:$Rm), (i64 VectorIndexS:$imm))), + (INSvi32lane + V128:$Rn, VectorIndexS:$imm, + (v4f32 (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FPR32:$Rm, ssub)), + (i64 0))>; +def : Pat<(v2f64 (vector_insert (v2f64 V128:$Rn), + (f64 FPR64:$Rm), (i64 VectorIndexD:$imm))), + (INSvi64lane + V128:$Rn, VectorIndexD:$imm, + (v2f64 (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FPR64:$Rm, dsub)), + (i64 0))>; + +// Copy an element at a constant index in one vector into a constant indexed +// element of another. +// FIXME refactor to a shared class/dev parameterized on vector type, vector +// index type and INS extension +def : Pat<(v16i8 (int_aarch64_neon_vcopy_lane + (v16i8 V128:$Vd), VectorIndexB:$idx, (v16i8 V128:$Vs), + VectorIndexB:$idx2)), + (v16i8 (INSvi8lane + V128:$Vd, VectorIndexB:$idx, V128:$Vs, VectorIndexB:$idx2) + )>; +def : Pat<(v8i16 (int_aarch64_neon_vcopy_lane + (v8i16 V128:$Vd), VectorIndexH:$idx, (v8i16 V128:$Vs), + VectorIndexH:$idx2)), + (v8i16 (INSvi16lane + V128:$Vd, VectorIndexH:$idx, V128:$Vs, VectorIndexH:$idx2) + )>; +def : Pat<(v4i32 (int_aarch64_neon_vcopy_lane + (v4i32 V128:$Vd), VectorIndexS:$idx, (v4i32 V128:$Vs), + VectorIndexS:$idx2)), + (v4i32 (INSvi32lane + V128:$Vd, VectorIndexS:$idx, V128:$Vs, VectorIndexS:$idx2) + )>; +def : Pat<(v2i64 (int_aarch64_neon_vcopy_lane + (v2i64 V128:$Vd), VectorIndexD:$idx, (v2i64 V128:$Vs), + VectorIndexD:$idx2)), + (v2i64 (INSvi64lane + V128:$Vd, VectorIndexD:$idx, V128:$Vs, VectorIndexD:$idx2) + )>; + +multiclass Neon_INS_elt_pattern { + def : Pat<(VT128 (vector_insert V128:$src, + (VTScal (vector_extract (VT128 V128:$Rn), imm:$Immn)), + imm:$Immd)), + (INS V128:$src, imm:$Immd, V128:$Rn, imm:$Immn)>; + + def : Pat<(VT128 (vector_insert V128:$src, + (VTScal (vector_extract (VT64 V64:$Rn), imm:$Immn)), + imm:$Immd)), + (INS V128:$src, imm:$Immd, + (SUBREG_TO_REG (i64 0), V64:$Rn, dsub), imm:$Immn)>; + + def : Pat<(VT64 (vector_insert V64:$src, + (VTScal (vector_extract (VT128 V128:$Rn), imm:$Immn)), + imm:$Immd)), + (EXTRACT_SUBREG (INS (SUBREG_TO_REG (i64 0), V64:$src, dsub), + imm:$Immd, V128:$Rn, imm:$Immn), + dsub)>; + + def : Pat<(VT64 (vector_insert V64:$src, + (VTScal (vector_extract (VT64 V64:$Rn), imm:$Immn)), + imm:$Immd)), + (EXTRACT_SUBREG + (INS (SUBREG_TO_REG (i64 0), V64:$src, dsub), imm:$Immd, + (SUBREG_TO_REG (i64 0), V64:$Rn, dsub), imm:$Immn), + dsub)>; +} + +defm : Neon_INS_elt_pattern; +defm : Neon_INS_elt_pattern; +defm : Neon_INS_elt_pattern; + + +// Floating point vector extractions are codegen'd as either a sequence of +// subregister extractions, or a MOV (aka CPY here, alias for DUP) if +// the lane number is anything other than zero. +def : Pat<(vector_extract (v2f64 V128:$Rn), 0), + (f64 (EXTRACT_SUBREG V128:$Rn, dsub))>; +def : Pat<(vector_extract (v4f32 V128:$Rn), 0), + (f32 (EXTRACT_SUBREG V128:$Rn, ssub))>; +def : Pat<(vector_extract (v8f16 V128:$Rn), 0), + (f16 (EXTRACT_SUBREG V128:$Rn, hsub))>; + +def : Pat<(vector_extract (v2f64 V128:$Rn), VectorIndexD:$idx), + (f64 (CPYi64 V128:$Rn, VectorIndexD:$idx))>; +def : Pat<(vector_extract (v4f32 V128:$Rn), VectorIndexS:$idx), + (f32 (CPYi32 V128:$Rn, VectorIndexS:$idx))>; +def : Pat<(vector_extract (v8f16 V128:$Rn), VectorIndexH:$idx), + (f16 (CPYi16 V128:$Rn, VectorIndexH:$idx))>; + +// All concat_vectors operations are canonicalised to act on i64 vectors for +// AArch64. In the general case we need an instruction, which had just as well be +// INS. +class ConcatPat + : Pat<(DstTy (concat_vectors (SrcTy V64:$Rd), V64:$Rn)), + (INSvi64lane (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), 1, + (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rn, dsub), 0)>; + +def : ConcatPat; +def : ConcatPat; +def : ConcatPat; +def : ConcatPat; +def : ConcatPat; +def : ConcatPat; +def : ConcatPat; + +// If the high lanes are undef, though, we can just ignore them: +class ConcatUndefPat + : Pat<(DstTy (concat_vectors (SrcTy V64:$Rn), undef)), + (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rn, dsub)>; + +def : ConcatUndefPat; +def : ConcatUndefPat; +def : ConcatUndefPat; +def : ConcatUndefPat; +def : ConcatUndefPat; +def : ConcatUndefPat; + +//---------------------------------------------------------------------------- +// AdvSIMD across lanes instructions +//---------------------------------------------------------------------------- + +defm ADDV : SIMDAcrossLanesBHS<0, 0b11011, "addv">; +defm SMAXV : SIMDAcrossLanesBHS<0, 0b01010, "smaxv">; +defm SMINV : SIMDAcrossLanesBHS<0, 0b11010, "sminv">; +defm UMAXV : SIMDAcrossLanesBHS<1, 0b01010, "umaxv">; +defm UMINV : SIMDAcrossLanesBHS<1, 0b11010, "uminv">; +defm SADDLV : SIMDAcrossLanesHSD<0, 0b00011, "saddlv">; +defm UADDLV : SIMDAcrossLanesHSD<1, 0b00011, "uaddlv">; +defm FMAXNMV : SIMDFPAcrossLanes<0b01100, 0, "fmaxnmv", int_aarch64_neon_fmaxnmv>; +defm FMAXV : SIMDFPAcrossLanes<0b01111, 0, "fmaxv", int_aarch64_neon_fmaxv>; +defm FMINNMV : SIMDFPAcrossLanes<0b01100, 1, "fminnmv", int_aarch64_neon_fminnmv>; +defm FMINV : SIMDFPAcrossLanes<0b01111, 1, "fminv", int_aarch64_neon_fminv>; + +// Patterns for across-vector intrinsics, that have a node equivalent, that +// returns a vector (with only the low lane defined) instead of a scalar. +// In effect, opNode is the same as (scalar_to_vector (IntNode)). +multiclass SIMDAcrossLanesIntrinsic { +// If a lane instruction caught the vector_extract around opNode, we can +// directly match the latter to the instruction. +def : Pat<(v8i8 (opNode V64:$Rn)), + (INSERT_SUBREG (v8i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), bsub)>; +def : Pat<(v16i8 (opNode V128:$Rn)), + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), bsub)>; +def : Pat<(v4i16 (opNode V64:$Rn)), + (INSERT_SUBREG (v4i16 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), hsub)>; +def : Pat<(v8i16 (opNode V128:$Rn)), + (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), hsub)>; +def : Pat<(v4i32 (opNode V128:$Rn)), + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i32v")) V128:$Rn), ssub)>; + + +// If none did, fallback to the explicit patterns, consuming the vector_extract. +def : Pat<(i32 (vector_extract (insert_subvector undef, (v8i8 (opNode V64:$Rn)), + (i32 0)), (i64 0))), + (EXTRACT_SUBREG (INSERT_SUBREG (v8i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), + bsub), ssub)>; +def : Pat<(i32 (vector_extract (v16i8 (opNode V128:$Rn)), (i64 0))), + (EXTRACT_SUBREG (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), + bsub), ssub)>; +def : Pat<(i32 (vector_extract (insert_subvector undef, + (v4i16 (opNode V64:$Rn)), (i32 0)), (i64 0))), + (EXTRACT_SUBREG (INSERT_SUBREG (v4i16 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), + hsub), ssub)>; +def : Pat<(i32 (vector_extract (v8i16 (opNode V128:$Rn)), (i64 0))), + (EXTRACT_SUBREG (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), + hsub), ssub)>; +def : Pat<(i32 (vector_extract (v4i32 (opNode V128:$Rn)), (i64 0))), + (EXTRACT_SUBREG (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i32v")) V128:$Rn), + ssub), ssub)>; + +} + +multiclass SIMDAcrossLanesSignedIntrinsic + : SIMDAcrossLanesIntrinsic { +// If there is a sign extension after this intrinsic, consume it as smov already +// performed it +def : Pat<(i32 (sext_inreg (i32 (vector_extract (insert_subvector undef, + (opNode (v8i8 V64:$Rn)), (i32 0)), (i64 0))), i8)), + (i32 (SMOVvi8to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), bsub), + (i64 0)))>; +def : Pat<(i32 (sext_inreg (i32 (vector_extract + (opNode (v16i8 V128:$Rn)), (i64 0))), i8)), + (i32 (SMOVvi8to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), bsub), + (i64 0)))>; +def : Pat<(i32 (sext_inreg (i32 (vector_extract (insert_subvector undef, + (opNode (v4i16 V64:$Rn)), (i32 0)), (i64 0))), i16)), + (i32 (SMOVvi16to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), hsub), + (i64 0)))>; +def : Pat<(i32 (sext_inreg (i32 (vector_extract + (opNode (v8i16 V128:$Rn)), (i64 0))), i16)), + (i32 (SMOVvi16to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), hsub), + (i64 0)))>; +} + +multiclass SIMDAcrossLanesUnsignedIntrinsic + : SIMDAcrossLanesIntrinsic { +// If there is a masking operation keeping only what has been actually +// generated, consume it. +def : Pat<(i32 (and (i32 (vector_extract (insert_subvector undef, + (opNode (v8i8 V64:$Rn)), (i32 0)), (i64 0))), maski8_or_more)), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), bsub), + ssub))>; +def : Pat<(i32 (and (i32 (vector_extract (opNode (v16i8 V128:$Rn)), (i64 0))), + maski8_or_more)), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), bsub), + ssub))>; +def : Pat<(i32 (and (i32 (vector_extract (insert_subvector undef, + (opNode (v4i16 V64:$Rn)), (i32 0)), (i64 0))), maski16_or_more)), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), hsub), + ssub))>; +def : Pat<(i32 (and (i32 (vector_extract (opNode (v8i16 V128:$Rn)), (i64 0))), + maski16_or_more)), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), hsub), + ssub))>; +} + +defm : SIMDAcrossLanesSignedIntrinsic<"ADDV", AArch64saddv>; +// vaddv_[su]32 is special; -> ADDP Vd.2S,Vn.2S,Vm.2S; return Vd.s[0];Vn==Vm +def : Pat<(v2i32 (AArch64saddv (v2i32 V64:$Rn))), + (ADDPv2i32 V64:$Rn, V64:$Rn)>; + +defm : SIMDAcrossLanesUnsignedIntrinsic<"ADDV", AArch64uaddv>; +// vaddv_[su]32 is special; -> ADDP Vd.2S,Vn.2S,Vm.2S; return Vd.s[0];Vn==Vm +def : Pat<(v2i32 (AArch64uaddv (v2i32 V64:$Rn))), + (ADDPv2i32 V64:$Rn, V64:$Rn)>; + +defm : SIMDAcrossLanesSignedIntrinsic<"SMAXV", AArch64smaxv>; +def : Pat<(v2i32 (AArch64smaxv (v2i32 V64:$Rn))), + (SMAXPv2i32 V64:$Rn, V64:$Rn)>; + +defm : SIMDAcrossLanesSignedIntrinsic<"SMINV", AArch64sminv>; +def : Pat<(v2i32 (AArch64sminv (v2i32 V64:$Rn))), + (SMINPv2i32 V64:$Rn, V64:$Rn)>; + +defm : SIMDAcrossLanesUnsignedIntrinsic<"UMAXV", AArch64umaxv>; +def : Pat<(v2i32 (AArch64umaxv (v2i32 V64:$Rn))), + (UMAXPv2i32 V64:$Rn, V64:$Rn)>; + +defm : SIMDAcrossLanesUnsignedIntrinsic<"UMINV", AArch64uminv>; +def : Pat<(v2i32 (AArch64uminv (v2i32 V64:$Rn))), + (UMINPv2i32 V64:$Rn, V64:$Rn)>; + +multiclass SIMDAcrossLanesSignedLongIntrinsic { + def : Pat<(i32 (intOp (v8i8 V64:$Rn))), + (i32 (SMOVvi16to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), hsub), + (i64 0)))>; +def : Pat<(i32 (intOp (v16i8 V128:$Rn))), + (i32 (SMOVvi16to32 + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), hsub), + (i64 0)))>; + +def : Pat<(i32 (intOp (v4i16 V64:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), ssub), + ssub))>; +def : Pat<(i32 (intOp (v8i16 V128:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), ssub), + ssub))>; + +def : Pat<(i64 (intOp (v4i32 V128:$Rn))), + (i64 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i32v")) V128:$Rn), dsub), + dsub))>; +} + +multiclass SIMDAcrossLanesUnsignedLongIntrinsic { + def : Pat<(i32 (intOp (v8i8 V64:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i8v")) V64:$Rn), hsub), + ssub))>; +def : Pat<(i32 (intOp (v16i8 V128:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v16i8v")) V128:$Rn), hsub), + ssub))>; + +def : Pat<(i32 (intOp (v4i16 V64:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i16v")) V64:$Rn), ssub), + ssub))>; +def : Pat<(i32 (intOp (v8i16 V128:$Rn))), + (i32 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v8i16v")) V128:$Rn), ssub), + ssub))>; + +def : Pat<(i64 (intOp (v4i32 V128:$Rn))), + (i64 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (!cast(!strconcat(baseOpc, "v4i32v")) V128:$Rn), dsub), + dsub))>; +} + +defm : SIMDAcrossLanesSignedLongIntrinsic<"SADDLV", int_aarch64_neon_saddlv>; +defm : SIMDAcrossLanesUnsignedLongIntrinsic<"UADDLV", int_aarch64_neon_uaddlv>; + +// The vaddlv_s32 intrinsic gets mapped to SADDLP. +def : Pat<(i64 (int_aarch64_neon_saddlv (v2i32 V64:$Rn))), + (i64 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (SADDLPv2i32_v1i64 V64:$Rn), dsub), + dsub))>; +// The vaddlv_u32 intrinsic gets mapped to UADDLP. +def : Pat<(i64 (int_aarch64_neon_uaddlv (v2i32 V64:$Rn))), + (i64 (EXTRACT_SUBREG + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), + (UADDLPv2i32_v1i64 V64:$Rn), dsub), + dsub))>; + +//------------------------------------------------------------------------------ +// AdvSIMD modified immediate instructions +//------------------------------------------------------------------------------ + +// AdvSIMD BIC +defm BIC : SIMDModifiedImmVectorShiftTied<1, 0b11, 0b01, "bic", AArch64bici>; +// AdvSIMD ORR +defm ORR : SIMDModifiedImmVectorShiftTied<0, 0b11, 0b01, "orr", AArch64orri>; + +def : InstAlias<"bic $Vd.4h, $imm", (BICv4i16 V64:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"bic $Vd.8h, $imm", (BICv8i16 V128:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"bic $Vd.2s, $imm", (BICv2i32 V64:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"bic $Vd.4s, $imm", (BICv4i32 V128:$Vd, imm0_255:$imm, 0)>; + +def : InstAlias<"bic.4h $Vd, $imm", (BICv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"bic.8h $Vd, $imm", (BICv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"bic.2s $Vd, $imm", (BICv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"bic.4s $Vd, $imm", (BICv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +def : InstAlias<"orr $Vd.4h, $imm", (ORRv4i16 V64:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"orr $Vd.8h, $imm", (ORRv8i16 V128:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"orr $Vd.2s, $imm", (ORRv2i32 V64:$Vd, imm0_255:$imm, 0)>; +def : InstAlias<"orr $Vd.4s, $imm", (ORRv4i32 V128:$Vd, imm0_255:$imm, 0)>; + +def : InstAlias<"orr.4h $Vd, $imm", (ORRv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"orr.8h $Vd, $imm", (ORRv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"orr.2s $Vd, $imm", (ORRv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"orr.4s $Vd, $imm", (ORRv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +// AdvSIMD FMOV +def FMOVv2f64_ns : SIMDModifiedImmVectorNoShift<1, 1, 0, 0b1111, V128, fpimm8, + "fmov", ".2d", + [(set (v2f64 V128:$Rd), (AArch64fmov imm0_255:$imm8))]>; +def FMOVv2f32_ns : SIMDModifiedImmVectorNoShift<0, 0, 0, 0b1111, V64, fpimm8, + "fmov", ".2s", + [(set (v2f32 V64:$Rd), (AArch64fmov imm0_255:$imm8))]>; +def FMOVv4f32_ns : SIMDModifiedImmVectorNoShift<1, 0, 0, 0b1111, V128, fpimm8, + "fmov", ".4s", + [(set (v4f32 V128:$Rd), (AArch64fmov imm0_255:$imm8))]>; +let Predicates = [HasNEON, HasFullFP16] in { +def FMOVv4f16_ns : SIMDModifiedImmVectorNoShift<0, 0, 1, 0b1111, V64, fpimm8, + "fmov", ".4h", + [(set (v4f16 V64:$Rd), (AArch64fmov imm0_255:$imm8))]>; +def FMOVv8f16_ns : SIMDModifiedImmVectorNoShift<1, 0, 1, 0b1111, V128, fpimm8, + "fmov", ".8h", + [(set (v8f16 V128:$Rd), (AArch64fmov imm0_255:$imm8))]>; +} // Predicates = [HasNEON, HasFullFP16] + +// AdvSIMD MOVI + +// EDIT byte mask: scalar +let isReMaterializable = 1, isAsCheapAsAMove = 1 in +def MOVID : SIMDModifiedImmScalarNoShift<0, 1, 0b1110, "movi", + [(set FPR64:$Rd, simdimmtype10:$imm8)]>; +// The movi_edit node has the immediate value already encoded, so we use +// a plain imm0_255 here. +def : Pat<(f64 (AArch64movi_edit imm0_255:$shift)), + (MOVID imm0_255:$shift)>; + +def : Pat<(v1i64 immAllZerosV), (MOVID (i32 0))>; +def : Pat<(v2i32 immAllZerosV), (MOVID (i32 0))>; +def : Pat<(v4i16 immAllZerosV), (MOVID (i32 0))>; +def : Pat<(v8i8 immAllZerosV), (MOVID (i32 0))>; + +def : Pat<(v1i64 immAllOnesV), (MOVID (i32 255))>; +def : Pat<(v2i32 immAllOnesV), (MOVID (i32 255))>; +def : Pat<(v4i16 immAllOnesV), (MOVID (i32 255))>; +def : Pat<(v8i8 immAllOnesV), (MOVID (i32 255))>; + +// EDIT byte mask: 2d + +// The movi_edit node has the immediate value already encoded, so we use +// a plain imm0_255 in the pattern +let isReMaterializable = 1, isAsCheapAsAMove = 1 in +def MOVIv2d_ns : SIMDModifiedImmVectorNoShift<1, 1, 0, 0b1110, V128, + simdimmtype10, + "movi", ".2d", + [(set (v2i64 V128:$Rd), (AArch64movi_edit imm0_255:$imm8))]>; + + +// Use movi.2d to materialize 0.0 if the HW does zero-cycle zeroing. +// Complexity is added to break a tie with a plain MOVI. +let AddedComplexity = 1 in { +def : Pat<(f32 fpimm0), + (f32 (EXTRACT_SUBREG (v2i64 (MOVIv2d_ns (i32 0))), ssub))>, + Requires<[HasZCZ]>; +def : Pat<(f64 fpimm0), + (f64 (EXTRACT_SUBREG (v2i64 (MOVIv2d_ns (i32 0))), dsub))>, + Requires<[HasZCZ]>; +} + +def : Pat<(v2i64 immAllZerosV), (MOVIv2d_ns (i32 0))>; +def : Pat<(v4i32 immAllZerosV), (MOVIv2d_ns (i32 0))>; +def : Pat<(v8i16 immAllZerosV), (MOVIv2d_ns (i32 0))>; +def : Pat<(v16i8 immAllZerosV), (MOVIv2d_ns (i32 0))>; + +def : Pat<(v2i64 immAllOnesV), (MOVIv2d_ns (i32 255))>; +def : Pat<(v4i32 immAllOnesV), (MOVIv2d_ns (i32 255))>; +def : Pat<(v8i16 immAllOnesV), (MOVIv2d_ns (i32 255))>; +def : Pat<(v16i8 immAllOnesV), (MOVIv2d_ns (i32 255))>; + +def : Pat<(v2f64 (AArch64dup (f64 fpimm0))), (MOVIv2d_ns (i32 0))>; +def : Pat<(v4f32 (AArch64dup (f32 fpimm0))), (MOVIv2d_ns (i32 0))>; + +// EDIT per word & halfword: 2s, 4h, 4s, & 8h +defm MOVI : SIMDModifiedImmVectorShift<0, 0b10, 0b00, "movi">; + +def : InstAlias<"movi $Vd.4h, $imm", (MOVIv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi $Vd.8h, $imm", (MOVIv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi $Vd.2s, $imm", (MOVIv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi $Vd.4s, $imm", (MOVIv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +def : InstAlias<"movi.4h $Vd, $imm", (MOVIv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi.8h $Vd, $imm", (MOVIv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi.2s $Vd, $imm", (MOVIv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"movi.4s $Vd, $imm", (MOVIv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +def : Pat<(v2i32 (AArch64movi_shift imm0_255:$imm8, (i32 imm:$shift))), + (MOVIv2i32 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v4i32 (AArch64movi_shift imm0_255:$imm8, (i32 imm:$shift))), + (MOVIv4i32 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v4i16 (AArch64movi_shift imm0_255:$imm8, (i32 imm:$shift))), + (MOVIv4i16 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v8i16 (AArch64movi_shift imm0_255:$imm8, (i32 imm:$shift))), + (MOVIv8i16 imm0_255:$imm8, imm:$shift)>; + +// EDIT per word: 2s & 4s with MSL shifter +def MOVIv2s_msl : SIMDModifiedImmMoveMSL<0, 0, {1,1,0,?}, V64, "movi", ".2s", + [(set (v2i32 V64:$Rd), + (AArch64movi_msl imm0_255:$imm8, (i32 imm:$shift)))]>; +def MOVIv4s_msl : SIMDModifiedImmMoveMSL<1, 0, {1,1,0,?}, V128, "movi", ".4s", + [(set (v4i32 V128:$Rd), + (AArch64movi_msl imm0_255:$imm8, (i32 imm:$shift)))]>; + +// Per byte: 8b & 16b +def MOVIv8b_ns : SIMDModifiedImmVectorNoShift<0, 0, 0, 0b1110, V64, imm0_255, + "movi", ".8b", + [(set (v8i8 V64:$Rd), (AArch64movi imm0_255:$imm8))]>; +def MOVIv16b_ns : SIMDModifiedImmVectorNoShift<1, 0, 0, 0b1110, V128, imm0_255, + "movi", ".16b", + [(set (v16i8 V128:$Rd), (AArch64movi imm0_255:$imm8))]>; + +// AdvSIMD MVNI + +// EDIT per word & halfword: 2s, 4h, 4s, & 8h +defm MVNI : SIMDModifiedImmVectorShift<1, 0b10, 0b00, "mvni">; + +def : InstAlias<"mvni $Vd.4h, $imm", (MVNIv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni $Vd.8h, $imm", (MVNIv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni $Vd.2s, $imm", (MVNIv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni $Vd.4s, $imm", (MVNIv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +def : InstAlias<"mvni.4h $Vd, $imm", (MVNIv4i16 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni.8h $Vd, $imm", (MVNIv8i16 V128:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni.2s $Vd, $imm", (MVNIv2i32 V64:$Vd, imm0_255:$imm, 0), 0>; +def : InstAlias<"mvni.4s $Vd, $imm", (MVNIv4i32 V128:$Vd, imm0_255:$imm, 0), 0>; + +def : Pat<(v2i32 (AArch64mvni_shift imm0_255:$imm8, (i32 imm:$shift))), + (MVNIv2i32 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v4i32 (AArch64mvni_shift imm0_255:$imm8, (i32 imm:$shift))), + (MVNIv4i32 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v4i16 (AArch64mvni_shift imm0_255:$imm8, (i32 imm:$shift))), + (MVNIv4i16 imm0_255:$imm8, imm:$shift)>; +def : Pat<(v8i16 (AArch64mvni_shift imm0_255:$imm8, (i32 imm:$shift))), + (MVNIv8i16 imm0_255:$imm8, imm:$shift)>; + +// EDIT per word: 2s & 4s with MSL shifter +def MVNIv2s_msl : SIMDModifiedImmMoveMSL<0, 1, {1,1,0,?}, V64, "mvni", ".2s", + [(set (v2i32 V64:$Rd), + (AArch64mvni_msl imm0_255:$imm8, (i32 imm:$shift)))]>; +def MVNIv4s_msl : SIMDModifiedImmMoveMSL<1, 1, {1,1,0,?}, V128, "mvni", ".4s", + [(set (v4i32 V128:$Rd), + (AArch64mvni_msl imm0_255:$imm8, (i32 imm:$shift)))]>; + +//---------------------------------------------------------------------------- +// AdvSIMD indexed element +//---------------------------------------------------------------------------- -// The MCInst is expected to store the bit-wise encoding of the value, -// which amounts to lopping off the extended sign bits. -def SDXF_simm9 : SDNodeXFormgetTargetConstant(N->getZExtValue() & 0x1ff, MVT::i32); +let hasSideEffects = 0 in { + defm FMLA : SIMDFPIndexedTied<0, 0b0001, "fmla">; + defm FMLS : SIMDFPIndexedTied<0, 0b0101, "fmls">; +} + +// NOTE: Operands are reordered in the FMLA/FMLS PatFrags because the +// instruction expects the addend first, while the intrinsic expects it last. + +// On the other hand, there are quite a few valid combinatorial options due to +// the commutativity of multiplication and the fact that (-x) * y = x * (-y). +defm : SIMDFPIndexedTiedPatterns<"FMLA", + TriOpFrag<(fma node:$RHS, node:$MHS, node:$LHS)>>; +defm : SIMDFPIndexedTiedPatterns<"FMLA", + TriOpFrag<(fma node:$MHS, node:$RHS, node:$LHS)>>; + +defm : SIMDFPIndexedTiedPatterns<"FMLS", + TriOpFrag<(fma node:$MHS, (fneg node:$RHS), node:$LHS)> >; +defm : SIMDFPIndexedTiedPatterns<"FMLS", + TriOpFrag<(fma node:$RHS, (fneg node:$MHS), node:$LHS)> >; +defm : SIMDFPIndexedTiedPatterns<"FMLS", + TriOpFrag<(fma (fneg node:$RHS), node:$MHS, node:$LHS)> >; +defm : SIMDFPIndexedTiedPatterns<"FMLS", + TriOpFrag<(fma (fneg node:$MHS), node:$RHS, node:$LHS)> >; + +multiclass FMLSIndexedAfterNegPatterns { + // 3 variants for the .2s version: DUPLANE from 128-bit, DUPLANE from 64-bit + // and DUP scalar. + def : Pat<(v2f32 (OpNode (v2f32 V64:$Rd), (v2f32 V64:$Rn), + (AArch64duplane32 (v4f32 (fneg V128:$Rm)), + VectorIndexS:$idx))), + (FMLSv2i32_indexed V64:$Rd, V64:$Rn, V128:$Rm, VectorIndexS:$idx)>; + def : Pat<(v2f32 (OpNode (v2f32 V64:$Rd), (v2f32 V64:$Rn), + (v2f32 (AArch64duplane32 + (v4f32 (insert_subvector undef, + (v2f32 (fneg V64:$Rm)), + (i32 0))), + VectorIndexS:$idx)))), + (FMLSv2i32_indexed V64:$Rd, V64:$Rn, + (SUBREG_TO_REG (i32 0), V64:$Rm, dsub), + VectorIndexS:$idx)>; + def : Pat<(v2f32 (OpNode (v2f32 V64:$Rd), (v2f32 V64:$Rn), + (AArch64dup (f32 (fneg FPR32Op:$Rm))))), + (FMLSv2i32_indexed V64:$Rd, V64:$Rn, + (SUBREG_TO_REG (i32 0), FPR32Op:$Rm, ssub), (i64 0))>; + + // 3 variants for the .4s version: DUPLANE from 128-bit, DUPLANE from 64-bit + // and DUP scalar. + def : Pat<(v4f32 (OpNode (v4f32 V128:$Rd), (v4f32 V128:$Rn), + (AArch64duplane32 (v4f32 (fneg V128:$Rm)), + VectorIndexS:$idx))), + (FMLSv4i32_indexed V128:$Rd, V128:$Rn, V128:$Rm, + VectorIndexS:$idx)>; + def : Pat<(v4f32 (OpNode (v4f32 V128:$Rd), (v4f32 V128:$Rn), + (v4f32 (AArch64duplane32 + (v4f32 (insert_subvector undef, + (v2f32 (fneg V64:$Rm)), + (i32 0))), + VectorIndexS:$idx)))), + (FMLSv4i32_indexed V128:$Rd, V128:$Rn, + (SUBREG_TO_REG (i32 0), V64:$Rm, dsub), + VectorIndexS:$idx)>; + def : Pat<(v4f32 (OpNode (v4f32 V128:$Rd), (v4f32 V128:$Rn), + (AArch64dup (f32 (fneg FPR32Op:$Rm))))), + (FMLSv4i32_indexed V128:$Rd, V128:$Rn, + (SUBREG_TO_REG (i32 0), FPR32Op:$Rm, ssub), (i64 0))>; + + // 2 variants for the .2d version: DUPLANE from 128-bit, and DUP scalar + // (DUPLANE from 64-bit would be trivial). + def : Pat<(v2f64 (OpNode (v2f64 V128:$Rd), (v2f64 V128:$Rn), + (AArch64duplane64 (v2f64 (fneg V128:$Rm)), + VectorIndexD:$idx))), + (FMLSv2i64_indexed + V128:$Rd, V128:$Rn, V128:$Rm, VectorIndexS:$idx)>; + def : Pat<(v2f64 (OpNode (v2f64 V128:$Rd), (v2f64 V128:$Rn), + (AArch64dup (f64 (fneg FPR64Op:$Rm))))), + (FMLSv2i64_indexed V128:$Rd, V128:$Rn, + (SUBREG_TO_REG (i32 0), FPR64Op:$Rm, dsub), (i64 0))>; + + // 2 variants for 32-bit scalar version: extract from .2s or from .4s + def : Pat<(f32 (OpNode (f32 FPR32:$Rd), (f32 FPR32:$Rn), + (vector_extract (v4f32 (fneg V128:$Rm)), + VectorIndexS:$idx))), + (FMLSv1i32_indexed FPR32:$Rd, FPR32:$Rn, + V128:$Rm, VectorIndexS:$idx)>; + def : Pat<(f32 (OpNode (f32 FPR32:$Rd), (f32 FPR32:$Rn), + (vector_extract (v4f32 (insert_subvector undef, + (v2f32 (fneg V64:$Rm)), + (i32 0))), + VectorIndexS:$idx))), + (FMLSv1i32_indexed FPR32:$Rd, FPR32:$Rn, + (SUBREG_TO_REG (i32 0), V64:$Rm, dsub), VectorIndexS:$idx)>; + + // 1 variant for 64-bit scalar version: extract from .1d or from .2d + def : Pat<(f64 (OpNode (f64 FPR64:$Rd), (f64 FPR64:$Rn), + (vector_extract (v2f64 (fneg V128:$Rm)), + VectorIndexS:$idx))), + (FMLSv1i64_indexed FPR64:$Rd, FPR64:$Rn, + V128:$Rm, VectorIndexS:$idx)>; +} + +defm : FMLSIndexedAfterNegPatterns< + TriOpFrag<(fma node:$RHS, node:$MHS, node:$LHS)> >; +defm : FMLSIndexedAfterNegPatterns< + TriOpFrag<(fma node:$MHS, node:$RHS, node:$LHS)> >; + +defm FMULX : SIMDFPIndexed<1, 0b1001, "fmulx", int_aarch64_neon_fmulx>; +defm FMUL : SIMDFPIndexed<0, 0b1001, "fmul", fmul>; + +def : Pat<(v2f32 (fmul V64:$Rn, (AArch64dup (f32 FPR32:$Rm)))), + (FMULv2i32_indexed V64:$Rn, + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR32:$Rm, ssub), + (i64 0))>; +def : Pat<(v4f32 (fmul V128:$Rn, (AArch64dup (f32 FPR32:$Rm)))), + (FMULv4i32_indexed V128:$Rn, + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR32:$Rm, ssub), + (i64 0))>; +def : Pat<(v2f64 (fmul V128:$Rn, (AArch64dup (f64 FPR64:$Rm)))), + (FMULv2i64_indexed V128:$Rn, + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR64:$Rm, dsub), + (i64 0))>; + +defm SQDMULH : SIMDIndexedHS<0, 0b1100, "sqdmulh", int_aarch64_neon_sqdmulh>; +defm SQRDMULH : SIMDIndexedHS<0, 0b1101, "sqrdmulh", int_aarch64_neon_sqrdmulh>; +defm MLA : SIMDVectorIndexedHSTied<1, 0b0000, "mla", + TriOpFrag<(add node:$LHS, (mul node:$MHS, node:$RHS))>>; +defm MLS : SIMDVectorIndexedHSTied<1, 0b0100, "mls", + TriOpFrag<(sub node:$LHS, (mul node:$MHS, node:$RHS))>>; +defm MUL : SIMDVectorIndexedHS<0, 0b1000, "mul", mul>; +defm SMLAL : SIMDVectorIndexedLongSDTied<0, 0b0010, "smlal", + TriOpFrag<(add node:$LHS, (int_aarch64_neon_smull node:$MHS, node:$RHS))>>; +defm SMLSL : SIMDVectorIndexedLongSDTied<0, 0b0110, "smlsl", + TriOpFrag<(sub node:$LHS, (int_aarch64_neon_smull node:$MHS, node:$RHS))>>; +defm SMULL : SIMDVectorIndexedLongSD<0, 0b1010, "smull", + int_aarch64_neon_smull>; +defm SQDMLAL : SIMDIndexedLongSQDMLXSDTied<0, 0b0011, "sqdmlal", + int_aarch64_neon_sqadd>; +defm SQDMLSL : SIMDIndexedLongSQDMLXSDTied<0, 0b0111, "sqdmlsl", + int_aarch64_neon_sqsub>; +defm SQRDMLAH : SIMDIndexedSQRDMLxHSDTied<1, 0b1101, "sqrdmlah", + int_aarch64_neon_sqadd>; +defm SQRDMLSH : SIMDIndexedSQRDMLxHSDTied<1, 0b1111, "sqrdmlsh", + int_aarch64_neon_sqsub>; +defm SQDMULL : SIMDIndexedLongSD<0, 0b1011, "sqdmull", int_aarch64_neon_sqdmull>; +defm UMLAL : SIMDVectorIndexedLongSDTied<1, 0b0010, "umlal", + TriOpFrag<(add node:$LHS, (int_aarch64_neon_umull node:$MHS, node:$RHS))>>; +defm UMLSL : SIMDVectorIndexedLongSDTied<1, 0b0110, "umlsl", + TriOpFrag<(sub node:$LHS, (int_aarch64_neon_umull node:$MHS, node:$RHS))>>; +defm UMULL : SIMDVectorIndexedLongSD<1, 0b1010, "umull", + int_aarch64_neon_umull>; + +// A scalar sqdmull with the second operand being a vector lane can be +// handled directly with the indexed instruction encoding. +def : Pat<(int_aarch64_neon_sqdmulls_scalar (i32 FPR32:$Rn), + (vector_extract (v4i32 V128:$Vm), + VectorIndexS:$idx)), + (SQDMULLv1i64_indexed FPR32:$Rn, V128:$Vm, VectorIndexS:$idx)>; + +//---------------------------------------------------------------------------- +// AdvSIMD scalar shift instructions +//---------------------------------------------------------------------------- +defm FCVTZS : SIMDFPScalarRShift<0, 0b11111, "fcvtzs">; +defm FCVTZU : SIMDFPScalarRShift<1, 0b11111, "fcvtzu">; +defm SCVTF : SIMDFPScalarRShift<0, 0b11100, "scvtf">; +defm UCVTF : SIMDFPScalarRShift<1, 0b11100, "ucvtf">; +// Codegen patterns for the above. We don't put these directly on the +// instructions because TableGen's type inference can't handle the truth. +// Having the same base pattern for fp <--> int totally freaks it out. +def : Pat<(int_aarch64_neon_vcvtfp2fxs FPR32:$Rn, vecshiftR32:$imm), + (FCVTZSs FPR32:$Rn, vecshiftR32:$imm)>; +def : Pat<(int_aarch64_neon_vcvtfp2fxu FPR32:$Rn, vecshiftR32:$imm), + (FCVTZUs FPR32:$Rn, vecshiftR32:$imm)>; +def : Pat<(i64 (int_aarch64_neon_vcvtfp2fxs (f64 FPR64:$Rn), vecshiftR64:$imm)), + (FCVTZSd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(i64 (int_aarch64_neon_vcvtfp2fxu (f64 FPR64:$Rn), vecshiftR64:$imm)), + (FCVTZUd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(v1i64 (int_aarch64_neon_vcvtfp2fxs (v1f64 FPR64:$Rn), + vecshiftR64:$imm)), + (FCVTZSd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(v1i64 (int_aarch64_neon_vcvtfp2fxu (v1f64 FPR64:$Rn), + vecshiftR64:$imm)), + (FCVTZUd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(int_aarch64_neon_vcvtfxs2fp FPR32:$Rn, vecshiftR32:$imm), + (SCVTFs FPR32:$Rn, vecshiftR32:$imm)>; +def : Pat<(int_aarch64_neon_vcvtfxu2fp FPR32:$Rn, vecshiftR32:$imm), + (UCVTFs FPR32:$Rn, vecshiftR32:$imm)>; +def : Pat<(f64 (int_aarch64_neon_vcvtfxs2fp (i64 FPR64:$Rn), vecshiftR64:$imm)), + (SCVTFd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(f64 (int_aarch64_neon_vcvtfxu2fp (i64 FPR64:$Rn), vecshiftR64:$imm)), + (UCVTFd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(v1f64 (int_aarch64_neon_vcvtfxs2fp (v1i64 FPR64:$Rn), + vecshiftR64:$imm)), + (SCVTFd FPR64:$Rn, vecshiftR64:$imm)>; +def : Pat<(v1f64 (int_aarch64_neon_vcvtfxu2fp (v1i64 FPR64:$Rn), + vecshiftR64:$imm)), + (UCVTFd FPR64:$Rn, vecshiftR64:$imm)>; + +defm SHL : SIMDScalarLShiftD< 0, 0b01010, "shl", AArch64vshl>; +defm SLI : SIMDScalarLShiftDTied<1, 0b01010, "sli">; +defm SQRSHRN : SIMDScalarRShiftBHS< 0, 0b10011, "sqrshrn", + int_aarch64_neon_sqrshrn>; +defm SQRSHRUN : SIMDScalarRShiftBHS< 1, 0b10001, "sqrshrun", + int_aarch64_neon_sqrshrun>; +defm SQSHLU : SIMDScalarLShiftBHSD<1, 0b01100, "sqshlu", AArch64sqshlui>; +defm SQSHL : SIMDScalarLShiftBHSD<0, 0b01110, "sqshl", AArch64sqshli>; +defm SQSHRN : SIMDScalarRShiftBHS< 0, 0b10010, "sqshrn", + int_aarch64_neon_sqshrn>; +defm SQSHRUN : SIMDScalarRShiftBHS< 1, 0b10000, "sqshrun", + int_aarch64_neon_sqshrun>; +defm SRI : SIMDScalarRShiftDTied< 1, 0b01000, "sri">; +defm SRSHR : SIMDScalarRShiftD< 0, 0b00100, "srshr", AArch64srshri>; +defm SRSRA : SIMDScalarRShiftDTied< 0, 0b00110, "srsra", + TriOpFrag<(add node:$LHS, + (AArch64srshri node:$MHS, node:$RHS))>>; +defm SSHR : SIMDScalarRShiftD< 0, 0b00000, "sshr", AArch64vashr>; +defm SSRA : SIMDScalarRShiftDTied< 0, 0b00010, "ssra", + TriOpFrag<(add node:$LHS, + (AArch64vashr node:$MHS, node:$RHS))>>; +defm UQRSHRN : SIMDScalarRShiftBHS< 1, 0b10011, "uqrshrn", + int_aarch64_neon_uqrshrn>; +defm UQSHL : SIMDScalarLShiftBHSD<1, 0b01110, "uqshl", AArch64uqshli>; +defm UQSHRN : SIMDScalarRShiftBHS< 1, 0b10010, "uqshrn", + int_aarch64_neon_uqshrn>; +defm URSHR : SIMDScalarRShiftD< 1, 0b00100, "urshr", AArch64urshri>; +defm URSRA : SIMDScalarRShiftDTied< 1, 0b00110, "ursra", + TriOpFrag<(add node:$LHS, + (AArch64urshri node:$MHS, node:$RHS))>>; +defm USHR : SIMDScalarRShiftD< 1, 0b00000, "ushr", AArch64vlshr>; +defm USRA : SIMDScalarRShiftDTied< 1, 0b00010, "usra", + TriOpFrag<(add node:$LHS, + (AArch64vlshr node:$MHS, node:$RHS))>>; + +//---------------------------------------------------------------------------- +// AdvSIMD vector shift instructions +//---------------------------------------------------------------------------- +defm FCVTZS:SIMDVectorRShiftSD<0, 0b11111, "fcvtzs", int_aarch64_neon_vcvtfp2fxs>; +defm FCVTZU:SIMDVectorRShiftSD<1, 0b11111, "fcvtzu", int_aarch64_neon_vcvtfp2fxu>; +defm SCVTF: SIMDVectorRShiftToFP<0, 0b11100, "scvtf", + int_aarch64_neon_vcvtfxs2fp>; +defm RSHRN : SIMDVectorRShiftNarrowBHS<0, 0b10001, "rshrn", + int_aarch64_neon_rshrn>; +defm SHL : SIMDVectorLShiftBHSD<0, 0b01010, "shl", AArch64vshl>; +defm SHRN : SIMDVectorRShiftNarrowBHS<0, 0b10000, "shrn", + BinOpFrag<(trunc (AArch64vashr node:$LHS, node:$RHS))>>; +defm SLI : SIMDVectorLShiftBHSDTied<1, 0b01010, "sli", int_aarch64_neon_vsli>; +def : Pat<(v1i64 (int_aarch64_neon_vsli (v1i64 FPR64:$Rd), (v1i64 FPR64:$Rn), + (i32 vecshiftL64:$imm))), + (SLId FPR64:$Rd, FPR64:$Rn, vecshiftL64:$imm)>; +defm SQRSHRN : SIMDVectorRShiftNarrowBHS<0, 0b10011, "sqrshrn", + int_aarch64_neon_sqrshrn>; +defm SQRSHRUN: SIMDVectorRShiftNarrowBHS<1, 0b10001, "sqrshrun", + int_aarch64_neon_sqrshrun>; +defm SQSHLU : SIMDVectorLShiftBHSD<1, 0b01100, "sqshlu", AArch64sqshlui>; +defm SQSHL : SIMDVectorLShiftBHSD<0, 0b01110, "sqshl", AArch64sqshli>; +defm SQSHRN : SIMDVectorRShiftNarrowBHS<0, 0b10010, "sqshrn", + int_aarch64_neon_sqshrn>; +defm SQSHRUN : SIMDVectorRShiftNarrowBHS<1, 0b10000, "sqshrun", + int_aarch64_neon_sqshrun>; +defm SRI : SIMDVectorRShiftBHSDTied<1, 0b01000, "sri", int_aarch64_neon_vsri>; +def : Pat<(v1i64 (int_aarch64_neon_vsri (v1i64 FPR64:$Rd), (v1i64 FPR64:$Rn), + (i32 vecshiftR64:$imm))), + (SRId FPR64:$Rd, FPR64:$Rn, vecshiftR64:$imm)>; +defm SRSHR : SIMDVectorRShiftBHSD<0, 0b00100, "srshr", AArch64srshri>; +defm SRSRA : SIMDVectorRShiftBHSDTied<0, 0b00110, "srsra", + TriOpFrag<(add node:$LHS, + (AArch64srshri node:$MHS, node:$RHS))> >; +defm SSHLL : SIMDVectorLShiftLongBHSD<0, 0b10100, "sshll", + BinOpFrag<(AArch64vshl (sext node:$LHS), node:$RHS)>>; + +defm SSHR : SIMDVectorRShiftBHSD<0, 0b00000, "sshr", AArch64vashr>; +defm SSRA : SIMDVectorRShiftBHSDTied<0, 0b00010, "ssra", + TriOpFrag<(add node:$LHS, (AArch64vashr node:$MHS, node:$RHS))>>; +defm UCVTF : SIMDVectorRShiftToFP<1, 0b11100, "ucvtf", + int_aarch64_neon_vcvtfxu2fp>; +defm UQRSHRN : SIMDVectorRShiftNarrowBHS<1, 0b10011, "uqrshrn", + int_aarch64_neon_uqrshrn>; +defm UQSHL : SIMDVectorLShiftBHSD<1, 0b01110, "uqshl", AArch64uqshli>; +defm UQSHRN : SIMDVectorRShiftNarrowBHS<1, 0b10010, "uqshrn", + int_aarch64_neon_uqshrn>; +defm URSHR : SIMDVectorRShiftBHSD<1, 0b00100, "urshr", AArch64urshri>; +defm URSRA : SIMDVectorRShiftBHSDTied<1, 0b00110, "ursra", + TriOpFrag<(add node:$LHS, + (AArch64urshri node:$MHS, node:$RHS))> >; +defm USHLL : SIMDVectorLShiftLongBHSD<1, 0b10100, "ushll", + BinOpFrag<(AArch64vshl (zext node:$LHS), node:$RHS)>>; +defm USHR : SIMDVectorRShiftBHSD<1, 0b00000, "ushr", AArch64vlshr>; +defm USRA : SIMDVectorRShiftBHSDTied<1, 0b00010, "usra", + TriOpFrag<(add node:$LHS, (AArch64vlshr node:$MHS, node:$RHS))> >; + +// SHRN patterns for when a logical right shift was used instead of arithmetic +// (the immediate guarantees no sign bits actually end up in the result so it +// doesn't matter). +def : Pat<(v8i8 (trunc (AArch64vlshr (v8i16 V128:$Rn), vecshiftR16Narrow:$imm))), + (SHRNv8i8_shift V128:$Rn, vecshiftR16Narrow:$imm)>; +def : Pat<(v4i16 (trunc (AArch64vlshr (v4i32 V128:$Rn), vecshiftR32Narrow:$imm))), + (SHRNv4i16_shift V128:$Rn, vecshiftR32Narrow:$imm)>; +def : Pat<(v2i32 (trunc (AArch64vlshr (v2i64 V128:$Rn), vecshiftR64Narrow:$imm))), + (SHRNv2i32_shift V128:$Rn, vecshiftR64Narrow:$imm)>; + +def : Pat<(v16i8 (concat_vectors (v8i8 V64:$Rd), + (trunc (AArch64vlshr (v8i16 V128:$Rn), + vecshiftR16Narrow:$imm)))), + (SHRNv16i8_shift (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), + V128:$Rn, vecshiftR16Narrow:$imm)>; +def : Pat<(v8i16 (concat_vectors (v4i16 V64:$Rd), + (trunc (AArch64vlshr (v4i32 V128:$Rn), + vecshiftR32Narrow:$imm)))), + (SHRNv8i16_shift (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), + V128:$Rn, vecshiftR32Narrow:$imm)>; +def : Pat<(v4i32 (concat_vectors (v2i32 V64:$Rd), + (trunc (AArch64vlshr (v2i64 V128:$Rn), + vecshiftR64Narrow:$imm)))), + (SHRNv4i32_shift (INSERT_SUBREG (IMPLICIT_DEF), V64:$Rd, dsub), + V128:$Rn, vecshiftR32Narrow:$imm)>; + +// Vector sign and zero extensions are implemented with SSHLL and USSHLL. +// Anyexts are implemented as zexts. +def : Pat<(v8i16 (sext (v8i8 V64:$Rn))), (SSHLLv8i8_shift V64:$Rn, (i32 0))>; +def : Pat<(v8i16 (zext (v8i8 V64:$Rn))), (USHLLv8i8_shift V64:$Rn, (i32 0))>; +def : Pat<(v8i16 (anyext (v8i8 V64:$Rn))), (USHLLv8i8_shift V64:$Rn, (i32 0))>; +def : Pat<(v4i32 (sext (v4i16 V64:$Rn))), (SSHLLv4i16_shift V64:$Rn, (i32 0))>; +def : Pat<(v4i32 (zext (v4i16 V64:$Rn))), (USHLLv4i16_shift V64:$Rn, (i32 0))>; +def : Pat<(v4i32 (anyext (v4i16 V64:$Rn))), (USHLLv4i16_shift V64:$Rn, (i32 0))>; +def : Pat<(v2i64 (sext (v2i32 V64:$Rn))), (SSHLLv2i32_shift V64:$Rn, (i32 0))>; +def : Pat<(v2i64 (zext (v2i32 V64:$Rn))), (USHLLv2i32_shift V64:$Rn, (i32 0))>; +def : Pat<(v2i64 (anyext (v2i32 V64:$Rn))), (USHLLv2i32_shift V64:$Rn, (i32 0))>; +// Also match an extend from the upper half of a 128 bit source register. +def : Pat<(v8i16 (anyext (v8i8 (extract_subvector V128:$Rn, (i64 8)) ))), + (USHLLv16i8_shift V128:$Rn, (i32 0))>; +def : Pat<(v8i16 (zext (v8i8 (extract_subvector V128:$Rn, (i64 8)) ))), + (USHLLv16i8_shift V128:$Rn, (i32 0))>; +def : Pat<(v8i16 (sext (v8i8 (extract_subvector V128:$Rn, (i64 8)) ))), + (SSHLLv16i8_shift V128:$Rn, (i32 0))>; +def : Pat<(v4i32 (anyext (v4i16 (extract_subvector V128:$Rn, (i64 4)) ))), + (USHLLv8i16_shift V128:$Rn, (i32 0))>; +def : Pat<(v4i32 (zext (v4i16 (extract_subvector V128:$Rn, (i64 4)) ))), + (USHLLv8i16_shift V128:$Rn, (i32 0))>; +def : Pat<(v4i32 (sext (v4i16 (extract_subvector V128:$Rn, (i64 4)) ))), + (SSHLLv8i16_shift V128:$Rn, (i32 0))>; +def : Pat<(v2i64 (anyext (v2i32 (extract_subvector V128:$Rn, (i64 2)) ))), + (USHLLv4i32_shift V128:$Rn, (i32 0))>; +def : Pat<(v2i64 (zext (v2i32 (extract_subvector V128:$Rn, (i64 2)) ))), + (USHLLv4i32_shift V128:$Rn, (i32 0))>; +def : Pat<(v2i64 (sext (v2i32 (extract_subvector V128:$Rn, (i64 2)) ))), + (SSHLLv4i32_shift V128:$Rn, (i32 0))>; + +// Vector shift sxtl aliases +def : InstAlias<"sxtl.8h $dst, $src1", + (SSHLLv8i8_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"sxtl $dst.8h, $src1.8b", + (SSHLLv8i8_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"sxtl.4s $dst, $src1", + (SSHLLv4i16_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"sxtl $dst.4s, $src1.4h", + (SSHLLv4i16_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"sxtl.2d $dst, $src1", + (SSHLLv2i32_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"sxtl $dst.2d, $src1.2s", + (SSHLLv2i32_shift V128:$dst, V64:$src1, 0)>; + +// Vector shift sxtl2 aliases +def : InstAlias<"sxtl2.8h $dst, $src1", + (SSHLLv16i8_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"sxtl2 $dst.8h, $src1.16b", + (SSHLLv16i8_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"sxtl2.4s $dst, $src1", + (SSHLLv8i16_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"sxtl2 $dst.4s, $src1.8h", + (SSHLLv8i16_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"sxtl2.2d $dst, $src1", + (SSHLLv4i32_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"sxtl2 $dst.2d, $src1.4s", + (SSHLLv4i32_shift V128:$dst, V128:$src1, 0)>; + +// Vector shift uxtl aliases +def : InstAlias<"uxtl.8h $dst, $src1", + (USHLLv8i8_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"uxtl $dst.8h, $src1.8b", + (USHLLv8i8_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"uxtl.4s $dst, $src1", + (USHLLv4i16_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"uxtl $dst.4s, $src1.4h", + (USHLLv4i16_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"uxtl.2d $dst, $src1", + (USHLLv2i32_shift V128:$dst, V64:$src1, 0)>; +def : InstAlias<"uxtl $dst.2d, $src1.2s", + (USHLLv2i32_shift V128:$dst, V64:$src1, 0)>; + +// Vector shift uxtl2 aliases +def : InstAlias<"uxtl2.8h $dst, $src1", + (USHLLv16i8_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"uxtl2 $dst.8h, $src1.16b", + (USHLLv16i8_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"uxtl2.4s $dst, $src1", + (USHLLv8i16_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"uxtl2 $dst.4s, $src1.8h", + (USHLLv8i16_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"uxtl2.2d $dst, $src1", + (USHLLv4i32_shift V128:$dst, V128:$src1, 0)>; +def : InstAlias<"uxtl2 $dst.2d, $src1.4s", + (USHLLv4i32_shift V128:$dst, V128:$src1, 0)>; + +// If an integer is about to be converted to a floating point value, +// just load it on the floating point unit. +// These patterns are more complex because floating point loads do not +// support sign extension. +// The sign extension has to be explicitly added and is only supported for +// one step: byte-to-half, half-to-word, word-to-doubleword. +// SCVTF GPR -> FPR is 9 cycles. +// SCVTF FPR -> FPR is 4 cyclces. +// (sign extension with lengthen) SXTL FPR -> FPR is 2 cycles. +// Therefore, we can do 2 sign extensions and one SCVTF FPR -> FPR +// and still being faster. +// However, this is not good for code size. +// 8-bits -> float. 2 sizes step-up. +class SExtLoadi8CVTf32Pat + : Pat<(f32 (sint_to_fp (i32 (sextloadi8 addrmode)))), + (SCVTFv1i32 (f32 (EXTRACT_SUBREG + (SSHLLv4i16_shift + (f64 + (EXTRACT_SUBREG + (SSHLLv8i8_shift + (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + INST, + bsub), + 0), + dsub)), + 0), + ssub)))>, Requires<[NotForCodeSize, IsCyclone]>; + +def : SExtLoadi8CVTf32Pat<(ro8.Wpat GPR64sp:$Rn, GPR32:$Rm, ro8.Wext:$ext), + (LDRBroW GPR64sp:$Rn, GPR32:$Rm, ro8.Wext:$ext)>; +def : SExtLoadi8CVTf32Pat<(ro8.Xpat GPR64sp:$Rn, GPR64:$Rm, ro8.Xext:$ext), + (LDRBroX GPR64sp:$Rn, GPR64:$Rm, ro8.Xext:$ext)>; +def : SExtLoadi8CVTf32Pat<(am_indexed8 GPR64sp:$Rn, uimm12s1:$offset), + (LDRBui GPR64sp:$Rn, uimm12s1:$offset)>; +def : SExtLoadi8CVTf32Pat<(am_unscaled8 GPR64sp:$Rn, simm9:$offset), + (LDURBi GPR64sp:$Rn, simm9:$offset)>; + +// 16-bits -> float. 1 size step-up. +class SExtLoadi16CVTf32Pat + : Pat<(f32 (sint_to_fp (i32 (sextloadi16 addrmode)))), + (SCVTFv1i32 (f32 (EXTRACT_SUBREG + (SSHLLv4i16_shift + (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + INST, + hsub), + 0), + ssub)))>, Requires<[NotForCodeSize]>; + +def : SExtLoadi16CVTf32Pat<(ro16.Wpat GPR64sp:$Rn, GPR32:$Rm, ro16.Wext:$ext), + (LDRHroW GPR64sp:$Rn, GPR32:$Rm, ro16.Wext:$ext)>; +def : SExtLoadi16CVTf32Pat<(ro16.Xpat GPR64sp:$Rn, GPR64:$Rm, ro16.Xext:$ext), + (LDRHroX GPR64sp:$Rn, GPR64:$Rm, ro16.Xext:$ext)>; +def : SExtLoadi16CVTf32Pat<(am_indexed16 GPR64sp:$Rn, uimm12s2:$offset), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset)>; +def : SExtLoadi16CVTf32Pat<(am_unscaled16 GPR64sp:$Rn, simm9:$offset), + (LDURHi GPR64sp:$Rn, simm9:$offset)>; + +// 32-bits to 32-bits are handled in target specific dag combine: +// performIntToFpCombine. +// 64-bits integer to 32-bits floating point, not possible with +// SCVTF on floating point registers (both source and destination +// must have the same size). + +// Here are the patterns for 8, 16, 32, and 64-bits to double. +// 8-bits -> double. 3 size step-up: give up. +// 16-bits -> double. 2 size step. +class SExtLoadi16CVTf64Pat + : Pat <(f64 (sint_to_fp (i32 (sextloadi16 addrmode)))), + (SCVTFv1i64 (f64 (EXTRACT_SUBREG + (SSHLLv2i32_shift + (f64 + (EXTRACT_SUBREG + (SSHLLv4i16_shift + (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + INST, + hsub), + 0), + dsub)), + 0), + dsub)))>, Requires<[NotForCodeSize, IsCyclone]>; + +def : SExtLoadi16CVTf64Pat<(ro16.Wpat GPR64sp:$Rn, GPR32:$Rm, ro16.Wext:$ext), + (LDRHroW GPR64sp:$Rn, GPR32:$Rm, ro16.Wext:$ext)>; +def : SExtLoadi16CVTf64Pat<(ro16.Xpat GPR64sp:$Rn, GPR64:$Rm, ro16.Xext:$ext), + (LDRHroX GPR64sp:$Rn, GPR64:$Rm, ro16.Xext:$ext)>; +def : SExtLoadi16CVTf64Pat<(am_indexed16 GPR64sp:$Rn, uimm12s2:$offset), + (LDRHui GPR64sp:$Rn, uimm12s2:$offset)>; +def : SExtLoadi16CVTf64Pat<(am_unscaled16 GPR64sp:$Rn, simm9:$offset), + (LDURHi GPR64sp:$Rn, simm9:$offset)>; +// 32-bits -> double. 1 size step-up. +class SExtLoadi32CVTf64Pat + : Pat <(f64 (sint_to_fp (i32 (load addrmode)))), + (SCVTFv1i64 (f64 (EXTRACT_SUBREG + (SSHLLv2i32_shift + (INSERT_SUBREG (f64 (IMPLICIT_DEF)), + INST, + ssub), + 0), + dsub)))>, Requires<[NotForCodeSize]>; + +def : SExtLoadi32CVTf64Pat<(ro32.Wpat GPR64sp:$Rn, GPR32:$Rm, ro32.Wext:$ext), + (LDRSroW GPR64sp:$Rn, GPR32:$Rm, ro32.Wext:$ext)>; +def : SExtLoadi32CVTf64Pat<(ro32.Xpat GPR64sp:$Rn, GPR64:$Rm, ro32.Xext:$ext), + (LDRSroX GPR64sp:$Rn, GPR64:$Rm, ro32.Xext:$ext)>; +def : SExtLoadi32CVTf64Pat<(am_indexed32 GPR64sp:$Rn, uimm12s4:$offset), + (LDRSui GPR64sp:$Rn, uimm12s4:$offset)>; +def : SExtLoadi32CVTf64Pat<(am_unscaled32 GPR64sp:$Rn, simm9:$offset), + (LDURSi GPR64sp:$Rn, simm9:$offset)>; + +// 64-bits -> double are handled in target specific dag combine: +// performIntToFpCombine. + + +//---------------------------------------------------------------------------- +// AdvSIMD Load-Store Structure +//---------------------------------------------------------------------------- +defm LD1 : SIMDLd1Multiple<"ld1">; +defm LD2 : SIMDLd2Multiple<"ld2">; +defm LD3 : SIMDLd3Multiple<"ld3">; +defm LD4 : SIMDLd4Multiple<"ld4">; + +defm ST1 : SIMDSt1Multiple<"st1">; +defm ST2 : SIMDSt2Multiple<"st2">; +defm ST3 : SIMDSt3Multiple<"st3">; +defm ST4 : SIMDSt4Multiple<"st4">; + +class Ld1Pat + : Pat<(ty (load GPR64sp:$Rn)), (INST GPR64sp:$Rn)>; + +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; +def : Ld1Pat; + +class St1Pat + : Pat<(store ty:$Vt, GPR64sp:$Rn), + (INST ty:$Vt, GPR64sp:$Rn)>; + +def : St1Pat; +def : St1Pat; +def : St1Pat; +def : St1Pat; +def : St1Pat; +def : St1Pat; +def : St1Pat; +def : St1Pat; + +//--- +// Single-element +//--- + +defm LD1R : SIMDLdR<0, 0b110, 0, "ld1r", "One", 1, 2, 4, 8>; +defm LD2R : SIMDLdR<1, 0b110, 0, "ld2r", "Two", 2, 4, 8, 16>; +defm LD3R : SIMDLdR<0, 0b111, 0, "ld3r", "Three", 3, 6, 12, 24>; +defm LD4R : SIMDLdR<1, 0b111, 0, "ld4r", "Four", 4, 8, 16, 32>; +let mayLoad = 1, hasSideEffects = 0 in { +defm LD1 : SIMDLdSingleBTied<0, 0b000, "ld1", VecListOneb, GPR64pi1>; +defm LD1 : SIMDLdSingleHTied<0, 0b010, 0, "ld1", VecListOneh, GPR64pi2>; +defm LD1 : SIMDLdSingleSTied<0, 0b100, 0b00, "ld1", VecListOnes, GPR64pi4>; +defm LD1 : SIMDLdSingleDTied<0, 0b100, 0b01, "ld1", VecListOned, GPR64pi8>; +defm LD2 : SIMDLdSingleBTied<1, 0b000, "ld2", VecListTwob, GPR64pi2>; +defm LD2 : SIMDLdSingleHTied<1, 0b010, 0, "ld2", VecListTwoh, GPR64pi4>; +defm LD2 : SIMDLdSingleSTied<1, 0b100, 0b00, "ld2", VecListTwos, GPR64pi8>; +defm LD2 : SIMDLdSingleDTied<1, 0b100, 0b01, "ld2", VecListTwod, GPR64pi16>; +defm LD3 : SIMDLdSingleBTied<0, 0b001, "ld3", VecListThreeb, GPR64pi3>; +defm LD3 : SIMDLdSingleHTied<0, 0b011, 0, "ld3", VecListThreeh, GPR64pi6>; +defm LD3 : SIMDLdSingleSTied<0, 0b101, 0b00, "ld3", VecListThrees, GPR64pi12>; +defm LD3 : SIMDLdSingleDTied<0, 0b101, 0b01, "ld3", VecListThreed, GPR64pi24>; +defm LD4 : SIMDLdSingleBTied<1, 0b001, "ld4", VecListFourb, GPR64pi4>; +defm LD4 : SIMDLdSingleHTied<1, 0b011, 0, "ld4", VecListFourh, GPR64pi8>; +defm LD4 : SIMDLdSingleSTied<1, 0b101, 0b00, "ld4", VecListFours, GPR64pi16>; +defm LD4 : SIMDLdSingleDTied<1, 0b101, 0b01, "ld4", VecListFourd, GPR64pi32>; +} + +def : Pat<(v8i8 (AArch64dup (i32 (extloadi8 GPR64sp:$Rn)))), + (LD1Rv8b GPR64sp:$Rn)>; +def : Pat<(v16i8 (AArch64dup (i32 (extloadi8 GPR64sp:$Rn)))), + (LD1Rv16b GPR64sp:$Rn)>; +def : Pat<(v4i16 (AArch64dup (i32 (extloadi16 GPR64sp:$Rn)))), + (LD1Rv4h GPR64sp:$Rn)>; +def : Pat<(v8i16 (AArch64dup (i32 (extloadi16 GPR64sp:$Rn)))), + (LD1Rv8h GPR64sp:$Rn)>; +def : Pat<(v2i32 (AArch64dup (i32 (load GPR64sp:$Rn)))), + (LD1Rv2s GPR64sp:$Rn)>; +def : Pat<(v4i32 (AArch64dup (i32 (load GPR64sp:$Rn)))), + (LD1Rv4s GPR64sp:$Rn)>; +def : Pat<(v2i64 (AArch64dup (i64 (load GPR64sp:$Rn)))), + (LD1Rv2d GPR64sp:$Rn)>; +def : Pat<(v1i64 (AArch64dup (i64 (load GPR64sp:$Rn)))), + (LD1Rv1d GPR64sp:$Rn)>; +// Grab the floating point version too +def : Pat<(v2f32 (AArch64dup (f32 (load GPR64sp:$Rn)))), + (LD1Rv2s GPR64sp:$Rn)>; +def : Pat<(v4f32 (AArch64dup (f32 (load GPR64sp:$Rn)))), + (LD1Rv4s GPR64sp:$Rn)>; +def : Pat<(v2f64 (AArch64dup (f64 (load GPR64sp:$Rn)))), + (LD1Rv2d GPR64sp:$Rn)>; +def : Pat<(v1f64 (AArch64dup (f64 (load GPR64sp:$Rn)))), + (LD1Rv1d GPR64sp:$Rn)>; +def : Pat<(v4f16 (AArch64dup (f16 (load GPR64sp:$Rn)))), + (LD1Rv4h GPR64sp:$Rn)>; +def : Pat<(v8f16 (AArch64dup (f16 (load GPR64sp:$Rn)))), + (LD1Rv8h GPR64sp:$Rn)>; + +class Ld1Lane128Pat + : Pat<(vector_insert (VTy VecListOne128:$Rd), + (STy (scalar_load GPR64sp:$Rn)), VecIndex:$idx), + (LD1 VecListOne128:$Rd, VecIndex:$idx, GPR64sp:$Rn)>; + +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; +def : Ld1Lane128Pat; + +class Ld1Lane64Pat + : Pat<(vector_insert (VTy VecListOne64:$Rd), + (STy (scalar_load GPR64sp:$Rn)), VecIndex:$idx), + (EXTRACT_SUBREG + (LD1 (SUBREG_TO_REG (i32 0), VecListOne64:$Rd, dsub), + VecIndex:$idx, GPR64sp:$Rn), + dsub)>; + +def : Ld1Lane64Pat; +def : Ld1Lane64Pat; +def : Ld1Lane64Pat; +def : Ld1Lane64Pat; +def : Ld1Lane64Pat; + + +defm LD1 : SIMDLdSt1SingleAliases<"ld1">; +defm LD2 : SIMDLdSt2SingleAliases<"ld2">; +defm LD3 : SIMDLdSt3SingleAliases<"ld3">; +defm LD4 : SIMDLdSt4SingleAliases<"ld4">; + +// Stores +defm ST1 : SIMDStSingleB<0, 0b000, "st1", VecListOneb, GPR64pi1>; +defm ST1 : SIMDStSingleH<0, 0b010, 0, "st1", VecListOneh, GPR64pi2>; +defm ST1 : SIMDStSingleS<0, 0b100, 0b00, "st1", VecListOnes, GPR64pi4>; +defm ST1 : SIMDStSingleD<0, 0b100, 0b01, "st1", VecListOned, GPR64pi8>; + +let AddedComplexity = 19 in +class St1Lane128Pat + : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne128:$Vt), VecIndex:$idx)), + GPR64sp:$Rn), + (ST1 VecListOne128:$Vt, VecIndex:$idx, GPR64sp:$Rn)>; + +def : St1Lane128Pat; +def : St1Lane128Pat; +def : St1Lane128Pat; +def : St1Lane128Pat; +def : St1Lane128Pat; +def : St1Lane128Pat; +def : St1Lane128Pat; + +let AddedComplexity = 19 in +class St1Lane64Pat + : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne64:$Vt), VecIndex:$idx)), + GPR64sp:$Rn), + (ST1 (SUBREG_TO_REG (i32 0), VecListOne64:$Vt, dsub), + VecIndex:$idx, GPR64sp:$Rn)>; + +def : St1Lane64Pat; +def : St1Lane64Pat; +def : St1Lane64Pat; +def : St1Lane64Pat; +def : St1Lane64Pat; + +multiclass St1LanePost64Pat { + def : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne64:$Vt), VecIndex:$idx)), + GPR64sp:$Rn, offset), + (ST1 (SUBREG_TO_REG (i32 0), VecListOne64:$Vt, dsub), + VecIndex:$idx, GPR64sp:$Rn, XZR)>; + + def : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne64:$Vt), VecIndex:$idx)), + GPR64sp:$Rn, GPR64:$Rm), + (ST1 (SUBREG_TO_REG (i32 0), VecListOne64:$Vt, dsub), + VecIndex:$idx, GPR64sp:$Rn, $Rm)>; +} + +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; +defm : St1LanePost64Pat; + +multiclass St1LanePost128Pat { + def : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne128:$Vt), VecIndex:$idx)), + GPR64sp:$Rn, offset), + (ST1 VecListOne128:$Vt, VecIndex:$idx, GPR64sp:$Rn, XZR)>; + + def : Pat<(scalar_store + (STy (vector_extract (VTy VecListOne128:$Vt), VecIndex:$idx)), + GPR64sp:$Rn, GPR64:$Rm), + (ST1 VecListOne128:$Vt, VecIndex:$idx, GPR64sp:$Rn, $Rm)>; +} + +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; +defm : St1LanePost128Pat; + +let mayStore = 1, hasSideEffects = 0 in { +defm ST2 : SIMDStSingleB<1, 0b000, "st2", VecListTwob, GPR64pi2>; +defm ST2 : SIMDStSingleH<1, 0b010, 0, "st2", VecListTwoh, GPR64pi4>; +defm ST2 : SIMDStSingleS<1, 0b100, 0b00, "st2", VecListTwos, GPR64pi8>; +defm ST2 : SIMDStSingleD<1, 0b100, 0b01, "st2", VecListTwod, GPR64pi16>; +defm ST3 : SIMDStSingleB<0, 0b001, "st3", VecListThreeb, GPR64pi3>; +defm ST3 : SIMDStSingleH<0, 0b011, 0, "st3", VecListThreeh, GPR64pi6>; +defm ST3 : SIMDStSingleS<0, 0b101, 0b00, "st3", VecListThrees, GPR64pi12>; +defm ST3 : SIMDStSingleD<0, 0b101, 0b01, "st3", VecListThreed, GPR64pi24>; +defm ST4 : SIMDStSingleB<1, 0b001, "st4", VecListFourb, GPR64pi4>; +defm ST4 : SIMDStSingleH<1, 0b011, 0, "st4", VecListFourh, GPR64pi8>; +defm ST4 : SIMDStSingleS<1, 0b101, 0b00, "st4", VecListFours, GPR64pi16>; +defm ST4 : SIMDStSingleD<1, 0b101, 0b01, "st4", VecListFourd, GPR64pi32>; +} + +defm ST1 : SIMDLdSt1SingleAliases<"st1">; +defm ST2 : SIMDLdSt2SingleAliases<"st2">; +defm ST3 : SIMDLdSt3SingleAliases<"st3">; +defm ST4 : SIMDLdSt4SingleAliases<"st4">; + +//---------------------------------------------------------------------------- +// Crypto extensions +//---------------------------------------------------------------------------- + +def AESErr : AESTiedInst<0b0100, "aese", int_aarch64_crypto_aese>; +def AESDrr : AESTiedInst<0b0101, "aesd", int_aarch64_crypto_aesd>; +def AESMCrr : AESInst< 0b0110, "aesmc", int_aarch64_crypto_aesmc>; +def AESIMCrr : AESInst< 0b0111, "aesimc", int_aarch64_crypto_aesimc>; + +def SHA1Crrr : SHATiedInstQSV<0b000, "sha1c", int_aarch64_crypto_sha1c>; +def SHA1Prrr : SHATiedInstQSV<0b001, "sha1p", int_aarch64_crypto_sha1p>; +def SHA1Mrrr : SHATiedInstQSV<0b010, "sha1m", int_aarch64_crypto_sha1m>; +def SHA1SU0rrr : SHATiedInstVVV<0b011, "sha1su0", int_aarch64_crypto_sha1su0>; +def SHA256Hrrr : SHATiedInstQQV<0b100, "sha256h", int_aarch64_crypto_sha256h>; +def SHA256H2rrr : SHATiedInstQQV<0b101, "sha256h2",int_aarch64_crypto_sha256h2>; +def SHA256SU1rrr :SHATiedInstVVV<0b110, "sha256su1",int_aarch64_crypto_sha256su1>; + +def SHA1Hrr : SHAInstSS< 0b0000, "sha1h", int_aarch64_crypto_sha1h>; +def SHA1SU1rr : SHATiedInstVV<0b0001, "sha1su1", int_aarch64_crypto_sha1su1>; +def SHA256SU0rr : SHATiedInstVV<0b0010, "sha256su0",int_aarch64_crypto_sha256su0>; + +//---------------------------------------------------------------------------- +// Compiler-pseudos +//---------------------------------------------------------------------------- +// FIXME: Like for X86, these should go in their own separate .td file. + +// Any instruction that defines a 32-bit result leaves the high half of the +// register. Truncate can be lowered to EXTRACT_SUBREG. CopyFromReg may +// be copying from a truncate. But any other 32-bit operation will zero-extend +// up to 64 bits. +// FIXME: X86 also checks for CMOV here. Do we need something similar? +def def32 : PatLeaf<(i32 GPR32:$src), [{ + return N->getOpcode() != ISD::TRUNCATE && + N->getOpcode() != TargetOpcode::EXTRACT_SUBREG && + N->getOpcode() != ISD::CopyFromReg; }]>; -def simm9_asmoperand : AsmOperandClass { - let Name = "SImm9"; - let PredicateMethod = "isSImm<9>"; - let RenderMethod = "addSImmOperands<9>"; - let DiagnosticType = "LoadStoreSImm9"; -} - -def simm9 : Operand, - ImmLeaf= -0x100 && Imm <= 0xff; }], - SDXF_simm9> { - let PrintMethod = "printOffsetSImm9Operand"; - let ParserMatchClass = simm9_asmoperand; -} - - -//===------------------------------- -// 1.3 Register offset extensions -//===------------------------------- - -// The assembly-syntax for these addressing-modes is: -// [, {, {}}] +// In the case of a 32-bit def that is known to implicitly zero-extend, +// we can use a SUBREG_TO_REG. +def : Pat<(i64 (zext def32:$src)), (SUBREG_TO_REG (i64 0), GPR32:$src, sub_32)>; + +// For an anyext, we don't care what the high bits are, so we can perform an +// INSERT_SUBREF into an IMPLICIT_DEF. +def : Pat<(i64 (anyext GPR32:$src)), + (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GPR32:$src, sub_32)>; + +// When we need to explicitly zero-extend, we use a 32-bit MOV instruction and +// then assert the extension has happened. +def : Pat<(i64 (zext GPR32:$src)), + (SUBREG_TO_REG (i32 0), (ORRWrs WZR, GPR32:$src, 0), sub_32)>; + +// To sign extend, we use a signed bitfield move instruction (SBFM) on the +// containing super-reg. +def : Pat<(i64 (sext GPR32:$src)), + (SBFMXri (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GPR32:$src, sub_32), 0, 31)>; +def : Pat<(i64 (sext_inreg GPR64:$src, i32)), (SBFMXri GPR64:$src, 0, 31)>; +def : Pat<(i64 (sext_inreg GPR64:$src, i16)), (SBFMXri GPR64:$src, 0, 15)>; +def : Pat<(i64 (sext_inreg GPR64:$src, i8)), (SBFMXri GPR64:$src, 0, 7)>; +def : Pat<(i64 (sext_inreg GPR64:$src, i1)), (SBFMXri GPR64:$src, 0, 0)>; +def : Pat<(i32 (sext_inreg GPR32:$src, i16)), (SBFMWri GPR32:$src, 0, 15)>; +def : Pat<(i32 (sext_inreg GPR32:$src, i8)), (SBFMWri GPR32:$src, 0, 7)>; +def : Pat<(i32 (sext_inreg GPR32:$src, i1)), (SBFMWri GPR32:$src, 0, 0)>; + +def : Pat<(shl (sext_inreg GPR32:$Rn, i8), (i64 imm0_31:$imm)), + (SBFMWri GPR32:$Rn, (i64 (i32shift_a imm0_31:$imm)), + (i64 (i32shift_sext_i8 imm0_31:$imm)))>; +def : Pat<(shl (sext_inreg GPR64:$Rn, i8), (i64 imm0_63:$imm)), + (SBFMXri GPR64:$Rn, (i64 (i64shift_a imm0_63:$imm)), + (i64 (i64shift_sext_i8 imm0_63:$imm)))>; + +def : Pat<(shl (sext_inreg GPR32:$Rn, i16), (i64 imm0_31:$imm)), + (SBFMWri GPR32:$Rn, (i64 (i32shift_a imm0_31:$imm)), + (i64 (i32shift_sext_i16 imm0_31:$imm)))>; +def : Pat<(shl (sext_inreg GPR64:$Rn, i16), (i64 imm0_63:$imm)), + (SBFMXri GPR64:$Rn, (i64 (i64shift_a imm0_63:$imm)), + (i64 (i64shift_sext_i16 imm0_63:$imm)))>; + +def : Pat<(shl (i64 (sext GPR32:$Rn)), (i64 imm0_63:$imm)), + (SBFMXri (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GPR32:$Rn, sub_32), + (i64 (i64shift_a imm0_63:$imm)), + (i64 (i64shift_sext_i32 imm0_63:$imm)))>; + +// sra patterns have an AddedComplexity of 10, so make sure we have a higher +// AddedComplexity for the following patterns since we want to match sext + sra +// patterns before we attempt to match a single sra node. +let AddedComplexity = 20 in { +// We support all sext + sra combinations which preserve at least one bit of the +// original value which is to be sign extended. E.g. we support shifts up to +// bitwidth-1 bits. +def : Pat<(sra (sext_inreg GPR32:$Rn, i8), (i64 imm0_7:$imm)), + (SBFMWri GPR32:$Rn, (i64 imm0_7:$imm), 7)>; +def : Pat<(sra (sext_inreg GPR64:$Rn, i8), (i64 imm0_7:$imm)), + (SBFMXri GPR64:$Rn, (i64 imm0_7:$imm), 7)>; + +def : Pat<(sra (sext_inreg GPR32:$Rn, i16), (i64 imm0_15:$imm)), + (SBFMWri GPR32:$Rn, (i64 imm0_15:$imm), 15)>; +def : Pat<(sra (sext_inreg GPR64:$Rn, i16), (i64 imm0_15:$imm)), + (SBFMXri GPR64:$Rn, (i64 imm0_15:$imm), 15)>; + +def : Pat<(sra (i64 (sext GPR32:$Rn)), (i64 imm0_31:$imm)), + (SBFMXri (INSERT_SUBREG (i64 (IMPLICIT_DEF)), GPR32:$Rn, sub_32), + (i64 imm0_31:$imm), 31)>; +} // AddedComplexity = 20 + +// To truncate, we can simply extract from a subregister. +def : Pat<(i32 (trunc GPR64sp:$src)), + (i32 (EXTRACT_SUBREG GPR64sp:$src, sub_32))>; + +// __builtin_trap() uses the BRK instruction on AArch64. +def : Pat<(trap), (BRK 1)>; + +// Conversions within AdvSIMD types in the same register size are free. +// But because we need a consistent lane ordering, in big endian many +// conversions require one or more REV instructions. // -// The essential semantics are: -// + is a shift: # or #0 -// + can be W or X. -// + If is W, can be UXTW or SXTW -// + If is X, can be LSL or SXTX +// Consider a simple memory load followed by a bitconvert then a store. +// v0 = load v2i32 +// v1 = BITCAST v2i32 v0 to v4i16 +// store v4i16 v2 // -// The trickiest of those constraints is that Rm can be either GPR32 or GPR64, -// which will need separate instructions for LLVM type-consistency. We'll also -// need separate operands, of course. -multiclass regexts { - def regext_asmoperand : AsmOperandClass { - let Name = "AddrRegExtend_" # MemSize # "_" # Rm; - let PredicateMethod = "isAddrRegExtend<" # MemSize # "," # RmSize # ">"; - let RenderMethod = "addAddrRegExtendOperands<" # MemSize # ">"; - let DiagnosticType = "LoadStoreExtend" # RmSize # "_" # MemSize; - } - - def regext : Operand { - let PrintMethod - = "printAddrRegExtendOperand<" # MemSize # ", " # RmSize # ">"; - - let DecoderMethod = "DecodeAddrRegExtendOperand"; - let ParserMatchClass - = !cast(prefix # regext_asmoperand); - } -} - -multiclass regexts_wx { - // Rm is an X-register if LSL or SXTX are specified as the shift. - defm Xm_ : regexts; - - // Rm is a W-register if UXTW or SXTW are specified as the shift. - defm Wm_ : regexts; -} - -defm byte_ : regexts_wx<1, "byte_">; -defm hword_ : regexts_wx<2, "hword_">; -defm word_ : regexts_wx<4, "word_">; -defm dword_ : regexts_wx<8, "dword_">; -defm qword_ : regexts_wx<16, "qword_">; - - -//===------------------------------ -// 2. The instructions themselves. -//===------------------------------ - -// We have the following instructions to implement: -// | | B | H | W | X | -// |-----------------+-------+-------+-------+--------| -// | unsigned str | STRB | STRH | STR | STR | -// | unsigned ldr | LDRB | LDRH | LDR | LDR | -// | signed ldr to W | LDRSB | LDRSH | - | - | -// | signed ldr to X | LDRSB | LDRSH | LDRSW | (PRFM) | - -// This will instantiate the LDR/STR instructions you'd expect to use for an -// unsigned datatype (first two rows above) or floating-point register, which is -// reasonably uniform across all access sizes. - - -//===------------------------------ -// 2.1 Regular instructions -//===------------------------------ - -// This class covers the basic unsigned or irrelevantly-signed loads and stores, -// to general-purpose and floating-point registers. - -class AddrParams { - Operand uimm12 = !cast(prefix # "_uimm12"); - - Operand regextWm = !cast(prefix # "_Wm_regext"); - Operand regextXm = !cast(prefix # "_Xm_regext"); -} - -def byte_addrparams : AddrParams<"byte">; -def hword_addrparams : AddrParams<"hword">; -def word_addrparams : AddrParams<"word">; -def dword_addrparams : AddrParams<"dword">; -def qword_addrparams : AddrParams<"qword">; - -multiclass A64I_LDRSTR_unsigned size, bit v, - bit high_opc, string asmsuffix, - RegisterClass GPR, AddrParams params> { - // Unsigned immediate - def _STR : A64I_LSunsigimm { - let mayStore = 1; - } - def : InstAlias<"str" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_STR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - - def _LDR : A64I_LSunsigimm { - let mayLoad = 1; - } - def : InstAlias<"ldr" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_LDR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - - // Register offset (four of these: load/store and Wm/Xm). - let mayLoad = 1 in { - def _Wm_RegOffset_LDR : A64I_LSregoff; - - def _Xm_RegOffset_LDR : A64I_LSregoff; - } - def : InstAlias<"ldr" # asmsuffix # " $Rt, [$Rn, $Rm]", - (!cast(prefix # "_Xm_RegOffset_LDR") GPR:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, 2)>; - - let mayStore = 1 in { - def _Wm_RegOffset_STR : A64I_LSregoff; - - def _Xm_RegOffset_STR : A64I_LSregoff; - } - def : InstAlias<"str" # asmsuffix # " $Rt, [$Rn, $Rm]", - (!cast(prefix # "_Xm_RegOffset_STR") GPR:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, 2)>; - - // Unaligned immediate - def _STUR : A64I_LSunalimm { - let mayStore = 1; - } - def : InstAlias<"stur" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_STUR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - - def _LDUR : A64I_LSunalimm { - let mayLoad = 1; - } - def : InstAlias<"ldur" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_LDUR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - - // Post-indexed - def _PostInd_STR : A64I_LSpostind { - let Constraints = "$Rn = $Rn_wb"; - let mayStore = 1; - - // Decoder only needed for unpredictability checking (FIXME). - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - def _PostInd_LDR : A64I_LSpostind { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - // Pre-indexed - def _PreInd_STR : A64I_LSpreind { - let Constraints = "$Rn = $Rn_wb"; - let mayStore = 1; - - // Decoder only needed for unpredictability checking (FIXME). - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - def _PreInd_LDR : A64I_LSpreind { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - -} - -// STRB/LDRB: First define the instructions -defm LS8 - : A64I_LDRSTR_unsigned<"LS8", 0b00, 0b0, 0b0, "b", GPR32, byte_addrparams>; - -// STRH/LDRH -defm LS16 - : A64I_LDRSTR_unsigned<"LS16", 0b01, 0b0, 0b0, "h", GPR32, hword_addrparams>; - - -// STR/LDR to/from a W register -defm LS32 - : A64I_LDRSTR_unsigned<"LS32", 0b10, 0b0, 0b0, "", GPR32, word_addrparams>; - -// STR/LDR to/from an X register -defm LS64 - : A64I_LDRSTR_unsigned<"LS64", 0b11, 0b0, 0b0, "", GPR64, dword_addrparams>; - -// STR/LDR to/from a B register -defm LSFP8 - : A64I_LDRSTR_unsigned<"LSFP8", 0b00, 0b1, 0b0, "", FPR8, byte_addrparams>; - -// STR/LDR to/from an H register -defm LSFP16 - : A64I_LDRSTR_unsigned<"LSFP16", 0b01, 0b1, 0b0, "", FPR16, hword_addrparams>; - -// STR/LDR to/from an S register -defm LSFP32 - : A64I_LDRSTR_unsigned<"LSFP32", 0b10, 0b1, 0b0, "", FPR32, word_addrparams>; -// STR/LDR to/from a D register -defm LSFP64 - : A64I_LDRSTR_unsigned<"LSFP64", 0b11, 0b1, 0b0, "", FPR64, dword_addrparams>; -// STR/LDR to/from a Q register -defm LSFP128 - : A64I_LDRSTR_unsigned<"LSFP128", 0b00, 0b1, 0b1, "", FPR128, - qword_addrparams>; - -//===------------------------------ -// 2.3 Signed loads -//===------------------------------ - -// Byte and half-word signed loads can both go into either an X or a W register, -// so it's worth factoring out. Signed word loads don't fit because there is no -// W version. -multiclass A64I_LDR_signed size, string asmopcode, AddrParams params, - string prefix> { - // Unsigned offset - def w : A64I_LSunsigimm { - let mayLoad = 1; - } - def : InstAlias<"ldrs" # asmopcode # " $Rt, [$Rn]", - (!cast(prefix # w) GPR32:$Rt, GPR64xsp:$Rn, 0)>; - - def x : A64I_LSunsigimm { - let mayLoad = 1; - } - def : InstAlias<"ldrs" # asmopcode # " $Rt, [$Rn]", - (!cast(prefix # x) GPR64:$Rt, GPR64xsp:$Rn, 0)>; - - // Register offset - let mayLoad = 1 in { - def w_Wm_RegOffset : A64I_LSregoff; - - def w_Xm_RegOffset : A64I_LSregoff; - - def x_Wm_RegOffset : A64I_LSregoff; - - def x_Xm_RegOffset : A64I_LSregoff; - } - def : InstAlias<"ldrs" # asmopcode # " $Rt, [$Rn, $Rm]", - (!cast(prefix # "w_Xm_RegOffset") GPR32:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, 2)>; - - def : InstAlias<"ldrs" # asmopcode # " $Rt, [$Rn, $Rm]", - (!cast(prefix # "x_Xm_RegOffset") GPR64:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, 2)>; - - - let mayLoad = 1 in { - // Unaligned offset - def w_U : A64I_LSunalimm; - - def x_U : A64I_LSunalimm; - - - // Post-indexed - def w_PostInd : A64I_LSpostind { - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - def x_PostInd : A64I_LSpostind { - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - // Pre-indexed - def w_PreInd : A64I_LSpreind { - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - - def x_PreInd : A64I_LSpreind { - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; - } - } // let mayLoad = 1 -} - -// LDRSB -defm LDRSB : A64I_LDR_signed<0b00, "b", byte_addrparams, "LDRSB">; -// LDRSH -defm LDRSH : A64I_LDR_signed<0b01, "h", hword_addrparams, "LDRSH">; - -// LDRSW: load a 32-bit register, sign-extending to 64-bits. -def LDRSWx - : A64I_LSunsigimm<0b10, 0b0, 0b10, - (outs GPR64:$Rt), - (ins GPR64xsp:$Rn, word_uimm12:$UImm12), - "ldrsw\t$Rt, [$Rn, $UImm12]", - [], NoItinerary> { - let mayLoad = 1; -} -def : InstAlias<"ldrsw $Rt, [$Rn]", (LDRSWx GPR64:$Rt, GPR64xsp:$Rn, 0)>; - -let mayLoad = 1 in { - def LDRSWx_Wm_RegOffset : A64I_LSregoff<0b10, 0b0, 0b10, 0b0, - (outs GPR64:$Rt), - (ins GPR64xsp:$Rn, GPR32:$Rm, word_Wm_regext:$Ext), - "ldrsw\t$Rt, [$Rn, $Rm, $Ext]", - [], NoItinerary>; - - def LDRSWx_Xm_RegOffset : A64I_LSregoff<0b10, 0b0, 0b10, 0b1, - (outs GPR64:$Rt), - (ins GPR64xsp:$Rn, GPR64:$Rm, word_Xm_regext:$Ext), - "ldrsw\t$Rt, [$Rn, $Rm, $Ext]", - [], NoItinerary>; -} -def : InstAlias<"ldrsw $Rt, [$Rn, $Rm]", - (LDRSWx_Xm_RegOffset GPR64:$Rt, GPR64xsp:$Rn, GPR64:$Rm, 2)>; - - -def LDURSWx - : A64I_LSunalimm<0b10, 0b0, 0b10, - (outs GPR64:$Rt), - (ins GPR64xsp:$Rn, simm9:$SImm9), - "ldursw\t$Rt, [$Rn, $SImm9]", - [], NoItinerary> { - let mayLoad = 1; -} -def : InstAlias<"ldursw $Rt, [$Rn]", (LDURSWx GPR64:$Rt, GPR64xsp:$Rn, 0)>; - -def LDRSWx_PostInd - : A64I_LSpostind<0b10, 0b0, 0b10, - (outs GPR64:$Rt, GPR64xsp:$Rn_wb), - (ins GPR64xsp:$Rn, simm9:$SImm9), - "ldrsw\t$Rt, [$Rn], $SImm9", - [], NoItinerary> { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; -} - -def LDRSWx_PreInd : A64I_LSpreind<0b10, 0b0, 0b10, - (outs GPR64:$Rt, GPR64xsp:$Rn_wb), - (ins GPR64xsp:$Rn, simm9:$SImm9), - "ldrsw\t$Rt, [$Rn, $SImm9]!", - [], NoItinerary> { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeSingleIndexedInstruction"; -} - -//===------------------------------ -// 2.4 Prefetch operations -//===------------------------------ - -def PRFM : A64I_LSunsigimm<0b11, 0b0, 0b10, (outs), - (ins prefetch_op:$Rt, GPR64xsp:$Rn, dword_uimm12:$UImm12), - "prfm\t$Rt, [$Rn, $UImm12]", - [], NoItinerary> { - let mayLoad = 1; -} -def : InstAlias<"prfm $Rt, [$Rn]", - (PRFM prefetch_op:$Rt, GPR64xsp:$Rn, 0)>; - -let mayLoad = 1 in { - def PRFM_Wm_RegOffset : A64I_LSregoff<0b11, 0b0, 0b10, 0b0, (outs), - (ins prefetch_op:$Rt, GPR64xsp:$Rn, - GPR32:$Rm, dword_Wm_regext:$Ext), - "prfm\t$Rt, [$Rn, $Rm, $Ext]", - [], NoItinerary>; - def PRFM_Xm_RegOffset : A64I_LSregoff<0b11, 0b0, 0b10, 0b1, (outs), - (ins prefetch_op:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, dword_Xm_regext:$Ext), - "prfm\t$Rt, [$Rn, $Rm, $Ext]", - [], NoItinerary>; -} - -def : InstAlias<"prfm $Rt, [$Rn, $Rm]", - (PRFM_Xm_RegOffset prefetch_op:$Rt, GPR64xsp:$Rn, - GPR64:$Rm, 2)>; - - -def PRFUM : A64I_LSunalimm<0b11, 0b0, 0b10, (outs), - (ins prefetch_op:$Rt, GPR64xsp:$Rn, simm9:$SImm9), - "prfum\t$Rt, [$Rn, $SImm9]", - [], NoItinerary> { - let mayLoad = 1; -} -def : InstAlias<"prfum $Rt, [$Rn]", - (PRFUM prefetch_op:$Rt, GPR64xsp:$Rn, 0)>; - -//===----------------------------------------------------------------------===// -// Load-store register (unprivileged) instructions -//===----------------------------------------------------------------------===// -// Contains: LDTRB, LDTRH, LDTRSB, LDTRSH, LDTRSW, STTR, STTRB and STTRH - -// These instructions very much mirror the "unscaled immediate" loads, but since -// there are no floating-point variants we need to split them out into their own -// section to avoid instantiation of "ldtr d0, [sp]" etc. - -multiclass A64I_LDTRSTTR size, string asmsuffix, RegisterClass GPR, - string prefix> { - def _UnPriv_STR : A64I_LSunpriv { - let mayStore = 1; - } - - def : InstAlias<"sttr" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_UnPriv_STR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - - def _UnPriv_LDR : A64I_LSunpriv { - let mayLoad = 1; - } - - def : InstAlias<"ldtr" # asmsuffix # " $Rt, [$Rn]", - (!cast(prefix # "_UnPriv_LDR") GPR:$Rt, GPR64xsp:$Rn, 0)>; - -} - -// STTRB/LDTRB: First define the instructions -defm LS8 : A64I_LDTRSTTR<0b00, "b", GPR32, "LS8">; - -// STTRH/LDTRH -defm LS16 : A64I_LDTRSTTR<0b01, "h", GPR32, "LS16">; - -// STTR/LDTR to/from a W register -defm LS32 : A64I_LDTRSTTR<0b10, "", GPR32, "LS32">; - -// STTR/LDTR to/from an X register -defm LS64 : A64I_LDTRSTTR<0b11, "", GPR64, "LS64">; - -// Now a class for the signed instructions that can go to either 32 or 64 -// bits... -multiclass A64I_LDTR_signed size, string asmopcode, string prefix> { - let mayLoad = 1 in { - def w : A64I_LSunpriv; - - def x : A64I_LSunpriv; - } - - def : InstAlias<"ldtrs" # asmopcode # " $Rt, [$Rn]", - (!cast(prefix # "w") GPR32:$Rt, GPR64xsp:$Rn, 0)>; - - def : InstAlias<"ldtrs" # asmopcode # " $Rt, [$Rn]", - (!cast(prefix # "x") GPR64:$Rt, GPR64xsp:$Rn, 0)>; - -} - -// LDTRSB -defm LDTRSB : A64I_LDTR_signed<0b00, "b", "LDTRSB">; -// LDTRSH -defm LDTRSH : A64I_LDTR_signed<0b01, "h", "LDTRSH">; - -// And finally LDTRSW which only goes to 64 bits. -def LDTRSWx : A64I_LSunpriv<0b10, 0b0, 0b10, - (outs GPR64:$Rt), - (ins GPR64xsp:$Rn, simm9:$SImm9), - "ldtrsw\t$Rt, [$Rn, $SImm9]", - [], NoItinerary> { - let mayLoad = 1; -} -def : InstAlias<"ldtrsw $Rt, [$Rn]", (LDTRSWx GPR64:$Rt, GPR64xsp:$Rn, 0)>; - -//===----------------------------------------------------------------------===// -// Load-store register pair (offset) instructions -//===----------------------------------------------------------------------===// +// In big endian mode every memory access has an implicit byte swap. LDR and +// STR do a 64-bit byte swap, whereas LD1/ST1 do a byte swap per lane - that +// is, they treat the vector as a sequence of elements to be byte-swapped. +// The two pairs of instructions are fundamentally incompatible. We've decided +// to use LD1/ST1 only to simplify compiler implementation. // -// and +// LD1/ST1 perform the equivalent of a sequence of LDR/STR + REV. This makes +// the original code sequence: +// v0 = load v2i32 +// v1 = REV v2i32 (implicit) +// v2 = BITCAST v2i32 v1 to v4i16 +// v3 = REV v4i16 v2 (implicit) +// store v4i16 v3 // -//===----------------------------------------------------------------------===// -// Load-store register pair (post-indexed) instructions -//===----------------------------------------------------------------------===// -// Contains: STP, LDP, LDPSW +// But this is now broken - the value stored is different to the value loaded +// due to lane reordering. To fix this, on every BITCAST we must perform two +// other REVs: +// v0 = load v2i32 +// v1 = REV v2i32 (implicit) +// v2 = REV v2i32 +// v3 = BITCAST v2i32 v2 to v4i16 +// v4 = REV v4i16 +// v5 = REV v4i16 v4 (implicit) +// store v4i16 v5 // -// and +// This means an extra two instructions, but actually in most cases the two REV +// instructions can be combined into one. For example: +// (REV64_2s (REV64_4h X)) === (REV32_4h X) // -//===----------------------------------------------------------------------===// -// Load-store register pair (pre-indexed) instructions -//===----------------------------------------------------------------------===// -// Contains: STP, LDP, LDPSW +// There is also no 128-bit REV instruction. This must be synthesized with an +// EXT instruction. // -// and +// Most bitconverts require some sort of conversion. The only exceptions are: +// a) Identity conversions - vNfX <-> vNiX +// b) Single-lane-to-scalar - v1fX <-> fX or v1iX <-> iX // -//===----------------------------------------------------------------------===// -// Load-store non-temporal register pair (offset) instructions -//===----------------------------------------------------------------------===// -// Contains: STNP, LDNP - - -// Anything that creates an MCInst (Decoding, selection and AsmParsing) has to -// know the access size via some means. An isolated operand does not have this -// information unless told from here, which means we need separate tablegen -// Operands for each access size. This multiclass takes care of instantiating -// the correct template functions in the rest of the backend. - -multiclass offsets_simm7 { - // The bare signed 7-bit immediate is used in post-indexed instructions, but - // because of the scaling performed a generic "simm7" operand isn't - // appropriate here either. - def simm7_asmoperand : AsmOperandClass { - let Name = "SImm7_Scaled" # MemSize; - let PredicateMethod = "isSImm7Scaled<" # MemSize # ">"; - let RenderMethod = "addSImm7ScaledOperands<" # MemSize # ">"; - let DiagnosticType = "LoadStoreSImm7_" # MemSize; - } - - def simm7 : Operand { - let PrintMethod = "printSImm7ScaledOperand<" # MemSize # ">"; - let ParserMatchClass = !cast(prefix # "simm7_asmoperand"); - } -} - -defm word_ : offsets_simm7<"4", "word_">; -defm dword_ : offsets_simm7<"8", "dword_">; -defm qword_ : offsets_simm7<"16", "qword_">; - -multiclass A64I_LSPsimple opc, bit v, RegisterClass SomeReg, - Operand simm7, string prefix> { - def _STR : A64I_LSPoffset { - let mayStore = 1; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - def : InstAlias<"stp $Rt, $Rt2, [$Rn]", - (!cast(prefix # "_STR") SomeReg:$Rt, - SomeReg:$Rt2, GPR64xsp:$Rn, 0)>; - - def _LDR : A64I_LSPoffset { - let mayLoad = 1; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - def : InstAlias<"ldp $Rt, $Rt2, [$Rn]", - (!cast(prefix # "_LDR") SomeReg:$Rt, - SomeReg:$Rt2, GPR64xsp:$Rn, 0)>; - - def _PostInd_STR : A64I_LSPpostind { - let mayStore = 1; - let Constraints = "$Rn = $Rn_wb"; - - // Decoder only needed for unpredictability checking (FIXME). - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - - def _PostInd_LDR : A64I_LSPpostind { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - - def _PreInd_STR : A64I_LSPpreind { - let mayStore = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - - def _PreInd_LDR : A64I_LSPpreind { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - - def _NonTemp_STR : A64I_LSPnontemp { - let mayStore = 1; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - def : InstAlias<"stnp $Rt, $Rt2, [$Rn]", - (!cast(prefix # "_NonTemp_STR") SomeReg:$Rt, - SomeReg:$Rt2, GPR64xsp:$Rn, 0)>; - - def _NonTemp_LDR : A64I_LSPnontemp { - let mayLoad = 1; - let DecoderMethod = "DecodeLDSTPairInstruction"; - } - def : InstAlias<"ldnp $Rt, $Rt2, [$Rn]", - (!cast(prefix # "_NonTemp_LDR") SomeReg:$Rt, - SomeReg:$Rt2, GPR64xsp:$Rn, 0)>; - -} - - -defm LSPair32 : A64I_LSPsimple<0b00, 0b0, GPR32, word_simm7, "LSPair32">; -defm LSPair64 : A64I_LSPsimple<0b10, 0b0, GPR64, dword_simm7, "LSPair64">; -defm LSFPPair32 : A64I_LSPsimple<0b00, 0b1, FPR32, word_simm7, "LSFPPair32">; -defm LSFPPair64 : A64I_LSPsimple<0b01, 0b1, FPR64, dword_simm7, "LSFPPair64">; -defm LSFPPair128 : A64I_LSPsimple<0b10, 0b1, FPR128, qword_simm7, - "LSFPPair128">; - - -def LDPSWx : A64I_LSPoffset<0b01, 0b0, 0b1, - (outs GPR64:$Rt, GPR64:$Rt2), - (ins GPR64xsp:$Rn, word_simm7:$SImm7), - "ldpsw\t$Rt, $Rt2, [$Rn, $SImm7]", [], NoItinerary> { - let mayLoad = 1; - let DecoderMethod = "DecodeLDSTPairInstruction"; -} -def : InstAlias<"ldpsw $Rt, $Rt2, [$Rn]", - (LDPSWx GPR64:$Rt, GPR64:$Rt2, GPR64xsp:$Rn, 0)>; - -def LDPSWx_PostInd : A64I_LSPpostind<0b01, 0b0, 0b1, - (outs GPR64:$Rt, GPR64:$Rt2, GPR64:$Rn_wb), - (ins GPR64xsp:$Rn, word_simm7:$SImm7), - "ldpsw\t$Rt, $Rt2, [$Rn], $SImm7", - [], NoItinerary> { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeLDSTPairInstruction"; -} - -def LDPSWx_PreInd : A64I_LSPpreind<0b01, 0b0, 0b1, - (outs GPR64:$Rt, GPR64:$Rt2, GPR64:$Rn_wb), - (ins GPR64xsp:$Rn, word_simm7:$SImm7), - "ldpsw\t$Rt, $Rt2, [$Rn, $SImm7]!", - [], NoItinerary> { - let mayLoad = 1; - let Constraints = "$Rn = $Rn_wb"; - let DecoderMethod = "DecodeLDSTPairInstruction"; -} - -//===----------------------------------------------------------------------===// -// Logical (immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: AND, ORR, EOR, ANDS, + aliases TST, MOV - -multiclass logical_imm_operands { - def _asmoperand : AsmOperandClass { - let Name = "LogicalImm" # note # size; - let PredicateMethod = "isLogicalImm" # note # "<" # size # ">"; - let RenderMethod = "addLogicalImmOperands<" # size # ">"; - let DiagnosticType = "LogicalSecondSource"; - } - - def _operand - : Operand, ComplexPattern { - let ParserMatchClass = !cast(prefix # "_asmoperand"); - let PrintMethod = "printLogicalImmOperand<" # size # ">"; - let DecoderMethod = "DecodeLogicalImmOperand<" # size # ">"; - } -} - -defm logical_imm32 : logical_imm_operands<"logical_imm32", "", 32, i32>; -defm logical_imm64 : logical_imm_operands<"logical_imm64", "", 64, i64>; - -// The mov versions only differ in assembly parsing, where they -// exclude values representable with either MOVZ or MOVN. -defm logical_imm32_mov - : logical_imm_operands<"logical_imm32_mov", "MOV", 32, i32>; -defm logical_imm64_mov - : logical_imm_operands<"logical_imm64_mov", "MOV", 64, i64>; - - -multiclass A64I_logimmSizes opc, string asmop, SDNode opnode> { - def wwi : A64I_logicalimm<0b0, opc, (outs GPR32wsp:$Rd), - (ins GPR32:$Rn, logical_imm32_operand:$Imm), - !strconcat(asmop, "\t$Rd, $Rn, $Imm"), - [(set GPR32wsp:$Rd, - (opnode GPR32:$Rn, logical_imm32_operand:$Imm))], - NoItinerary>; - - def xxi : A64I_logicalimm<0b1, opc, (outs GPR64xsp:$Rd), - (ins GPR64:$Rn, logical_imm64_operand:$Imm), - !strconcat(asmop, "\t$Rd, $Rn, $Imm"), - [(set GPR64xsp:$Rd, - (opnode GPR64:$Rn, logical_imm64_operand:$Imm))], - NoItinerary>; -} - -defm AND : A64I_logimmSizes<0b00, "and", and>; -defm ORR : A64I_logimmSizes<0b01, "orr", or>; -defm EOR : A64I_logimmSizes<0b10, "eor", xor>; - -let Defs = [NZCV] in { - def ANDSwwi : A64I_logicalimm<0b0, 0b11, (outs GPR32:$Rd), - (ins GPR32:$Rn, logical_imm32_operand:$Imm), - "ands\t$Rd, $Rn, $Imm", - [], NoItinerary>; - - def ANDSxxi : A64I_logicalimm<0b1, 0b11, (outs GPR64:$Rd), - (ins GPR64:$Rn, logical_imm64_operand:$Imm), - "ands\t$Rd, $Rn, $Imm", - [], NoItinerary>; -} - - -def : InstAlias<"tst $Rn, $Imm", - (ANDSwwi WZR, GPR32:$Rn, logical_imm32_operand:$Imm)>; -def : InstAlias<"tst $Rn, $Imm", - (ANDSxxi XZR, GPR64:$Rn, logical_imm64_operand:$Imm)>; -def : InstAlias<"mov $Rd, $Imm", - (ORRwwi GPR32wsp:$Rd, WZR, logical_imm32_mov_operand:$Imm)>; -def : InstAlias<"mov $Rd, $Imm", - (ORRxxi GPR64xsp:$Rd, XZR, logical_imm64_mov_operand:$Imm)>; - -//===----------------------------------------------------------------------===// -// Logical (shifted register) instructions -//===----------------------------------------------------------------------===// -// Contains: AND, BIC, ORR, ORN, EOR, EON, ANDS, BICS + aliases TST, MVN, MOV - -// Operand for optimizing (icmp (and LHS, RHS), 0, SomeCode). In theory "ANDS" -// behaves differently for unsigned comparisons, so we defensively only allow -// signed or n/a as the operand. In practice "unsigned greater than 0" is "not -// equal to 0" and LLVM gives us this. -def signed_cond : PatLeaf<(cond), [{ - return !isUnsignedIntSetCC(N->get()); -}]>; - - -// These instructions share their "shift" operands with add/sub (shifted -// register instructions). They are defined there. - -// N.b. the commutable parameter is just !N. It will be first against the wall -// when the revolution comes. -multiclass logical_shifts opc, - bit N, bit commutable, - string asmop, SDPatternOperator opfrag, string sty, - RegisterClass GPR, list defs> { - let isCommutable = commutable, Defs = defs in { - def _lsl : A64I_logicalshift("lsl_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (shl GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _lsr : A64I_logicalshift("lsr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (srl GPR:$Rm, - !cast("lsr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _asr : A64I_logicalshift("asr_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (sra GPR:$Rm, - !cast("asr_operand_" # sty):$Imm6)) - )], - NoItinerary>; - - def _ror : A64I_logicalshift("ror_operand_" # sty):$Imm6), - !strconcat(asmop, "\t$Rd, $Rn, $Rm, $Imm6"), - [(set GPR:$Rd, (opfrag GPR:$Rn, (rotr GPR:$Rm, - !cast("ror_operand_" # sty):$Imm6)) - )], - NoItinerary>; - } - - def _noshift - : InstAlias(prefix # "_lsl") GPR:$Rd, GPR:$Rn, - GPR:$Rm, 0)>; - - def : Pat<(opfrag GPR:$Rn, GPR:$Rm), - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; -} - -multiclass logical_sizes opc, bit N, bit commutable, - string asmop, SDPatternOperator opfrag, - list defs> { - defm xxx : logical_shifts; - defm www : logical_shifts; -} - - -defm AND : logical_sizes<"AND", 0b00, 0b0, 0b1, "and", and, []>; -defm ORR : logical_sizes<"ORR", 0b01, 0b0, 0b1, "orr", or, []>; -defm EOR : logical_sizes<"EOR", 0b10, 0b0, 0b1, "eor", xor, []>; -defm ANDS : logical_sizes<"ANDS", 0b11, 0b0, 0b1, "ands", - PatFrag<(ops node:$lhs, node:$rhs), (and node:$lhs, node:$rhs), - [{ (void)N; return false; }]>, - [NZCV]>; - -defm BIC : logical_sizes<"BIC", 0b00, 0b1, 0b0, "bic", - PatFrag<(ops node:$lhs, node:$rhs), - (and node:$lhs, (not node:$rhs))>, []>; -defm ORN : logical_sizes<"ORN", 0b01, 0b1, 0b0, "orn", - PatFrag<(ops node:$lhs, node:$rhs), - (or node:$lhs, (not node:$rhs))>, []>; -defm EON : logical_sizes<"EON", 0b10, 0b1, 0b0, "eon", - PatFrag<(ops node:$lhs, node:$rhs), - (xor node:$lhs, (not node:$rhs))>, []>; -defm BICS : logical_sizes<"BICS", 0b11, 0b1, 0b0, "bics", - PatFrag<(ops node:$lhs, node:$rhs), - (and node:$lhs, (not node:$rhs)), - [{ (void)N; return false; }]>, - [NZCV]>; - -multiclass tst_shifts { - let isCommutable = 1, Rd = 0b11111, Defs = [NZCV] in { - def _lsl : A64I_logicalshift("lsl_operand_" # sty):$Imm6), - "tst\t$Rn, $Rm, $Imm6", - [(set NZCV, (A64setcc (and GPR:$Rn, (shl GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)), - 0, signed_cond))], - NoItinerary>; - - - def _lsr : A64I_logicalshift("lsr_operand_" # sty):$Imm6), - "tst\t$Rn, $Rm, $Imm6", - [(set NZCV, (A64setcc (and GPR:$Rn, (srl GPR:$Rm, - !cast("lsr_operand_" # sty):$Imm6)), - 0, signed_cond))], - NoItinerary>; - - def _asr : A64I_logicalshift("asr_operand_" # sty):$Imm6), - "tst\t$Rn, $Rm, $Imm6", - [(set NZCV, (A64setcc (and GPR:$Rn, (sra GPR:$Rm, - !cast("asr_operand_" # sty):$Imm6)), - 0, signed_cond))], - NoItinerary>; - - def _ror : A64I_logicalshift("ror_operand_" # sty):$Imm6), - "tst\t$Rn, $Rm, $Imm6", - [(set NZCV, (A64setcc (and GPR:$Rn, (rotr GPR:$Rm, - !cast("ror_operand_" # sty):$Imm6)), - 0, signed_cond))], - NoItinerary>; - } - - def _noshift : InstAlias<"tst $Rn, $Rm", - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; - - def : Pat<(A64setcc (and GPR:$Rn, GPR:$Rm), 0, signed_cond), - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; -} - -defm TSTxx : tst_shifts<"TSTxx", 0b1, "i64", GPR64>; -defm TSTww : tst_shifts<"TSTww", 0b0, "i32", GPR32>; - - -multiclass mvn_shifts { - let isCommutable = 0, Rn = 0b11111 in { - def _lsl : A64I_logicalshift("lsl_operand_" # sty):$Imm6), - "mvn\t$Rd, $Rm, $Imm6", - [(set GPR:$Rd, (not (shl GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)))], - NoItinerary>; - - - def _lsr : A64I_logicalshift("lsr_operand_" # sty):$Imm6), - "mvn\t$Rd, $Rm, $Imm6", - [(set GPR:$Rd, (not (srl GPR:$Rm, - !cast("lsr_operand_" # sty):$Imm6)))], - NoItinerary>; - - def _asr : A64I_logicalshift("asr_operand_" # sty):$Imm6), - "mvn\t$Rd, $Rm, $Imm6", - [(set GPR:$Rd, (not (sra GPR:$Rm, - !cast("asr_operand_" # sty):$Imm6)))], - NoItinerary>; - - def _ror : A64I_logicalshift("ror_operand_" # sty):$Imm6), - "mvn\t$Rd, $Rm, $Imm6", - [(set GPR:$Rd, (not (rotr GPR:$Rm, - !cast("lsl_operand_" # sty):$Imm6)))], - NoItinerary>; - } - - def _noshift : InstAlias<"mvn $Rn, $Rm", - (!cast(prefix # "_lsl") GPR:$Rn, GPR:$Rm, 0)>; - - def : Pat<(not GPR:$Rm), - (!cast(prefix # "_lsl") GPR:$Rm, 0)>; -} - -defm MVNxx : mvn_shifts<"MVNxx", 0b1, "i64", GPR64>; -defm MVNww : mvn_shifts<"MVNww", 0b0, "i32", GPR32>; - -def MOVxx :InstAlias<"mov $Rd, $Rm", (ORRxxx_lsl GPR64:$Rd, XZR, GPR64:$Rm, 0)>; -def MOVww :InstAlias<"mov $Rd, $Rm", (ORRwww_lsl GPR32:$Rd, WZR, GPR32:$Rm, 0)>; - -//===----------------------------------------------------------------------===// -// Move wide (immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: MOVN, MOVZ, MOVK + MOV aliases - -// A wide variety of different relocations are needed for variants of these -// instructions, so it turns out that we need a different operand for all of -// them. -multiclass movw_operands { - def _imm_asmoperand : AsmOperandClass { - let Name = instname # width # "Shifted" # shift; - let PredicateMethod = "is" # instname # width # "Imm"; - let RenderMethod = "addMoveWideImmOperands"; - let ParserMethod = "ParseImmWithLSLOperand"; - let DiagnosticType = "MOVWUImm16"; - } - - def _imm : Operand { - let ParserMatchClass = !cast(prefix # "_imm_asmoperand"); - let PrintMethod = "printMoveWideImmOperand"; - let EncoderMethod = "getMoveWideImmOpValue"; - let DecoderMethod = "DecodeMoveWideImmOperand<" # width # ">"; - - let MIOperandInfo = (ops uimm16:$UImm16, imm:$Shift); - } -} - -defm movn32 : movw_operands<"movn32", "MOVN", 32>; -defm movn64 : movw_operands<"movn64", "MOVN", 64>; -defm movz32 : movw_operands<"movz32", "MOVZ", 32>; -defm movz64 : movw_operands<"movz64", "MOVZ", 64>; -defm movk32 : movw_operands<"movk32", "MOVK", 32>; -defm movk64 : movw_operands<"movk64", "MOVK", 64>; - -multiclass A64I_movwSizes opc, string asmop, dag ins32bit, - dag ins64bit> { - - def wii : A64I_movw<0b0, opc, (outs GPR32:$Rd), ins32bit, - !strconcat(asmop, "\t$Rd, $FullImm"), - [], NoItinerary> { - bits<18> FullImm; - let UImm16 = FullImm{15-0}; - let Shift = FullImm{17-16}; - } - - def xii : A64I_movw<0b1, opc, (outs GPR64:$Rd), ins64bit, - !strconcat(asmop, "\t$Rd, $FullImm"), - [], NoItinerary> { - bits<18> FullImm; - let UImm16 = FullImm{15-0}; - let Shift = FullImm{17-16}; - } -} - -let isMoveImm = 1, isReMaterializable = 1, - isAsCheapAsAMove = 1, hasSideEffects = 0 in { - defm MOVN : A64I_movwSizes<0b00, "movn", - (ins movn32_imm:$FullImm), - (ins movn64_imm:$FullImm)>; - - // Some relocations are able to convert between a MOVZ and a MOVN. If these - // are applied the instruction must be emitted with the corresponding bits as - // 0, which means a MOVZ needs to override that bit from the default. - let PostEncoderMethod = "fixMOVZ" in - defm MOVZ : A64I_movwSizes<0b10, "movz", - (ins movz32_imm:$FullImm), - (ins movz64_imm:$FullImm)>; -} - -let Constraints = "$src = $Rd" in -defm MOVK : A64I_movwSizes<0b11, "movk", - (ins GPR32:$src, movk32_imm:$FullImm), - (ins GPR64:$src, movk64_imm:$FullImm)>; - - -// And now the "MOV" aliases. These also need their own operands because what -// they accept is completely different to what the base instructions accept. -multiclass movalias_operand { - def _asmoperand : AsmOperandClass { - let Name = basename # width # "MovAlias"; - let PredicateMethod - = "isMoveWideMovAlias<" # width # ", A64Imms::" # immpredicate # ">"; - let RenderMethod - = "addMoveWideMovAliasOperands<" # width # ", " - # "A64Imms::" # immpredicate # ">"; - } - - def _movimm : Operand { - let ParserMatchClass = !cast(prefix # "_asmoperand"); - - let MIOperandInfo = (ops uimm16:$UImm16, imm:$Shift); - } -} - -defm movz32 : movalias_operand<"movz32", "MOVZ", "isMOVZImm", 32>; -defm movz64 : movalias_operand<"movz64", "MOVZ", "isMOVZImm", 64>; -defm movn32 : movalias_operand<"movn32", "MOVN", "isOnlyMOVNImm", 32>; -defm movn64 : movalias_operand<"movn64", "MOVN", "isOnlyMOVNImm", 64>; - -// FIXME: these are officially canonical aliases, but TableGen is too limited to -// print them at the moment. I believe in this case an "AliasPredicate" method -// will need to be implemented. to allow it, as well as the more generally -// useful handling of non-register, non-constant operands. -class movalias - : InstAlias<"mov $Rd, $FullImm", (INST GPR:$Rd, operand:$FullImm)>; - -def : movalias; -def : movalias; -def : movalias; -def : movalias; - -//===----------------------------------------------------------------------===// -// PC-relative addressing instructions -//===----------------------------------------------------------------------===// -// Contains: ADR, ADRP - -def adr_label : Operand { - let EncoderMethod = "getLabelOpValue"; - - // This label is a 21-bit offset from PC, unscaled - let PrintMethod = "printLabelOperand<21, 1>"; - let ParserMatchClass = label_asmoperand<21, 1>; - let OperandType = "OPERAND_PCREL"; -} - -def adrp_label_asmoperand : AsmOperandClass { - let Name = "AdrpLabel"; - let RenderMethod = "addLabelOperands<21, 4096>"; - let DiagnosticType = "Label"; -} - -def adrp_label : Operand { - let EncoderMethod = "getAdrpLabelOpValue"; - - // This label is a 21-bit offset from PC, scaled by the page-size: 4096. - let PrintMethod = "printLabelOperand<21, 4096>"; - let ParserMatchClass = adrp_label_asmoperand; - let OperandType = "OPERAND_PCREL"; -} - -let hasSideEffects = 0 in { - def ADRxi : A64I_PCADR<0b0, (outs GPR64:$Rd), (ins adr_label:$Label), - "adr\t$Rd, $Label", [], NoItinerary>; - - def ADRPxi : A64I_PCADR<0b1, (outs GPR64:$Rd), (ins adrp_label:$Label), - "adrp\t$Rd, $Label", [], NoItinerary>; -} - -//===----------------------------------------------------------------------===// -// System instructions -//===----------------------------------------------------------------------===// -// Contains: HINT, CLREX, DSB, DMB, ISB, MSR, SYS, SYSL, MRS -// + aliases IC, DC, AT, TLBI, NOP, YIELD, WFE, WFI, SEV, SEVL - -// Op1 and Op2 fields are sometimes simple 3-bit unsigned immediate values. -def uimm3_asmoperand : AsmOperandClass { - let Name = "UImm3"; - let PredicateMethod = "isUImm<3>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm3"; -} - -def uimm3 : Operand { - let ParserMatchClass = uimm3_asmoperand; -} - -// The HINT alias can accept a simple unsigned 7-bit immediate. -def uimm7_asmoperand : AsmOperandClass { - let Name = "UImm7"; - let PredicateMethod = "isUImm<7>"; - let RenderMethod = "addImmOperands"; - let DiagnosticType = "UImm7"; -} - -def uimm7 : Operand { - let ParserMatchClass = uimm7_asmoperand; -} - -// Multiclass namedimm is defined with the prefetch operands. Most of these fit -// into the NamedImmMapper scheme well: they either accept a named operand or -// any immediate under a particular value (which may be 0, implying no immediate -// is allowed). -defm dbarrier : namedimm<"dbarrier", "A64DB::DBarrierMapper">; -defm isb : namedimm<"isb", "A64ISB::ISBMapper">; -defm ic : namedimm<"ic", "A64IC::ICMapper">; -defm dc : namedimm<"dc", "A64DC::DCMapper">; -defm at : namedimm<"at", "A64AT::ATMapper">; -defm tlbi : namedimm<"tlbi", "A64TLBI::TLBIMapper">; - -// However, MRS and MSR are more complicated for a few reasons: -// * There are ~1000 generic names S3____ which have an -// implementation-defined effect -// * Most registers are shared, but some are read-only or write-only. -// * There is a variant of MSR which accepts the same register name (SPSel), -// but which would have a different encoding. - -// In principle these could be resolved in with more complicated subclasses of -// NamedImmMapper, however that imposes an overhead on other "named -// immediates". Both in concrete terms with virtual tables and in unnecessary -// abstraction. - -// The solution adopted here is to take the MRS/MSR Mappers out of the usual -// hierarchy (they're not derived from NamedImmMapper) and to add logic for -// their special situation. -def mrs_asmoperand : AsmOperandClass { - let Name = "MRS"; - let ParserMethod = "ParseSysRegOperand"; - let DiagnosticType = "MRS"; -} - -def mrs_op : Operand { - let ParserMatchClass = mrs_asmoperand; - let PrintMethod = "printMRSOperand"; - let DecoderMethod = "DecodeMRSOperand"; -} - -def msr_asmoperand : AsmOperandClass { - let Name = "MSRWithReg"; - - // Note that SPSel is valid for both this and the pstate operands, but with - // different immediate encodings. This is why these operands provide a string - // AArch64Operand rather than an immediate. The overlap is small enough that - // it could be resolved with hackery now, but who can say in future? - let ParserMethod = "ParseSysRegOperand"; - let DiagnosticType = "MSR"; -} - -def msr_op : Operand { - let ParserMatchClass = msr_asmoperand; - let PrintMethod = "printMSROperand"; - let DecoderMethod = "DecodeMSROperand"; -} - -def pstate_asmoperand : AsmOperandClass { - let Name = "MSRPState"; - // See comment above about parser. - let ParserMethod = "ParseSysRegOperand"; - let DiagnosticType = "MSR"; -} - -def pstate_op : Operand { - let ParserMatchClass = pstate_asmoperand; - let PrintMethod = "printNamedImmOperand"; - let DecoderMethod = "DecodeNamedImmOperand"; -} - -// When is specified, an assembler should accept something like "C4", not -// the usual "#4" immediate. -def CRx_asmoperand : AsmOperandClass { - let Name = "CRx"; - let PredicateMethod = "isUImm<4>"; - let RenderMethod = "addImmOperands"; - let ParserMethod = "ParseCRxOperand"; - // Diagnostics are handled in all cases by ParseCRxOperand. -} - -def CRx : Operand { - let ParserMatchClass = CRx_asmoperand; - let PrintMethod = "printCRxOperand"; -} - - -// Finally, we can start defining the instructions. - -// HINT is straightforward, with a few aliases. -def HINTi : A64I_system<0b0, (outs), (ins uimm7:$UImm7), "hint\t$UImm7", - [], NoItinerary> { - bits<7> UImm7; - let CRm = UImm7{6-3}; - let Op2 = UImm7{2-0}; - - let Op0 = 0b00; - let Op1 = 0b011; - let CRn = 0b0010; - let Rt = 0b11111; -} - -def : InstAlias<"nop", (HINTi 0)>; -def : InstAlias<"yield", (HINTi 1)>; -def : InstAlias<"wfe", (HINTi 2)>; -def : InstAlias<"wfi", (HINTi 3)>; -def : InstAlias<"sev", (HINTi 4)>; -def : InstAlias<"sevl", (HINTi 5)>; - -// Quite a few instructions then follow a similar pattern of fixing common -// fields in the bitpattern, we'll define a helper-class for them. -class simple_sys op0, bits<3> op1, bits<4> crn, bits<3> op2, - Operand operand, string asmop> - : A64I_system<0b0, (outs), (ins operand:$CRm), !strconcat(asmop, "\t$CRm"), - [], NoItinerary> { - let Op0 = op0; - let Op1 = op1; - let CRn = crn; - let Op2 = op2; - let Rt = 0b11111; -} - - -def CLREXi : simple_sys<0b00, 0b011, 0b0011, 0b010, uimm4, "clrex">; -def DSBi : simple_sys<0b00, 0b011, 0b0011, 0b100, dbarrier_op, "dsb">; -def DMBi : simple_sys<0b00, 0b011, 0b0011, 0b101, dbarrier_op, "dmb">; -def ISBi : simple_sys<0b00, 0b011, 0b0011, 0b110, isb_op, "isb">; - -def : InstAlias<"clrex", (CLREXi 0b1111)>; -def : InstAlias<"isb", (ISBi 0b1111)>; - -// (DMBi 0xb) is a "DMB ISH" instruciton, appropriate for Linux SMP -// configurations at least. -def : Pat<(atomic_fence imm, imm), (DMBi 0xb)>; - -// Any SYS bitpattern can be represented with a complex and opaque "SYS" -// instruction. -def SYSiccix : A64I_system<0b0, (outs), - (ins uimm3:$Op1, CRx:$CRn, CRx:$CRm, - uimm3:$Op2, GPR64:$Rt), - "sys\t$Op1, $CRn, $CRm, $Op2, $Rt", - [], NoItinerary> { - let Op0 = 0b01; -} - -// You can skip the Xt argument whether it makes sense or not for the generic -// SYS instruction. -def : InstAlias<"sys $Op1, $CRn, $CRm, $Op2", - (SYSiccix uimm3:$Op1, CRx:$CRn, CRx:$CRm, uimm3:$Op2, XZR)>; - - -// But many have aliases, which obviously don't fit into -class SYSalias - : A64I_system<0b0, (outs), ins, asmstring, [], NoItinerary> { - let isAsmParserOnly = 1; - - bits<14> SysOp; - let Op0 = 0b01; - let Op1 = SysOp{13-11}; - let CRn = SysOp{10-7}; - let CRm = SysOp{6-3}; - let Op2 = SysOp{2-0}; -} - -def ICix : SYSalias<(ins ic_op:$SysOp, GPR64:$Rt), "ic\t$SysOp, $Rt">; - -def ICi : SYSalias<(ins ic_op:$SysOp), "ic\t$SysOp"> { - let Rt = 0b11111; -} - -def DCix : SYSalias<(ins dc_op:$SysOp, GPR64:$Rt), "dc\t$SysOp, $Rt">; -def ATix : SYSalias<(ins at_op:$SysOp, GPR64:$Rt), "at\t$SysOp, $Rt">; - -def TLBIix : SYSalias<(ins tlbi_op:$SysOp, GPR64:$Rt), "tlbi\t$SysOp, $Rt">; - -def TLBIi : SYSalias<(ins tlbi_op:$SysOp), "tlbi\t$SysOp"> { - let Rt = 0b11111; -} - - -def SYSLxicci : A64I_system<0b1, (outs GPR64:$Rt), - (ins uimm3:$Op1, CRx:$CRn, CRx:$CRm, uimm3:$Op2), - "sysl\t$Rt, $Op1, $CRn, $CRm, $Op2", - [], NoItinerary> { - let Op0 = 0b01; -} - -// The instructions themselves are rather simple for MSR and MRS. -def MSRix : A64I_system<0b0, (outs), (ins msr_op:$SysReg, GPR64:$Rt), - "msr\t$SysReg, $Rt", [], NoItinerary> { - bits<16> SysReg; - let Op0 = SysReg{15-14}; - let Op1 = SysReg{13-11}; - let CRn = SysReg{10-7}; - let CRm = SysReg{6-3}; - let Op2 = SysReg{2-0}; -} - -def MRSxi : A64I_system<0b1, (outs GPR64:$Rt), (ins mrs_op:$SysReg), - "mrs\t$Rt, $SysReg", [], NoItinerary> { - bits<16> SysReg; - let Op0 = SysReg{15-14}; - let Op1 = SysReg{13-11}; - let CRn = SysReg{10-7}; - let CRm = SysReg{6-3}; - let Op2 = SysReg{2-0}; -} - -def MSRii : A64I_system<0b0, (outs), (ins pstate_op:$PState, uimm4:$CRm), - "msr\t$PState, $CRm", [], NoItinerary> { - bits<6> PState; - - let Op0 = 0b00; - let Op1 = PState{5-3}; - let CRn = 0b0100; - let Op2 = PState{2-0}; - let Rt = 0b11111; -} - -//===----------------------------------------------------------------------===// -// Test & branch (immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: TBZ, TBNZ - -// The bit to test is a simple unsigned 6-bit immediate in the X-register -// versions. -def uimm6 : Operand { - let ParserMatchClass = uimm6_asmoperand; -} - -def label_wid14_scal4_asmoperand : label_asmoperand<14, 4>; - -def tbimm_target : Operand { - let EncoderMethod = "getLabelOpValue"; - - // This label is a 14-bit offset from PC, scaled by the instruction-width: 4. - let PrintMethod = "printLabelOperand<14, 4>"; - let ParserMatchClass = label_wid14_scal4_asmoperand; - - let OperandType = "OPERAND_PCREL"; -} - -def A64eq : ImmLeaf; -def A64ne : ImmLeaf; - -// These instructions correspond to patterns involving "and" with a power of -// two, which we need to be able to select. -def tstb64_pat : ComplexPattern">; -def tstb32_pat : ComplexPattern">; - -let isBranch = 1, isTerminator = 1 in { - def TBZxii : A64I_TBimm<0b0, (outs), - (ins GPR64:$Rt, uimm6:$Imm, tbimm_target:$Label), - "tbz\t$Rt, $Imm, $Label", - [(A64br_cc (A64cmp (and GPR64:$Rt, tstb64_pat:$Imm), 0), - A64eq, bb:$Label)], - NoItinerary>; - - def TBNZxii : A64I_TBimm<0b1, (outs), - (ins GPR64:$Rt, uimm6:$Imm, tbimm_target:$Label), - "tbnz\t$Rt, $Imm, $Label", - [(A64br_cc (A64cmp (and GPR64:$Rt, tstb64_pat:$Imm), 0), - A64ne, bb:$Label)], - NoItinerary>; - - - // Note, these instructions overlap with the above 64-bit patterns. This is - // intentional, "tbz x3, #1, somewhere" and "tbz w3, #1, somewhere" would both - // do the same thing and are both permitted assembly. They also both have - // sensible DAG patterns. - def TBZwii : A64I_TBimm<0b0, (outs), - (ins GPR32:$Rt, uimm5:$Imm, tbimm_target:$Label), - "tbz\t$Rt, $Imm, $Label", - [(A64br_cc (A64cmp (and GPR32:$Rt, tstb32_pat:$Imm), 0), - A64eq, bb:$Label)], - NoItinerary> { - let Imm{5} = 0b0; - } - - def TBNZwii : A64I_TBimm<0b1, (outs), - (ins GPR32:$Rt, uimm5:$Imm, tbimm_target:$Label), - "tbnz\t$Rt, $Imm, $Label", - [(A64br_cc (A64cmp (and GPR32:$Rt, tstb32_pat:$Imm), 0), - A64ne, bb:$Label)], - NoItinerary> { - let Imm{5} = 0b0; - } -} - -//===----------------------------------------------------------------------===// -// Unconditional branch (immediate) instructions -//===----------------------------------------------------------------------===// -// Contains: B, BL - -def label_wid26_scal4_asmoperand : label_asmoperand<26, 4>; - -def bimm_target : Operand { - let EncoderMethod = "getLabelOpValue"; - - // This label is a 26-bit offset from PC, scaled by the instruction-width: 4. - let PrintMethod = "printLabelOperand<26, 4>"; - let ParserMatchClass = label_wid26_scal4_asmoperand; - - let OperandType = "OPERAND_PCREL"; -} - -def blimm_target : Operand { - let EncoderMethod = "getLabelOpValue"; - - // This label is a 26-bit offset from PC, scaled by the instruction-width: 4. - let PrintMethod = "printLabelOperand<26, 4>"; - let ParserMatchClass = label_wid26_scal4_asmoperand; - - let OperandType = "OPERAND_PCREL"; -} - -class A64I_BimmImpl patterns, Operand lbl_type> - : A64I_Bimm; - -let isBranch = 1 in { - def Bimm : A64I_BimmImpl<0b0, "b", [(br bb:$Label)], bimm_target> { - let isTerminator = 1; - let isBarrier = 1; - } - - def BLimm : A64I_BimmImpl<0b1, "bl", - [(AArch64Call tglobaladdr:$Label)], blimm_target> { - let isCall = 1; - let Defs = [X30]; - } -} - -def : Pat<(AArch64Call texternalsym:$Label), (BLimm texternalsym:$Label)>; - -//===----------------------------------------------------------------------===// -// Unconditional branch (register) instructions -//===----------------------------------------------------------------------===// -// Contains: BR, BLR, RET, ERET, DRP. - -// Most of the notional opcode fields in the A64I_Breg format are fixed in A64 -// at the moment. -class A64I_BregImpl opc, - dag outs, dag ins, string asmstr, list patterns, - InstrItinClass itin = NoItinerary> - : A64I_Breg { - let isBranch = 1; - let isIndirectBranch = 1; -} - -// Note that these are not marked isCall or isReturn because as far as LLVM is -// concerned they're not. "ret" is just another jump unless it has been selected -// by LLVM as the function's return. - -let isBranch = 1 in { - def BRx : A64I_BregImpl<0b0000,(outs), (ins GPR64:$Rn), - "br\t$Rn", [(brind GPR64:$Rn)]> { - let isBarrier = 1; - let isTerminator = 1; - } - - def BLRx : A64I_BregImpl<0b0001, (outs), (ins GPR64:$Rn), - "blr\t$Rn", [(AArch64Call GPR64:$Rn)]> { - let isBarrier = 0; - let isCall = 1; - let Defs = [X30]; - } - - def RETx : A64I_BregImpl<0b0010, (outs), (ins GPR64:$Rn), - "ret\t$Rn", []> { - let isBarrier = 1; - let isTerminator = 1; - let isReturn = 1; - } - - // Create a separate pseudo-instruction for codegen to use so that we don't - // flag x30 as used in every function. It'll be restored before the RET by the - // epilogue if it's legitimately used. - def RET : A64PseudoExpand<(outs), (ins), [(A64ret)], (RETx (ops X30))> { - let isTerminator = 1; - let isBarrier = 1; - let isReturn = 1; - } - - def ERET : A64I_BregImpl<0b0100, (outs), (ins), "eret", []> { - let Rn = 0b11111; - let isBarrier = 1; - let isTerminator = 1; - let isReturn = 1; - } - - def DRPS : A64I_BregImpl<0b0101, (outs), (ins), "drps", []> { - let Rn = 0b11111; - let isBarrier = 1; - } -} - -def RETAlias : InstAlias<"ret", (RETx X30)>; - - -//===----------------------------------------------------------------------===// -// Address generation patterns -//===----------------------------------------------------------------------===// - -// Primary method of address generation for the small/absolute memory model is -// an ADRP/ADR pair: -// ADRP x0, some_variable -// ADD x0, x0, #:lo12:some_variable -// -// The load/store elision of the ADD is accomplished when selecting -// addressing-modes. This just mops up the cases where that doesn't work and we -// really need an address in some register. - -// This wrapper applies a LO12 modifier to the address. Otherwise we could just -// use the same address. - -class ADRP_ADD - : Pat<(Wrapper addrop:$Hi, addrop:$Lo12, (i32 imm)), - (ADDxxi_lsl0_s (ADRPxi addrop:$Hi), addrop:$Lo12)>; - -def : ADRP_ADD; -def : ADRP_ADD; -def : ADRP_ADD; -def : ADRP_ADD; -def : ADRP_ADD; - -//===----------------------------------------------------------------------===// -// GOT access patterns -//===----------------------------------------------------------------------===// - -// FIXME: Wibble - -class GOTLoadSmall - : Pat<(A64GOTLoad (A64WrapperSmall addrfrag:$Hi, addrfrag:$Lo12, 8)), - (LS64_LDR (ADRPxi addrfrag:$Hi), addrfrag:$Lo12)>; - -def : GOTLoadSmall; -def : GOTLoadSmall; -def : GOTLoadSmall; - -//===----------------------------------------------------------------------===// -// Tail call handling -//===----------------------------------------------------------------------===// - -let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1, Uses = [XSP] in { - def TC_RETURNdi - : PseudoInst<(outs), (ins i64imm:$dst, i32imm:$FPDiff), - [(AArch64tcret tglobaladdr:$dst, (i32 timm:$FPDiff))]>; - - def TC_RETURNxi - : PseudoInst<(outs), (ins tcGPR64:$dst, i32imm:$FPDiff), - [(AArch64tcret tcGPR64:$dst, (i32 timm:$FPDiff))]>; -} - -let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1, - Uses = [XSP] in { - def TAIL_Bimm : A64PseudoExpand<(outs), (ins bimm_target:$Label), [], - (Bimm bimm_target:$Label)>; - - def TAIL_BRx : A64PseudoExpand<(outs), (ins tcGPR64:$Rd), [], - (BRx GPR64:$Rd)>; -} - +// Natural vector casts (64 bit) +def : Pat<(v8i8 (AArch64NvCast (v2i32 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v4i16 (AArch64NvCast (v2i32 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4f16 (AArch64NvCast (v2i32 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v2i32 (AArch64NvCast (v2i32 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2f32 (AArch64NvCast (v2i32 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v1i64 (AArch64NvCast (v2i32 FPR64:$src))), (v1i64 FPR64:$src)>; + +def : Pat<(v8i8 (AArch64NvCast (v4i16 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v4i16 (AArch64NvCast (v4i16 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4f16 (AArch64NvCast (v4i16 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v2i32 (AArch64NvCast (v4i16 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v1i64 (AArch64NvCast (v4i16 FPR64:$src))), (v1i64 FPR64:$src)>; + +def : Pat<(v8i8 (AArch64NvCast (v8i8 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v4i16 (AArch64NvCast (v8i8 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4f16 (AArch64NvCast (v8i8 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v2i32 (AArch64NvCast (v8i8 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v1i64 (AArch64NvCast (v8i8 FPR64:$src))), (v1i64 FPR64:$src)>; + +def : Pat<(v8i8 (AArch64NvCast (f64 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v4i16 (AArch64NvCast (f64 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4f16 (AArch64NvCast (f64 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v2i32 (AArch64NvCast (f64 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2f32 (AArch64NvCast (f64 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v1i64 (AArch64NvCast (f64 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1f64 (AArch64NvCast (f64 FPR64:$src))), (v1f64 FPR64:$src)>; + +def : Pat<(v8i8 (AArch64NvCast (v2f32 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v4i16 (AArch64NvCast (v2f32 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v2i32 (AArch64NvCast (v2f32 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2f32 (AArch64NvCast (v2f32 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v1i64 (AArch64NvCast (v2f32 FPR64:$src))), (v1i64 FPR64:$src)>; + +// Natural vector casts (128 bit) +def : Pat<(v16i8 (AArch64NvCast (v4i32 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v4i32 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v4i32 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v4i32 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v4i32 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v4i32 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v4i32 FPR128:$src))), (v2f64 FPR128:$src)>; + +def : Pat<(v16i8 (AArch64NvCast (v8i16 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v8i16 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v8i16 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v8i16 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v8i16 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v8i16 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v8i16 FPR128:$src))), (v2f64 FPR128:$src)>; + +def : Pat<(v16i8 (AArch64NvCast (v16i8 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v16i8 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v16i8 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v16i8 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v16i8 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v16i8 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v16i8 FPR128:$src))), (v2f64 FPR128:$src)>; + +def : Pat<(v16i8 (AArch64NvCast (v2i64 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v2i64 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v2i64 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v2i64 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v2i64 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v2i64 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v2i64 FPR128:$src))), (v2f64 FPR128:$src)>; + +def : Pat<(v16i8 (AArch64NvCast (v4f32 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v4f32 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v4f32 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v4f32 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v4f32 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v4f32 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v4f32 FPR128:$src))), (v2f64 FPR128:$src)>; + +def : Pat<(v16i8 (AArch64NvCast (v2f64 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v8i16 (AArch64NvCast (v2f64 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v4i32 (AArch64NvCast (v2f64 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v2i64 (AArch64NvCast (v2f64 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2f64 (AArch64NvCast (v2f64 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v8f16 (AArch64NvCast (v2f64 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v4f32 (AArch64NvCast (v2f64 FPR128:$src))), (v4f32 FPR128:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v8i8 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v4i16 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v2i32 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v4f16 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v2f32 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; + +def : Pat<(i64 (bitconvert (v8i8 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(i64 (bitconvert (v4i16 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(i64 (bitconvert (v2i32 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(i64 (bitconvert (v4f16 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(i64 (bitconvert (v2f32 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(i64 (bitconvert (v1f64 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +} +let Predicates = [IsBE] in { +def : Pat<(v8i8 (bitconvert GPR64:$Xn)), + (REV64v8i8 (COPY_TO_REGCLASS GPR64:$Xn, FPR64))>; +def : Pat<(v4i16 (bitconvert GPR64:$Xn)), + (REV64v4i16 (COPY_TO_REGCLASS GPR64:$Xn, FPR64))>; +def : Pat<(v2i32 (bitconvert GPR64:$Xn)), + (REV64v2i32 (COPY_TO_REGCLASS GPR64:$Xn, FPR64))>; +def : Pat<(v4f16 (bitconvert GPR64:$Xn)), + (REV64v4i16 (COPY_TO_REGCLASS GPR64:$Xn, FPR64))>; +def : Pat<(v2f32 (bitconvert GPR64:$Xn)), + (REV64v2i32 (COPY_TO_REGCLASS GPR64:$Xn, FPR64))>; + +def : Pat<(i64 (bitconvert (v8i8 V64:$Vn))), + (REV64v8i8 (COPY_TO_REGCLASS V64:$Vn, GPR64))>; +def : Pat<(i64 (bitconvert (v4i16 V64:$Vn))), + (REV64v4i16 (COPY_TO_REGCLASS V64:$Vn, GPR64))>; +def : Pat<(i64 (bitconvert (v2i32 V64:$Vn))), + (REV64v2i32 (COPY_TO_REGCLASS V64:$Vn, GPR64))>; +def : Pat<(i64 (bitconvert (v4f16 V64:$Vn))), + (REV64v4i16 (COPY_TO_REGCLASS V64:$Vn, GPR64))>; +def : Pat<(i64 (bitconvert (v2f32 V64:$Vn))), + (REV64v2i32 (COPY_TO_REGCLASS V64:$Vn, GPR64))>; +} +def : Pat<(v1i64 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v1f64 (bitconvert GPR64:$Xn)), (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(i64 (bitconvert (v1i64 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; +def : Pat<(v1i64 (scalar_to_vector GPR64:$Xn)), + (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v1f64 (scalar_to_vector GPR64:$Xn)), + (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(v1f64 (scalar_to_vector (f64 FPR64:$Xn))), (v1f64 FPR64:$Xn)>; + +def : Pat<(f32 (bitconvert (i32 GPR32:$Xn))), + (COPY_TO_REGCLASS GPR32:$Xn, FPR32)>; +def : Pat<(i32 (bitconvert (f32 FPR32:$Xn))), + (COPY_TO_REGCLASS FPR32:$Xn, GPR32)>; +def : Pat<(f64 (bitconvert (i64 GPR64:$Xn))), + (COPY_TO_REGCLASS GPR64:$Xn, FPR64)>; +def : Pat<(i64 (bitconvert (f64 FPR64:$Xn))), + (COPY_TO_REGCLASS FPR64:$Xn, GPR64)>; +def : Pat<(i64 (bitconvert (v1f64 V64:$Vn))), + (COPY_TO_REGCLASS V64:$Vn, GPR64)>; + +let Predicates = [IsLE] in { +def : Pat<(v1i64 (bitconvert (v2i32 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1i64 (bitconvert (v4i16 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1i64 (bitconvert (v8i8 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1i64 (bitconvert (v4f16 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1i64 (bitconvert (v2f32 FPR64:$src))), (v1i64 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v1i64 (bitconvert (v2i32 FPR64:$src))), + (v1i64 (REV64v2i32 FPR64:$src))>; +def : Pat<(v1i64 (bitconvert (v4i16 FPR64:$src))), + (v1i64 (REV64v4i16 FPR64:$src))>; +def : Pat<(v1i64 (bitconvert (v8i8 FPR64:$src))), + (v1i64 (REV64v8i8 FPR64:$src))>; +def : Pat<(v1i64 (bitconvert (v4f16 FPR64:$src))), + (v1i64 (REV64v4i16 FPR64:$src))>; +def : Pat<(v1i64 (bitconvert (v2f32 FPR64:$src))), + (v1i64 (REV64v2i32 FPR64:$src))>; +} +def : Pat<(v1i64 (bitconvert (v1f64 FPR64:$src))), (v1i64 FPR64:$src)>; +def : Pat<(v1i64 (bitconvert (f64 FPR64:$src))), (v1i64 FPR64:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v2i32 (bitconvert (v1i64 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2i32 (bitconvert (v4i16 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2i32 (bitconvert (v8i8 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2i32 (bitconvert (f64 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2i32 (bitconvert (v1f64 FPR64:$src))), (v2i32 FPR64:$src)>; +def : Pat<(v2i32 (bitconvert (v4f16 FPR64:$src))), (v2i32 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v2i32 (bitconvert (v1i64 FPR64:$src))), + (v2i32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2i32 (bitconvert (v4i16 FPR64:$src))), + (v2i32 (REV32v4i16 FPR64:$src))>; +def : Pat<(v2i32 (bitconvert (v8i8 FPR64:$src))), + (v2i32 (REV32v8i8 FPR64:$src))>; +def : Pat<(v2i32 (bitconvert (f64 FPR64:$src))), + (v2i32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2i32 (bitconvert (v1f64 FPR64:$src))), + (v2i32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2i32 (bitconvert (v4f16 FPR64:$src))), + (v2i32 (REV64v4i16 FPR64:$src))>; +} +def : Pat<(v2i32 (bitconvert (v2f32 FPR64:$src))), (v2i32 FPR64:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v4i16 (bitconvert (v1i64 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (v2i32 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (v8i8 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (f64 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (v4f16 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (v2f32 FPR64:$src))), (v4i16 FPR64:$src)>; +def : Pat<(v4i16 (bitconvert (v1f64 FPR64:$src))), (v4i16 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v4i16 (bitconvert (v1i64 FPR64:$src))), + (v4i16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (v2i32 FPR64:$src))), + (v4i16 (REV32v4i16 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (v8i8 FPR64:$src))), + (v4i16 (REV16v8i8 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (f64 FPR64:$src))), + (v4i16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (v4f16 FPR64:$src))), + (v4i16 (REV32v4i16 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (v2f32 FPR64:$src))), + (v4i16 (REV32v4i16 FPR64:$src))>; +def : Pat<(v4i16 (bitconvert (v1f64 FPR64:$src))), + (v4i16 (REV64v4i16 FPR64:$src))>; +} + +let Predicates = [IsLE] in { +def : Pat<(v4f16 (bitconvert (v1i64 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (v2i32 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (v4i16 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (v8i8 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (f64 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (v2f32 FPR64:$src))), (v4f16 FPR64:$src)>; +def : Pat<(v4f16 (bitconvert (v1f64 FPR64:$src))), (v4f16 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v4f16 (bitconvert (v1i64 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (v2i32 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (v4i16 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (v8i8 FPR64:$src))), + (v4f16 (REV16v8i8 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (f64 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (v2f32 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +def : Pat<(v4f16 (bitconvert (v1f64 FPR64:$src))), + (v4f16 (REV64v4i16 FPR64:$src))>; +} + + + +let Predicates = [IsLE] in { +def : Pat<(v8i8 (bitconvert (v1i64 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (v2i32 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (v4i16 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (f64 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (v2f32 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (v1f64 FPR64:$src))), (v8i8 FPR64:$src)>; +def : Pat<(v8i8 (bitconvert (v4f16 FPR64:$src))), (v8i8 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v8i8 (bitconvert (v1i64 FPR64:$src))), + (v8i8 (REV64v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (v2i32 FPR64:$src))), + (v8i8 (REV32v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (v4i16 FPR64:$src))), + (v8i8 (REV16v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (f64 FPR64:$src))), + (v8i8 (REV64v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (v2f32 FPR64:$src))), + (v8i8 (REV32v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (v1f64 FPR64:$src))), + (v8i8 (REV64v8i8 FPR64:$src))>; +def : Pat<(v8i8 (bitconvert (v4f16 FPR64:$src))), + (v8i8 (REV16v8i8 FPR64:$src))>; +} + +let Predicates = [IsLE] in { +def : Pat<(f64 (bitconvert (v2i32 FPR64:$src))), (f64 FPR64:$src)>; +def : Pat<(f64 (bitconvert (v4i16 FPR64:$src))), (f64 FPR64:$src)>; +def : Pat<(f64 (bitconvert (v2f32 FPR64:$src))), (f64 FPR64:$src)>; +def : Pat<(f64 (bitconvert (v8i8 FPR64:$src))), (f64 FPR64:$src)>; +def : Pat<(f64 (bitconvert (v4f16 FPR64:$src))), (f64 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(f64 (bitconvert (v2i32 FPR64:$src))), + (f64 (REV64v2i32 FPR64:$src))>; +def : Pat<(f64 (bitconvert (v4i16 FPR64:$src))), + (f64 (REV64v4i16 FPR64:$src))>; +def : Pat<(f64 (bitconvert (v2f32 FPR64:$src))), + (f64 (REV64v2i32 FPR64:$src))>; +def : Pat<(f64 (bitconvert (v8i8 FPR64:$src))), + (f64 (REV64v8i8 FPR64:$src))>; +def : Pat<(f64 (bitconvert (v4f16 FPR64:$src))), + (f64 (REV64v4i16 FPR64:$src))>; +} +def : Pat<(f64 (bitconvert (v1i64 FPR64:$src))), (f64 FPR64:$src)>; +def : Pat<(f64 (bitconvert (v1f64 FPR64:$src))), (f64 FPR64:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v1f64 (bitconvert (v2i32 FPR64:$src))), (v1f64 FPR64:$src)>; +def : Pat<(v1f64 (bitconvert (v4i16 FPR64:$src))), (v1f64 FPR64:$src)>; +def : Pat<(v1f64 (bitconvert (v8i8 FPR64:$src))), (v1f64 FPR64:$src)>; +def : Pat<(v1f64 (bitconvert (v2f32 FPR64:$src))), (v1f64 FPR64:$src)>; +def : Pat<(v1f64 (bitconvert (v4f16 FPR64:$src))), (v1f64 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v1f64 (bitconvert (v2i32 FPR64:$src))), + (v1f64 (REV64v2i32 FPR64:$src))>; +def : Pat<(v1f64 (bitconvert (v4i16 FPR64:$src))), + (v1f64 (REV64v4i16 FPR64:$src))>; +def : Pat<(v1f64 (bitconvert (v8i8 FPR64:$src))), + (v1f64 (REV64v8i8 FPR64:$src))>; +def : Pat<(v1f64 (bitconvert (v2f32 FPR64:$src))), + (v1f64 (REV64v2i32 FPR64:$src))>; +def : Pat<(v1f64 (bitconvert (v4f16 FPR64:$src))), + (v1f64 (REV64v4i16 FPR64:$src))>; +} +def : Pat<(v1f64 (bitconvert (v1i64 FPR64:$src))), (v1f64 FPR64:$src)>; +def : Pat<(v1f64 (bitconvert (f64 FPR64:$src))), (v1f64 FPR64:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v2f32 (bitconvert (v1i64 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v2f32 (bitconvert (v4i16 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v2f32 (bitconvert (v8i8 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v2f32 (bitconvert (v1f64 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v2f32 (bitconvert (f64 FPR64:$src))), (v2f32 FPR64:$src)>; +def : Pat<(v2f32 (bitconvert (v4f16 FPR64:$src))), (v2f32 FPR64:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v2f32 (bitconvert (v1i64 FPR64:$src))), + (v2f32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2f32 (bitconvert (v4i16 FPR64:$src))), + (v2f32 (REV32v4i16 FPR64:$src))>; +def : Pat<(v2f32 (bitconvert (v8i8 FPR64:$src))), + (v2f32 (REV32v8i8 FPR64:$src))>; +def : Pat<(v2f32 (bitconvert (v1f64 FPR64:$src))), + (v2f32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2f32 (bitconvert (f64 FPR64:$src))), + (v2f32 (REV64v2i32 FPR64:$src))>; +def : Pat<(v2f32 (bitconvert (v4f16 FPR64:$src))), + (v2f32 (REV64v4i16 FPR64:$src))>; +} +def : Pat<(v2f32 (bitconvert (v2i32 FPR64:$src))), (v2f32 FPR64:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(f128 (bitconvert (v2i64 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v4i32 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v8i16 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v2f64 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v4f32 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v8f16 FPR128:$src))), (f128 FPR128:$src)>; +def : Pat<(f128 (bitconvert (v16i8 FPR128:$src))), (f128 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(f128 (bitconvert (v2i64 FPR128:$src))), + (f128 (EXTv16i8 FPR128:$src, FPR128:$src, (i32 8)))>; +def : Pat<(f128 (bitconvert (v4i32 FPR128:$src))), + (f128 (EXTv16i8 (REV64v4i32 FPR128:$src), + (REV64v4i32 FPR128:$src), (i32 8)))>; +def : Pat<(f128 (bitconvert (v8i16 FPR128:$src))), + (f128 (EXTv16i8 (REV64v8i16 FPR128:$src), + (REV64v8i16 FPR128:$src), (i32 8)))>; +def : Pat<(f128 (bitconvert (v8f16 FPR128:$src))), + (f128 (EXTv16i8 (REV64v8i16 FPR128:$src), + (REV64v8i16 FPR128:$src), (i32 8)))>; +def : Pat<(f128 (bitconvert (v2f64 FPR128:$src))), + (f128 (EXTv16i8 FPR128:$src, FPR128:$src, (i32 8)))>; +def : Pat<(f128 (bitconvert (v4f32 FPR128:$src))), + (f128 (EXTv16i8 (REV64v4i32 FPR128:$src), + (REV64v4i32 FPR128:$src), (i32 8)))>; +def : Pat<(f128 (bitconvert (v16i8 FPR128:$src))), + (f128 (EXTv16i8 (REV64v16i8 FPR128:$src), + (REV64v16i8 FPR128:$src), (i32 8)))>; +} + +let Predicates = [IsLE] in { +def : Pat<(v2f64 (bitconvert (f128 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v2f64 (bitconvert (v4i32 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v2f64 (bitconvert (v8i16 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v2f64 (bitconvert (v8f16 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v2f64 (bitconvert (v16i8 FPR128:$src))), (v2f64 FPR128:$src)>; +def : Pat<(v2f64 (bitconvert (v4f32 FPR128:$src))), (v2f64 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v2f64 (bitconvert (f128 FPR128:$src))), + (v2f64 (EXTv16i8 FPR128:$src, + FPR128:$src, (i32 8)))>; +def : Pat<(v2f64 (bitconvert (v4i32 FPR128:$src))), + (v2f64 (REV64v4i32 FPR128:$src))>; +def : Pat<(v2f64 (bitconvert (v8i16 FPR128:$src))), + (v2f64 (REV64v8i16 FPR128:$src))>; +def : Pat<(v2f64 (bitconvert (v8f16 FPR128:$src))), + (v2f64 (REV64v8i16 FPR128:$src))>; +def : Pat<(v2f64 (bitconvert (v16i8 FPR128:$src))), + (v2f64 (REV64v16i8 FPR128:$src))>; +def : Pat<(v2f64 (bitconvert (v4f32 FPR128:$src))), + (v2f64 (REV64v4i32 FPR128:$src))>; +} +def : Pat<(v2f64 (bitconvert (v2i64 FPR128:$src))), (v2f64 FPR128:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v4f32 (bitconvert (f128 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v4f32 (bitconvert (v8i16 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v4f32 (bitconvert (v8f16 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v4f32 (bitconvert (v16i8 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v4f32 (bitconvert (v2i64 FPR128:$src))), (v4f32 FPR128:$src)>; +def : Pat<(v4f32 (bitconvert (v2f64 FPR128:$src))), (v4f32 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v4f32 (bitconvert (f128 FPR128:$src))), + (v4f32 (EXTv16i8 (REV64v4i32 FPR128:$src), + (REV64v4i32 FPR128:$src), (i32 8)))>; +def : Pat<(v4f32 (bitconvert (v8i16 FPR128:$src))), + (v4f32 (REV32v8i16 FPR128:$src))>; +def : Pat<(v4f32 (bitconvert (v8f16 FPR128:$src))), + (v4f32 (REV32v8i16 FPR128:$src))>; +def : Pat<(v4f32 (bitconvert (v16i8 FPR128:$src))), + (v4f32 (REV32v16i8 FPR128:$src))>; +def : Pat<(v4f32 (bitconvert (v2i64 FPR128:$src))), + (v4f32 (REV64v4i32 FPR128:$src))>; +def : Pat<(v4f32 (bitconvert (v2f64 FPR128:$src))), + (v4f32 (REV64v4i32 FPR128:$src))>; +} +def : Pat<(v4f32 (bitconvert (v4i32 FPR128:$src))), (v4f32 FPR128:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v2i64 (bitconvert (f128 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2i64 (bitconvert (v4i32 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2i64 (bitconvert (v8i16 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2i64 (bitconvert (v16i8 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2i64 (bitconvert (v4f32 FPR128:$src))), (v2i64 FPR128:$src)>; +def : Pat<(v2i64 (bitconvert (v8f16 FPR128:$src))), (v2i64 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v2i64 (bitconvert (f128 FPR128:$src))), + (v2i64 (EXTv16i8 FPR128:$src, + FPR128:$src, (i32 8)))>; +def : Pat<(v2i64 (bitconvert (v4i32 FPR128:$src))), + (v2i64 (REV64v4i32 FPR128:$src))>; +def : Pat<(v2i64 (bitconvert (v8i16 FPR128:$src))), + (v2i64 (REV64v8i16 FPR128:$src))>; +def : Pat<(v2i64 (bitconvert (v16i8 FPR128:$src))), + (v2i64 (REV64v16i8 FPR128:$src))>; +def : Pat<(v2i64 (bitconvert (v4f32 FPR128:$src))), + (v2i64 (REV64v4i32 FPR128:$src))>; +def : Pat<(v2i64 (bitconvert (v8f16 FPR128:$src))), + (v2i64 (REV64v8i16 FPR128:$src))>; +} +def : Pat<(v2i64 (bitconvert (v2f64 FPR128:$src))), (v2i64 FPR128:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v4i32 (bitconvert (f128 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4i32 (bitconvert (v2i64 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4i32 (bitconvert (v8i16 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4i32 (bitconvert (v16i8 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4i32 (bitconvert (v2f64 FPR128:$src))), (v4i32 FPR128:$src)>; +def : Pat<(v4i32 (bitconvert (v8f16 FPR128:$src))), (v4i32 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v4i32 (bitconvert (f128 FPR128:$src))), + (v4i32 (EXTv16i8 (REV64v4i32 FPR128:$src), + (REV64v4i32 FPR128:$src), + (i32 8)))>; +def : Pat<(v4i32 (bitconvert (v2i64 FPR128:$src))), + (v4i32 (REV64v4i32 FPR128:$src))>; +def : Pat<(v4i32 (bitconvert (v8i16 FPR128:$src))), + (v4i32 (REV32v8i16 FPR128:$src))>; +def : Pat<(v4i32 (bitconvert (v16i8 FPR128:$src))), + (v4i32 (REV32v16i8 FPR128:$src))>; +def : Pat<(v4i32 (bitconvert (v2f64 FPR128:$src))), + (v4i32 (REV64v4i32 FPR128:$src))>; +def : Pat<(v4i32 (bitconvert (v8f16 FPR128:$src))), + (v4i32 (REV32v8i16 FPR128:$src))>; +} +def : Pat<(v4i32 (bitconvert (v4f32 FPR128:$src))), (v4i32 FPR128:$src)>; + +let Predicates = [IsLE] in { +def : Pat<(v8i16 (bitconvert (f128 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v2i64 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v4i32 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v16i8 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v2f64 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v4f32 FPR128:$src))), (v8i16 FPR128:$src)>; +def : Pat<(v8i16 (bitconvert (v8f16 FPR128:$src))), (v8i16 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v8i16 (bitconvert (f128 FPR128:$src))), + (v8i16 (EXTv16i8 (REV64v8i16 FPR128:$src), + (REV64v8i16 FPR128:$src), + (i32 8)))>; +def : Pat<(v8i16 (bitconvert (v2i64 FPR128:$src))), + (v8i16 (REV64v8i16 FPR128:$src))>; +def : Pat<(v8i16 (bitconvert (v4i32 FPR128:$src))), + (v8i16 (REV32v8i16 FPR128:$src))>; +def : Pat<(v8i16 (bitconvert (v16i8 FPR128:$src))), + (v8i16 (REV16v16i8 FPR128:$src))>; +def : Pat<(v8i16 (bitconvert (v2f64 FPR128:$src))), + (v8i16 (REV64v8i16 FPR128:$src))>; +def : Pat<(v8i16 (bitconvert (v4f32 FPR128:$src))), + (v8i16 (REV32v8i16 FPR128:$src))>; +def : Pat<(v8i16 (bitconvert (v8f16 FPR128:$src))), + (v8i16 (REV32v8i16 FPR128:$src))>; +} + +let Predicates = [IsLE] in { +def : Pat<(v8f16 (bitconvert (f128 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v2i64 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v4i32 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v8i16 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v16i8 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v2f64 FPR128:$src))), (v8f16 FPR128:$src)>; +def : Pat<(v8f16 (bitconvert (v4f32 FPR128:$src))), (v8f16 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v8f16 (bitconvert (f128 FPR128:$src))), + (v8f16 (EXTv16i8 (REV64v8i16 FPR128:$src), + (REV64v8i16 FPR128:$src), + (i32 8)))>; +def : Pat<(v8f16 (bitconvert (v2i64 FPR128:$src))), + (v8f16 (REV64v8i16 FPR128:$src))>; +def : Pat<(v8f16 (bitconvert (v4i32 FPR128:$src))), + (v8f16 (REV32v8i16 FPR128:$src))>; +def : Pat<(v8f16 (bitconvert (v8i16 FPR128:$src))), + (v8f16 (REV64v8i16 FPR128:$src))>; +def : Pat<(v8f16 (bitconvert (v16i8 FPR128:$src))), + (v8f16 (REV16v16i8 FPR128:$src))>; +def : Pat<(v8f16 (bitconvert (v2f64 FPR128:$src))), + (v8f16 (REV64v8i16 FPR128:$src))>; +def : Pat<(v8f16 (bitconvert (v4f32 FPR128:$src))), + (v8f16 (REV32v8i16 FPR128:$src))>; +} + +let Predicates = [IsLE] in { +def : Pat<(v16i8 (bitconvert (f128 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v2i64 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v4i32 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v8i16 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v2f64 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v4f32 FPR128:$src))), (v16i8 FPR128:$src)>; +def : Pat<(v16i8 (bitconvert (v8f16 FPR128:$src))), (v16i8 FPR128:$src)>; +} +let Predicates = [IsBE] in { +def : Pat<(v16i8 (bitconvert (f128 FPR128:$src))), + (v16i8 (EXTv16i8 (REV64v16i8 FPR128:$src), + (REV64v16i8 FPR128:$src), + (i32 8)))>; +def : Pat<(v16i8 (bitconvert (v2i64 FPR128:$src))), + (v16i8 (REV64v16i8 FPR128:$src))>; +def : Pat<(v16i8 (bitconvert (v4i32 FPR128:$src))), + (v16i8 (REV32v16i8 FPR128:$src))>; +def : Pat<(v16i8 (bitconvert (v8i16 FPR128:$src))), + (v16i8 (REV16v16i8 FPR128:$src))>; +def : Pat<(v16i8 (bitconvert (v2f64 FPR128:$src))), + (v16i8 (REV64v16i8 FPR128:$src))>; +def : Pat<(v16i8 (bitconvert (v4f32 FPR128:$src))), + (v16i8 (REV32v16i8 FPR128:$src))>; +def : Pat<(v16i8 (bitconvert (v8f16 FPR128:$src))), + (v16i8 (REV16v16i8 FPR128:$src))>; +} + +def : Pat<(v4i16 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v8i8 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v2f32 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v4f16 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v2i32 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v1i64 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; +def : Pat<(v1f64 (extract_subvector V128:$Rn, (i64 0))), + (EXTRACT_SUBREG V128:$Rn, dsub)>; + +def : Pat<(v8i8 (extract_subvector (v16i8 FPR128:$Rn), (i64 1))), + (EXTRACT_SUBREG (DUPv2i64lane FPR128:$Rn, 1), dsub)>; +def : Pat<(v4i16 (extract_subvector (v8i16 FPR128:$Rn), (i64 1))), + (EXTRACT_SUBREG (DUPv2i64lane FPR128:$Rn, 1), dsub)>; +def : Pat<(v2i32 (extract_subvector (v4i32 FPR128:$Rn), (i64 1))), + (EXTRACT_SUBREG (DUPv2i64lane FPR128:$Rn, 1), dsub)>; +def : Pat<(v1i64 (extract_subvector (v2i64 FPR128:$Rn), (i64 1))), + (EXTRACT_SUBREG (DUPv2i64lane FPR128:$Rn, 1), dsub)>; + +// A 64-bit subvector insert to the first 128-bit vector position +// is a subregister copy that needs no instruction. +def : Pat<(insert_subvector undef, (v1i64 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v2i64 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v1f64 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v2i32 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v4i32 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v2f32 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v4i16 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v8i16 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v4f16 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v8f16 (IMPLICIT_DEF)), FPR64:$src, dsub)>; +def : Pat<(insert_subvector undef, (v8i8 FPR64:$src), (i32 0)), + (INSERT_SUBREG (v16i8 (IMPLICIT_DEF)), FPR64:$src, dsub)>; + +// Use pair-wise add instructions when summing up the lanes for v2f64, v2i64 +// or v2f32. +def : Pat<(i64 (add (vector_extract (v2i64 FPR128:$Rn), (i64 0)), + (vector_extract (v2i64 FPR128:$Rn), (i64 1)))), + (i64 (ADDPv2i64p (v2i64 FPR128:$Rn)))>; +def : Pat<(f64 (fadd (vector_extract (v2f64 FPR128:$Rn), (i64 0)), + (vector_extract (v2f64 FPR128:$Rn), (i64 1)))), + (f64 (FADDPv2i64p (v2f64 FPR128:$Rn)))>; + // vector_extract on 64-bit vectors gets promoted to a 128 bit vector, + // so we match on v4f32 here, not v2f32. This will also catch adding + // the low two lanes of a true v4f32 vector. +def : Pat<(fadd (vector_extract (v4f32 FPR128:$Rn), (i64 0)), + (vector_extract (v4f32 FPR128:$Rn), (i64 1))), + (f32 (FADDPv2i32p (EXTRACT_SUBREG FPR128:$Rn, dsub)))>; + +// Scalar 64-bit shifts in FPR64 registers. +def : Pat<(i64 (int_aarch64_neon_sshl (i64 FPR64:$Rn), (i64 FPR64:$Rm))), + (SSHLv1i64 FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(i64 (int_aarch64_neon_ushl (i64 FPR64:$Rn), (i64 FPR64:$Rm))), + (USHLv1i64 FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(i64 (int_aarch64_neon_srshl (i64 FPR64:$Rn), (i64 FPR64:$Rm))), + (SRSHLv1i64 FPR64:$Rn, FPR64:$Rm)>; +def : Pat<(i64 (int_aarch64_neon_urshl (i64 FPR64:$Rn), (i64 FPR64:$Rm))), + (URSHLv1i64 FPR64:$Rn, FPR64:$Rm)>; + +// Patterns for nontemporal/no-allocate stores. +// We have to resort to tricks to turn a single-input store into a store pair, +// because there is no single-input nontemporal store, only STNP. +let Predicates = [IsLE] in { +let AddedComplexity = 15 in { +class NTStore128Pat : + Pat<(nontemporalstore (VT FPR128:$Rt), + (am_indexed7s64 GPR64sp:$Rn, simm7s8:$offset)), + (STNPDi (EXTRACT_SUBREG FPR128:$Rt, dsub), + (CPYi64 FPR128:$Rt, (i64 1)), + GPR64sp:$Rn, simm7s8:$offset)>; + +def : NTStore128Pat; +def : NTStore128Pat; +def : NTStore128Pat; +def : NTStore128Pat; + +class NTStore64Pat : + Pat<(nontemporalstore (VT FPR64:$Rt), + (am_indexed7s32 GPR64sp:$Rn, simm7s4:$offset)), + (STNPSi (EXTRACT_SUBREG FPR64:$Rt, ssub), + (CPYi32 (SUBREG_TO_REG (i64 0), FPR64:$Rt, dsub), (i64 1)), + GPR64sp:$Rn, simm7s4:$offset)>; + +// FIXME: Shouldn't v1f64 loads/stores be promoted to v1i64? +def : NTStore64Pat; +def : NTStore64Pat; +def : NTStore64Pat; +def : NTStore64Pat; +def : NTStore64Pat; + +def : Pat<(nontemporalstore GPR64:$Rt, + (am_indexed7s32 GPR64sp:$Rn, simm7s4:$offset)), + (STNPWi (EXTRACT_SUBREG GPR64:$Rt, sub_32), + (EXTRACT_SUBREG (UBFMXri GPR64:$Rt, 0, 31), sub_32), + GPR64sp:$Rn, simm7s4:$offset)>; +} // AddedComplexity=10 +} // Predicates = [IsLE] + +// Tail call return handling. These are all compiler pseudo-instructions, +// so no encoding information or anything like that. +let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1, Uses = [SP] in { + def TCRETURNdi : Pseudo<(outs), (ins i64imm:$dst, i32imm:$FPDiff),[]>; + def TCRETURNri : Pseudo<(outs), (ins tcGPR64:$dst, i32imm:$FPDiff), []>; +} + +def : Pat<(AArch64tcret tcGPR64:$dst, (i32 timm:$FPDiff)), + (TCRETURNri tcGPR64:$dst, imm:$FPDiff)>; +def : Pat<(AArch64tcret tglobaladdr:$dst, (i32 timm:$FPDiff)), + (TCRETURNdi texternalsym:$dst, imm:$FPDiff)>; def : Pat<(AArch64tcret texternalsym:$dst, (i32 timm:$FPDiff)), - (TC_RETURNdi texternalsym:$dst, imm:$FPDiff)>; - -//===----------------------------------------------------------------------===// -// Thread local storage -//===----------------------------------------------------------------------===// - -// This is a pseudo-instruction representing the ".tlsdesccall" directive in -// assembly. Its effect is to insert an R_AARCH64_TLSDESC_CALL relocation at the -// current location. It should always be immediately followed by a BLR -// instruction, and is intended solely for relaxation by the linker. - -def : Pat<(A64threadpointer), (MRSxi 0xde82)>; - -def TLSDESCCALL : PseudoInst<(outs), (ins i64imm:$Lbl), []> { - let hasSideEffects = 1; -} - -def TLSDESC_BLRx : PseudoInst<(outs), (ins GPR64:$Rn, i64imm:$Var), - [(A64tlsdesc_blr GPR64:$Rn, tglobaltlsaddr:$Var)]> { - let isCall = 1; - let Defs = [X30]; -} - -def : Pat<(A64tlsdesc_blr GPR64:$Rn, texternalsym:$Var), - (TLSDESC_BLRx GPR64:$Rn, texternalsym:$Var)>; - -//===----------------------------------------------------------------------===// -// Bitfield patterns -//===----------------------------------------------------------------------===// - -def bfi32_lsb_to_immr : SDNodeXFormgetTargetConstant((32 - N->getZExtValue()) % 32, MVT::i64); -}]>; - -def bfi64_lsb_to_immr : SDNodeXFormgetTargetConstant((64 - N->getZExtValue()) % 64, MVT::i64); -}]>; - -def bfi_width_to_imms : SDNodeXFormgetTargetConstant(N->getZExtValue() - 1, MVT::i64); -}]>; - - -// The simpler patterns deal with cases where no AND mask is actually needed -// (either all bits are used or the low 32 bits are used). -let AddedComplexity = 10 in { - -def : Pat<(A64Bfi GPR64:$src, GPR64:$Rn, imm:$ImmR, imm:$ImmS), - (BFIxxii GPR64:$src, GPR64:$Rn, - (bfi64_lsb_to_immr (i64 imm:$ImmR)), - (bfi_width_to_imms (i64 imm:$ImmS)))>; - -def : Pat<(A64Bfi GPR32:$src, GPR32:$Rn, imm:$ImmR, imm:$ImmS), - (BFIwwii GPR32:$src, GPR32:$Rn, - (bfi32_lsb_to_immr (i64 imm:$ImmR)), - (bfi_width_to_imms (i64 imm:$ImmS)))>; - - -def : Pat<(and (A64Bfi GPR64:$src, GPR64:$Rn, imm:$ImmR, imm:$ImmS), - (i64 4294967295)), - (SUBREG_TO_REG (i64 0), - (BFIwwii (EXTRACT_SUBREG GPR64:$src, sub_32), - (EXTRACT_SUBREG GPR64:$Rn, sub_32), - (bfi32_lsb_to_immr (i64 imm:$ImmR)), - (bfi_width_to_imms (i64 imm:$ImmS))), - sub_32)>; - -} - -//===----------------------------------------------------------------------===// -// Constant island entries -//===----------------------------------------------------------------------===// - -// The constant island pass needs to create "instructions" in the middle of the -// instruction stream to reresent its constants. - -def cpinst_operand : Operand; - -def CONSTPOOL_ENTRY : PseudoInst<(outs), (ins cpinst_operand:$instid, - cpinst_operand:$cpidx, - i32imm:$size), []> { - let hasSideEffects = 0; - let isNotDuplicable = 1; -} - -//===----------------------------------------------------------------------===// -// Miscellaneous patterns -//===----------------------------------------------------------------------===// - -// Truncation from 64 to 32-bits just involves renaming your register. -def : Pat<(i32 (trunc (i64 GPR64:$val))), (EXTRACT_SUBREG GPR64:$val, sub_32)>; - -// Similarly, extension where we don't care about the high bits is -// just a rename. -def : Pat<(i64 (anyext (i32 GPR32:$val))), - (INSERT_SUBREG (IMPLICIT_DEF), GPR32:$val, sub_32)>; - -// SELECT instructions providing f128 types need to be handled by a -// pseudo-instruction since the eventual code will need to introduce basic -// blocks and control flow. -def F128CSEL : PseudoInst<(outs FPR128:$Rd), - (ins FPR128:$Rn, FPR128:$Rm, cond_code_op:$Cond), - [(set FPR128:$Rd, (simple_select (f128 FPR128:$Rn), - FPR128:$Rm))]> { - let Uses = [NZCV]; - let usesCustomInserter = 1; -} - -//===----------------------------------------------------------------------===// -// Load/store patterns -//===----------------------------------------------------------------------===// - -// There are lots of patterns here, because we need to allow at least three -// parameters to vary independently. -// 1. Instruction: "ldrb w9, [sp]", "ldrh w9, [sp]", ... -// 2. LLVM source: zextloadi8, anyextloadi8, ... -// 3. Address-generation: A64Wrapper, (add BASE, OFFSET), ... -// -// The biggest problem turns out to be the address-generation variable. At the -// point of instantiation we need to produce two DAGs, one for the pattern and -// one for the instruction. Doing this at the lowest level of classes doesn't -// work. -// -// Consider the simple uimm12 addressing mode, and the desire to match both (add -// GPR64xsp:$Rn, uimm12:$Offset) and GPR64xsp:$Rn, particularly on the -// instruction side. We'd need to insert either "GPR64xsp" and "uimm12" or -// "GPR64xsp" and "0" into an unknown dag. !subst is not capable of this -// operation, and PatFrags are for selection not output. -// -// As a result, the address-generation patterns are the final -// instantiations. However, we do still need to vary the operand for the address -// further down (At the point we're deciding A64WrapperSmall, we don't know -// the memory width of the operation). - -//===------------------------------ -// 1. Basic infrastructural defs -//===------------------------------ - -// First, some simple classes for !foreach and !subst to use: -class Decls { - dag pattern; -} - -def decls : Decls; -def ALIGN; -def INST; -def OFFSET; -def SHIFT; - -// You can't use !subst on an actual immediate, but you *can* use it on an -// operand record that happens to match a single immediate. So we do. -def imm_eq0 : ImmLeaf; -def imm_eq1 : ImmLeaf; -def imm_eq2 : ImmLeaf; -def imm_eq3 : ImmLeaf; -def imm_eq4 : ImmLeaf; - -// If the low bits of a pointer are known to be 0 then an "or" is just as good -// as addition for computing an offset. This fragment forwards that check for -// TableGen's use. -def add_like_or : PatFrag<(ops node:$lhs, node:$rhs), (or node:$lhs, node:$rhs), -[{ - return CurDAG->isBaseWithConstantOffset(SDValue(N, 0)); -}]>; - -// Load/store (unsigned immediate) operations with relocations against global -// symbols (for lo12) are only valid if those symbols have correct alignment -// (since the immediate offset is divided by the access scale, it can't have a -// remainder). -// -// The guaranteed alignment is provided as part of the WrapperSmall -// operation, and checked against one of these. -def any_align : ImmLeaf; -def min_align2 : ImmLeaf= 2; }]>; -def min_align4 : ImmLeaf= 4; }]>; -def min_align8 : ImmLeaf= 8; }]>; -def min_align16 : ImmLeaf= 16; }]>; - -// "Normal" load/store instructions can be used on atomic operations, provided -// the ordering parameter is at most "monotonic". Anything above that needs -// special handling with acquire/release instructions. -class simple_load - : PatFrag<(ops node:$ptr), (base node:$ptr), [{ - return cast(N)->getOrdering() <= Monotonic; -}]>; - -def atomic_load_simple_i8 : simple_load; -def atomic_load_simple_i16 : simple_load; -def atomic_load_simple_i32 : simple_load; -def atomic_load_simple_i64 : simple_load; - -class simple_store - : PatFrag<(ops node:$ptr, node:$val), (base node:$ptr, node:$val), [{ - return cast(N)->getOrdering() <= Monotonic; -}]>; - -def atomic_store_simple_i8 : simple_store; -def atomic_store_simple_i16 : simple_store; -def atomic_store_simple_i32 : simple_store; -def atomic_store_simple_i64 : simple_store; - -//===------------------------------ -// 2. UImm12 and SImm9 -//===------------------------------ - -// These instructions have two operands providing the address so they can be -// treated similarly for most purposes. - -//===------------------------------ -// 2.1 Base patterns covering extend/truncate semantics -//===------------------------------ - -// Atomic patterns can be shared between integer operations of all sizes, a -// quick multiclass here allows reuse. -multiclass ls_atomic_pats { - def : Pat<(!cast("atomic_load_simple_" # sty) address), - (LOAD Base, Offset)>; - - def : Pat<(!cast("atomic_store_simple_" # sty) address, TPR:$Rt), - (STORE TPR:$Rt, Base, Offset)>; -} - -// Instructions accessing a memory chunk smaller than a register (or, in a -// pinch, the same size) have a characteristic set of patterns they want to -// match: extending loads and truncating stores. This class deals with the -// sign-neutral version of those patterns. -// -// It will be instantiated across multiple addressing-modes. -multiclass ls_small_pats - : ls_atomic_pats { - def : Pat<(!cast(zextload # sty) address), (LOAD Base, Offset)>; - - def : Pat<(!cast(extload # sty) address), (LOAD Base, Offset)>; - - // For zero-extension to 64-bits we have to tell LLVM that the whole 64-bit - // register was actually set. - def : Pat<(i64 (!cast(zextload # sty) address)), - (SUBREG_TO_REG (i64 0), (LOAD Base, Offset), sub_32)>; - - def : Pat<(i64 (!cast(extload # sty) address)), - (SUBREG_TO_REG (i64 0), (LOAD Base, Offset), sub_32)>; - - def : Pat<(!cast(truncstore # sty) GPR32:$Rt, address), - (STORE GPR32:$Rt, Base, Offset)>; - - // For truncating store from 64-bits, we have to manually tell LLVM to - // ignore the high bits of the x register. - def : Pat<(!cast(truncstore # sty) GPR64:$Rt, address), - (STORE (EXTRACT_SUBREG GPR64:$Rt, sub_32), Base, Offset)>; -} - -// Next come patterns for sign-extending loads. -multiclass load_signed_pats { - def : Pat<(i32 (!cast("sextload" # sty) address)), - (!cast("LDRS" # T # "w" # U) Base, Offset)>; - - def : Pat<(i64 (!cast("sextload" # sty) address)), - (!cast("LDRS" # T # "x" # U) Base, Offset)>; - -} - -// and finally "natural-width" loads and stores come next. -multiclass ls_neutral_pats { - def : Pat<(sty (load address)), (LOAD Base, Offset)>; - def : Pat<(store (sty TPR:$Rt), address), (STORE TPR:$Rt, Base, Offset)>; -} - -// Integer operations also get atomic instructions to select for. -multiclass ls_int_neutral_pats - : ls_neutral_pats, - ls_atomic_pats; - -//===------------------------------ -// 2.2. Addressing-mode instantiations -//===------------------------------ - -multiclass uimm12_pats { - defm : ls_small_pats; - defm : ls_small_pats; - defm : ls_small_pats; - - defm : ls_int_neutral_pats; - - defm : ls_int_neutral_pats; - - defm : ls_neutral_pats; - - defm : ls_neutral_pats; - - defm : ls_neutral_pats; - - defm : ls_neutral_pats; - - defm : load_signed_pats<"B", "", Base, - !foreach(decls.pattern, Offset, - !subst(OFFSET, byte_uimm12, decls.pattern)), - !foreach(decls.pattern, address, - !subst(OFFSET, byte_uimm12, - !subst(ALIGN, any_align, decls.pattern))), - i8>; - - defm : load_signed_pats<"H", "", Base, - !foreach(decls.pattern, Offset, - !subst(OFFSET, hword_uimm12, decls.pattern)), - !foreach(decls.pattern, address, - !subst(OFFSET, hword_uimm12, - !subst(ALIGN, min_align2, decls.pattern))), - i16>; - - def : Pat<(sextloadi32 !foreach(decls.pattern, address, - !subst(OFFSET, word_uimm12, - !subst(ALIGN, min_align4, decls.pattern)))), - (LDRSWx Base, !foreach(decls.pattern, Offset, - !subst(OFFSET, word_uimm12, decls.pattern)))>; -} - -// Straightforward patterns of last resort: a pointer with or without an -// appropriate offset. -defm : uimm12_pats<(i64 GPR64xsp:$Rn), (i64 GPR64xsp:$Rn), (i64 0)>; -defm : uimm12_pats<(add GPR64xsp:$Rn, OFFSET:$UImm12), - (i64 GPR64xsp:$Rn), (i64 OFFSET:$UImm12)>; - -// The offset could be hidden behind an "or", of course: -defm : uimm12_pats<(add_like_or GPR64xsp:$Rn, OFFSET:$UImm12), - (i64 GPR64xsp:$Rn), (i64 OFFSET:$UImm12)>; - -// Global addresses under the small-absolute model should use these -// instructions. There are ELF relocations specifically for it. -defm : uimm12_pats<(A64WrapperSmall tglobaladdr:$Hi, tglobaladdr:$Lo12, ALIGN), - (ADRPxi tglobaladdr:$Hi), (i64 tglobaladdr:$Lo12)>; - -defm : uimm12_pats<(A64WrapperSmall tglobaltlsaddr:$Hi, tglobaltlsaddr:$Lo12, - ALIGN), - (ADRPxi tglobaltlsaddr:$Hi), (i64 tglobaltlsaddr:$Lo12)>; - -// External symbols that make it this far should also get standard relocations. -defm : uimm12_pats<(A64WrapperSmall texternalsym:$Hi, texternalsym:$Lo12, - ALIGN), - (ADRPxi texternalsym:$Hi), (i64 texternalsym:$Lo12)>; - - -// We also want to use uimm12 instructions for local variables at the moment. -def tframeindex_XFORM : SDNodeXForm(N)->getIndex(); - return CurDAG->getTargetFrameIndex(FI, MVT::i64); -}]>; - -defm : uimm12_pats<(i64 frameindex:$Rn), - (tframeindex_XFORM tframeindex:$Rn), (i64 0)>; - -// These can be much simpler than uimm12 because we don't to change the operand -// type (e.g. LDURB and LDURH take the same operands). -multiclass simm9_pats { - defm : ls_small_pats; - defm : ls_small_pats; - - defm : ls_int_neutral_pats; - defm : ls_int_neutral_pats; - - defm : ls_neutral_pats; - defm : ls_neutral_pats; - defm : ls_neutral_pats; - defm : ls_neutral_pats; - - def : Pat<(i64 (zextloadi32 address)), - (SUBREG_TO_REG (i64 0), (LS32_LDUR Base, Offset), sub_32)>; - - def : Pat<(truncstorei32 GPR64:$Rt, address), - (LS32_STUR (EXTRACT_SUBREG GPR64:$Rt, sub_32), Base, Offset)>; - - defm : load_signed_pats<"B", "_U", Base, Offset, address, i8>; - defm : load_signed_pats<"H", "_U", Base, Offset, address, i16>; - def : Pat<(sextloadi32 address), (LDURSWx Base, Offset)>; -} - -defm : simm9_pats<(add GPR64xsp:$Rn, simm9:$SImm9), - (i64 GPR64xsp:$Rn), (SDXF_simm9 simm9:$SImm9)>; - -defm : simm9_pats<(add_like_or GPR64xsp:$Rn, simm9:$SImm9), - (i64 GPR64xsp:$Rn), (SDXF_simm9 simm9:$SImm9)>; - - -//===------------------------------ -// 3. Register offset patterns -//===------------------------------ - -// Atomic patterns can be shared between integer operations of all sizes, a -// quick multiclass here allows reuse. -multiclass ro_atomic_pats { - def : Pat<(!cast("atomic_load_simple_" # sty) address), - (LOAD Base, Offset, Extend)>; - - def : Pat<(!cast("atomic_store_simple_" # sty) address, TPR:$Rt), - (STORE TPR:$Rt, Base, Offset, Extend)>; -} - -// The register offset instructions take three operands giving the instruction, -// and have an annoying split between instructions where Rm is 32-bit and -// 64-bit. So we need a special hierarchy to describe them. Other than that the -// same operations should be supported as for simm9 and uimm12 addressing. - -multiclass ro_small_pats - : ro_atomic_pats { - def : Pat<(!cast(zextload # sty) address), - (LOAD Base, Offset, Extend)>; - - def : Pat<(!cast(extload # sty) address), - (LOAD Base, Offset, Extend)>; - - // For zero-extension to 64-bits we have to tell LLVM that the whole 64-bit - // register was actually set. - def : Pat<(i64 (!cast(zextload # sty) address)), - (SUBREG_TO_REG (i64 0), (LOAD Base, Offset, Extend), sub_32)>; - - def : Pat<(i64 (!cast(extload # sty) address)), - (SUBREG_TO_REG (i64 0), (LOAD Base, Offset, Extend), sub_32)>; - - def : Pat<(!cast(truncstore # sty) GPR32:$Rt, address), - (STORE GPR32:$Rt, Base, Offset, Extend)>; - - // For truncating store from 64-bits, we have to manually tell LLVM to - // ignore the high bits of the x register. - def : Pat<(!cast(truncstore # sty) GPR64:$Rt, address), - (STORE (EXTRACT_SUBREG GPR64:$Rt, sub_32), Base, Offset, Extend)>; - -} - -// Next come patterns for sign-extending loads. -multiclass ro_signed_pats { - def : Pat<(i32 (!cast("sextload" # sty) address)), - (!cast("LDRS" # T # "w_" # Rm # "_RegOffset") - Base, Offset, Extend)>; - - def : Pat<(i64 (!cast("sextload" # sty) address)), - (!cast("LDRS" # T # "x_" # Rm # "_RegOffset") - Base, Offset, Extend)>; -} - -// and finally "natural-width" loads and stores come next. -multiclass ro_neutral_pats { - def : Pat<(sty (load address)), (LOAD Base, Offset, Extend)>; - def : Pat<(store (sty TPR:$Rt), address), - (STORE TPR:$Rt, Base, Offset, Extend)>; -} - -multiclass ro_int_neutral_pats - : ro_neutral_pats, - ro_atomic_pats; - -multiclass regoff_pats { - defm : ro_small_pats("LS8_" # Rm # "_RegOffset_LDR"), - !cast("LS8_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq0, decls.pattern)), - i8>; - defm : ro_small_pats("LS16_" # Rm # "_RegOffset_LDR"), - !cast("LS16_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq1, decls.pattern)), - i16>; - defm : ro_small_pats("LS32_" # Rm # "_RegOffset_LDR"), - !cast("LS32_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq2, decls.pattern)), - i32>; - - defm : ro_int_neutral_pats< - !cast("LS32_" # Rm # "_RegOffset_LDR"), - !cast("LS32_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq2, decls.pattern)), - GPR32, i32>; - - defm : ro_int_neutral_pats< - !cast("LS64_" # Rm # "_RegOffset_LDR"), - !cast("LS64_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq3, decls.pattern)), - GPR64, i64>; - - defm : ro_neutral_pats("LSFP16_" # Rm # "_RegOffset_LDR"), - !cast("LSFP16_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq1, decls.pattern)), - FPR16, f16>; - - defm : ro_neutral_pats("LSFP32_" # Rm # "_RegOffset_LDR"), - !cast("LSFP32_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq2, decls.pattern)), - FPR32, f32>; - - defm : ro_neutral_pats("LSFP64_" # Rm # "_RegOffset_LDR"), - !cast("LSFP64_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq3, decls.pattern)), - FPR64, f64>; - - defm : ro_neutral_pats("LSFP128_" # Rm # "_RegOffset_LDR"), - !cast("LSFP128_" # Rm # "_RegOffset_STR"), - Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq4, decls.pattern)), - FPR128, f128>; - - defm : ro_signed_pats<"B", Rm, Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq0, decls.pattern)), - i8>; - - defm : ro_signed_pats<"H", Rm, Base, Offset, Extend, - !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq1, decls.pattern)), - i16>; - - def : Pat<(sextloadi32 !foreach(decls.pattern, address, - !subst(SHIFT, imm_eq2, decls.pattern))), - (!cast("LDRSWx_" # Rm # "_RegOffset") - Base, Offset, Extend)>; -} - - -// Finally we're in a position to tell LLVM exactly what addresses are reachable -// using register-offset instructions. Essentially a base plus a possibly -// extended, possibly shifted (by access size) offset. - -defm : regoff_pats<"Wm", (add GPR64xsp:$Rn, (sext GPR32:$Rm)), - (i64 GPR64xsp:$Rn), (i32 GPR32:$Rm), (i64 6)>; - -defm : regoff_pats<"Wm", (add GPR64xsp:$Rn, (shl (sext GPR32:$Rm), SHIFT)), - (i64 GPR64xsp:$Rn), (i32 GPR32:$Rm), (i64 7)>; - -defm : regoff_pats<"Wm", (add GPR64xsp:$Rn, (zext GPR32:$Rm)), - (i64 GPR64xsp:$Rn), (i32 GPR32:$Rm), (i64 2)>; - -defm : regoff_pats<"Wm", (add GPR64xsp:$Rn, (shl (zext GPR32:$Rm), SHIFT)), - (i64 GPR64xsp:$Rn), (i32 GPR32:$Rm), (i64 3)>; - -defm : regoff_pats<"Xm", (add GPR64xsp:$Rn, GPR64:$Rm), - (i64 GPR64xsp:$Rn), (i64 GPR64:$Rm), (i64 2)>; + (TCRETURNdi texternalsym:$dst, imm:$FPDiff)>; -defm : regoff_pats<"Xm", (add GPR64xsp:$Rn, (shl GPR64:$Rm, SHIFT)), - (i64 GPR64xsp:$Rn), (i64 GPR64:$Rm), (i64 3)>; +include "AArch64InstrAtomics.td"