X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FMips%2FMipsISelLowering.cpp;h=bee470062564b1a4c1308ca38713a3c1353a6239;hb=7225e27b4cd9a3e33478264caac765bf31e3179a;hp=e2219f257ecd9a1982d89732ec0fce597fdafbc9;hpb=2c2c33a167c82db9abd9b6173c1cdfdaa40c2071;p=oota-llvm.git diff --git a/lib/Target/Mips/MipsISelLowering.cpp b/lib/Target/Mips/MipsISelLowering.cpp index e2219f257ec..bee47006256 100644 --- a/lib/Target/Mips/MipsISelLowering.cpp +++ b/lib/Target/Mips/MipsISelLowering.cpp @@ -20,6 +20,7 @@ #include "MipsTargetMachine.h" #include "MipsTargetObjectFile.h" #include "llvm/ADT/Statistic.h" +#include "llvm/ADT/StringSwitch.h" #include "llvm/CodeGen/CallingConvLower.h" #include "llvm/CodeGen/MachineFrameInfo.h" #include "llvm/CodeGen/MachineFunction.h" @@ -30,11 +31,11 @@ #include "llvm/IR/CallingConv.h" #include "llvm/IR/DerivedTypes.h" #include "llvm/IR/GlobalVariable.h" -#include "llvm/IR/Intrinsics.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/ErrorHandling.h" #include "llvm/Support/raw_ostream.h" +#include using namespace llvm; @@ -44,6 +45,11 @@ static cl::opt LargeGOT("mxgot", cl::Hidden, cl::desc("MIPS: Enable GOT larger than 64k."), cl::init(false)); +static cl::opt +NoZeroDivCheck("mno-check-zero-division", cl::Hidden, + cl::desc("MIPS: Don't trap on integer division by zero."), + cl::init(false)); + static const uint16_t O32IntRegs[4] = { Mips::A0, Mips::A1, Mips::A2, Mips::A3 }; @@ -63,10 +69,10 @@ static const uint16_t Mips64DPRegs[8] = { // For example, if I is 0x003ff800, (Pos, Size) = (11, 11). static bool isShiftedMask(uint64_t I, uint64_t &Pos, uint64_t &Size) { if (!isShiftedMask_64(I)) - return false; + return false; Size = CountPopulation_64(I); - Pos = CountTrailingZeros_64(I); + Pos = countTrailingZeros(I); return true; } @@ -75,72 +81,35 @@ SDValue MipsTargetLowering::getGlobalReg(SelectionDAG &DAG, EVT Ty) const { return DAG.getRegister(FI->getGlobalBaseReg(), Ty); } -static SDValue getTargetNode(SDValue Op, SelectionDAG &DAG, unsigned Flag) { - EVT Ty = Op.getValueType(); +SDValue MipsTargetLowering::getTargetNode(GlobalAddressSDNode *N, EVT Ty, + SelectionDAG &DAG, + unsigned Flag) const { + return DAG.getTargetGlobalAddress(N->getGlobal(), SDLoc(N), Ty, 0, Flag); +} - if (GlobalAddressSDNode *N = dyn_cast(Op)) - return DAG.getTargetGlobalAddress(N->getGlobal(), Op.getDebugLoc(), Ty, 0, - Flag); - if (ExternalSymbolSDNode *N = dyn_cast(Op)) - return DAG.getTargetExternalSymbol(N->getSymbol(), Ty, Flag); - if (BlockAddressSDNode *N = dyn_cast(Op)) - return DAG.getTargetBlockAddress(N->getBlockAddress(), Ty, 0, Flag); - if (JumpTableSDNode *N = dyn_cast(Op)) - return DAG.getTargetJumpTable(N->getIndex(), Ty, Flag); - if (ConstantPoolSDNode *N = dyn_cast(Op)) - return DAG.getTargetConstantPool(N->getConstVal(), Ty, N->getAlignment(), - N->getOffset(), Flag); - - llvm_unreachable("Unexpected node type."); - return SDValue(); +SDValue MipsTargetLowering::getTargetNode(ExternalSymbolSDNode *N, EVT Ty, + SelectionDAG &DAG, + unsigned Flag) const { + return DAG.getTargetExternalSymbol(N->getSymbol(), Ty, Flag); } -static SDValue getAddrNonPIC(SDValue Op, SelectionDAG &DAG) { - DebugLoc DL = Op.getDebugLoc(); - EVT Ty = Op.getValueType(); - SDValue Hi = getTargetNode(Op, DAG, MipsII::MO_ABS_HI); - SDValue Lo = getTargetNode(Op, DAG, MipsII::MO_ABS_LO); - return DAG.getNode(ISD::ADD, DL, Ty, - DAG.getNode(MipsISD::Hi, DL, Ty, Hi), - DAG.getNode(MipsISD::Lo, DL, Ty, Lo)); +SDValue MipsTargetLowering::getTargetNode(BlockAddressSDNode *N, EVT Ty, + SelectionDAG &DAG, + unsigned Flag) const { + return DAG.getTargetBlockAddress(N->getBlockAddress(), Ty, 0, Flag); } -SDValue MipsTargetLowering::getAddrLocal(SDValue Op, SelectionDAG &DAG, - bool HasMips64) const { - DebugLoc DL = Op.getDebugLoc(); - EVT Ty = Op.getValueType(); - unsigned GOTFlag = HasMips64 ? MipsII::MO_GOT_PAGE : MipsII::MO_GOT; - SDValue GOT = DAG.getNode(MipsISD::Wrapper, DL, Ty, getGlobalReg(DAG, Ty), - getTargetNode(Op, DAG, GOTFlag)); - SDValue Load = DAG.getLoad(Ty, DL, DAG.getEntryNode(), GOT, - MachinePointerInfo::getGOT(), false, false, false, - 0); - unsigned LoFlag = HasMips64 ? MipsII::MO_GOT_OFST : MipsII::MO_ABS_LO; - SDValue Lo = DAG.getNode(MipsISD::Lo, DL, Ty, getTargetNode(Op, DAG, LoFlag)); - return DAG.getNode(ISD::ADD, DL, Ty, Load, Lo); -} - -SDValue MipsTargetLowering::getAddrGlobal(SDValue Op, SelectionDAG &DAG, +SDValue MipsTargetLowering::getTargetNode(JumpTableSDNode *N, EVT Ty, + SelectionDAG &DAG, unsigned Flag) const { - DebugLoc DL = Op.getDebugLoc(); - EVT Ty = Op.getValueType(); - SDValue Tgt = DAG.getNode(MipsISD::Wrapper, DL, Ty, getGlobalReg(DAG, Ty), - getTargetNode(Op, DAG, Flag)); - return DAG.getLoad(Ty, DL, DAG.getEntryNode(), Tgt, - MachinePointerInfo::getGOT(), false, false, false, 0); + return DAG.getTargetJumpTable(N->getIndex(), Ty, Flag); } -SDValue MipsTargetLowering::getAddrGlobalLargeGOT(SDValue Op, SelectionDAG &DAG, - unsigned HiFlag, - unsigned LoFlag) const { - DebugLoc DL = Op.getDebugLoc(); - EVT Ty = Op.getValueType(); - SDValue Hi = DAG.getNode(MipsISD::Hi, DL, Ty, getTargetNode(Op, DAG, HiFlag)); - Hi = DAG.getNode(ISD::ADD, DL, Ty, Hi, getGlobalReg(DAG, Ty)); - SDValue Wrapper = DAG.getNode(MipsISD::Wrapper, DL, Ty, Hi, - getTargetNode(Op, DAG, LoFlag)); - return DAG.getLoad(Ty, DL, DAG.getEntryNode(), Wrapper, - MachinePointerInfo::getGOT(), false, false, false, 0); +SDValue MipsTargetLowering::getTargetNode(ConstantPoolSDNode *N, EVT Ty, + SelectionDAG &DAG, + unsigned Flag) const { + return DAG.getTargetConstantPool(N->getConstVal(), Ty, N->getAlignment(), + N->getOffset(), Flag); } const char *MipsTargetLowering::getTargetNodeName(unsigned Opcode) const { @@ -157,9 +126,10 @@ const char *MipsTargetLowering::getTargetNodeName(unsigned Opcode) const { case MipsISD::FPCmp: return "MipsISD::FPCmp"; case MipsISD::CMovFP_T: return "MipsISD::CMovFP_T"; case MipsISD::CMovFP_F: return "MipsISD::CMovFP_F"; - case MipsISD::FPRound: return "MipsISD::FPRound"; - case MipsISD::ExtractLOHI: return "MipsISD::ExtractLOHI"; - case MipsISD::InsertLOHI: return "MipsISD::InsertLOHI"; + case MipsISD::TruncIntFP: return "MipsISD::TruncIntFP"; + case MipsISD::MFHI: return "MipsISD::MFHI"; + case MipsISD::MFLO: return "MipsISD::MFLO"; + case MipsISD::MTLOHI: return "MipsISD::MTLOHI"; case MipsISD::Mult: return "MipsISD::Mult"; case MipsISD::Multu: return "MipsISD::Multu"; case MipsISD::MAdd: return "MipsISD::MAdd"; @@ -198,6 +168,35 @@ const char *MipsTargetLowering::getTargetNodeName(unsigned Opcode) const { case MipsISD::MADDU_DSP: return "MipsISD::MADDU_DSP"; case MipsISD::MSUB_DSP: return "MipsISD::MSUB_DSP"; case MipsISD::MSUBU_DSP: return "MipsISD::MSUBU_DSP"; + case MipsISD::SHLL_DSP: return "MipsISD::SHLL_DSP"; + case MipsISD::SHRA_DSP: return "MipsISD::SHRA_DSP"; + case MipsISD::SHRL_DSP: return "MipsISD::SHRL_DSP"; + case MipsISD::SETCC_DSP: return "MipsISD::SETCC_DSP"; + case MipsISD::SELECT_CC_DSP: return "MipsISD::SELECT_CC_DSP"; + case MipsISD::VALL_ZERO: return "MipsISD::VALL_ZERO"; + case MipsISD::VANY_ZERO: return "MipsISD::VANY_ZERO"; + case MipsISD::VALL_NONZERO: return "MipsISD::VALL_NONZERO"; + case MipsISD::VANY_NONZERO: return "MipsISD::VANY_NONZERO"; + case MipsISD::VCEQ: return "MipsISD::VCEQ"; + case MipsISD::VCLE_S: return "MipsISD::VCLE_S"; + case MipsISD::VCLE_U: return "MipsISD::VCLE_U"; + case MipsISD::VCLT_S: return "MipsISD::VCLT_S"; + case MipsISD::VCLT_U: return "MipsISD::VCLT_U"; + case MipsISD::VSMAX: return "MipsISD::VSMAX"; + case MipsISD::VSMIN: return "MipsISD::VSMIN"; + case MipsISD::VUMAX: return "MipsISD::VUMAX"; + case MipsISD::VUMIN: return "MipsISD::VUMIN"; + case MipsISD::VEXTRACT_SEXT_ELT: return "MipsISD::VEXTRACT_SEXT_ELT"; + case MipsISD::VEXTRACT_ZEXT_ELT: return "MipsISD::VEXTRACT_ZEXT_ELT"; + case MipsISD::VNOR: return "MipsISD::VNOR"; + case MipsISD::VSHF: return "MipsISD::VSHF"; + case MipsISD::SHF: return "MipsISD::SHF"; + case MipsISD::ILVEV: return "MipsISD::ILVEV"; + case MipsISD::ILVOD: return "MipsISD::ILVOD"; + case MipsISD::ILVL: return "MipsISD::ILVL"; + case MipsISD::ILVR: return "MipsISD::ILVR"; + case MipsISD::PCKEV: return "MipsISD::PCKEV"; + case MipsISD::PCKOD: return "MipsISD::PCKOD"; default: return NULL; } } @@ -211,7 +210,7 @@ MipsTargetLowering(MipsTargetMachine &TM) // Mips does not have i1 type, so use i32 for // setcc operations results (slt, sgt, ...). setBooleanContents(ZeroOrOneBooleanContent); - setBooleanVectorContents(ZeroOrOneBooleanContent); // FIXME: Is this correct? + setBooleanVectorContents(ZeroOrNegativeOneBooleanContent); // Load extented operations for i1 types must be promoted setLoadExtAction(ISD::EXTLOAD, MVT::i1, Promote); @@ -246,6 +245,7 @@ MipsTargetLowering(MipsTargetMachine &TM) setOperationAction(ISD::VASTART, MVT::Other, Custom); setOperationAction(ISD::FCOPYSIGN, MVT::f32, Custom); setOperationAction(ISD::FCOPYSIGN, MVT::f64, Custom); + setOperationAction(ISD::FP_TO_SINT, MVT::i32, Custom); if (!TM.Options.NoNaNsFPMath) { setOperationAction(ISD::FABS, MVT::f32, Custom); @@ -261,6 +261,7 @@ MipsTargetLowering(MipsTargetMachine &TM) setOperationAction(ISD::SELECT, MVT::i64, Custom); setOperationAction(ISD::LOAD, MVT::i64, Custom); setOperationAction(ISD::STORE, MVT::i64, Custom); + setOperationAction(ISD::FP_TO_SINT, MVT::i64, Custom); } if (!HasMips64) { @@ -335,20 +336,12 @@ MipsTargetLowering(MipsTargetMachine &TM) setOperationAction(ISD::FNEG, MVT::f64, Expand); } - setOperationAction(ISD::EXCEPTIONADDR, MVT::i32, Expand); - setOperationAction(ISD::EXCEPTIONADDR, MVT::i64, Expand); - setOperationAction(ISD::EHSELECTION, MVT::i32, Expand); - setOperationAction(ISD::EHSELECTION, MVT::i64, Expand); - setOperationAction(ISD::EH_RETURN, MVT::Other, Custom); setOperationAction(ISD::VAARG, MVT::Other, Expand); setOperationAction(ISD::VACOPY, MVT::Other, Expand); setOperationAction(ISD::VAEND, MVT::Other, Expand); - setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::i64, Custom); - setOperationAction(ISD::INTRINSIC_W_CHAIN, MVT::i64, Custom); - // Use the default for now setOperationAction(ISD::STACKSAVE, MVT::Other, Expand); setOperationAction(ISD::STACKRESTORE, MVT::Other, Expand); @@ -382,6 +375,8 @@ MipsTargetLowering(MipsTargetMachine &TM) setTruncStoreAction(MVT::i64, MVT::i32, Custom); } + setOperationAction(ISD::TRAP, MVT::Other, Legal); + setTargetDAGCombine(ISD::SDIVREM); setTargetDAGCombine(ISD::UDIVREM); setTargetDAGCombine(ISD::SELECT); @@ -397,6 +392,8 @@ MipsTargetLowering(MipsTargetMachine &TM) setExceptionSelectorRegister(IsN64 ? Mips::A1_64 : Mips::A1); MaxStoresPerMemcpy = 16; + + isMicroMips = Subtarget->inMicroMipsMode(); } const MipsTargetLowering *MipsTargetLowering::create(MipsTargetMachine &TM) { @@ -406,7 +403,7 @@ const MipsTargetLowering *MipsTargetLowering::create(MipsTargetMachine &TM) { return llvm::createMipsSETargetLowering(TM); } -EVT MipsTargetLowering::getSetCCResultType(EVT VT) const { +EVT MipsTargetLowering::getSetCCResultType(LLVMContext &, EVT VT) const { if (!VT.isVector()) return MVT::i32; return VT.changeVectorElementTypeToInteger(); @@ -419,11 +416,11 @@ static SDValue performDivRemCombine(SDNode *N, SelectionDAG &DAG, return SDValue(); EVT Ty = N->getValueType(0); - unsigned LO = (Ty == MVT::i32) ? Mips::LO : Mips::LO64; - unsigned HI = (Ty == MVT::i32) ? Mips::HI : Mips::HI64; + unsigned LO = (Ty == MVT::i32) ? Mips::LO0 : Mips::LO0_64; + unsigned HI = (Ty == MVT::i32) ? Mips::HI0 : Mips::HI0_64; unsigned Opc = N->getOpcode() == ISD::SDIVREM ? MipsISD::DivRem16 : MipsISD::DivRemU16; - DebugLoc DL = N->getDebugLoc(); + SDLoc DL(N); SDValue DivRem = DAG.getNode(Opc, DL, MVT::Glue, N->getOperand(0), N->getOperand(1)); @@ -449,7 +446,7 @@ static SDValue performDivRemCombine(SDNode *N, SelectionDAG &DAG, return SDValue(); } -static Mips::CondCode FPCondCCodeToFCC(ISD::CondCode CC) { +static Mips::CondCode condCodeToFCC(ISD::CondCode CC) { switch (CC) { default: llvm_unreachable("Unknown fp condition code!"); case ISD::SETEQ: @@ -501,24 +498,25 @@ static SDValue createFPCmp(SelectionDAG &DAG, const SDValue &Op) { return Op; SDValue RHS = Op.getOperand(1); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); // Assume the 3rd operand is a CondCodeSDNode. Add code to check the type of // node if necessary. ISD::CondCode CC = cast(Op.getOperand(2))->get(); return DAG.getNode(MipsISD::FPCmp, DL, MVT::Glue, LHS, RHS, - DAG.getConstant(FPCondCCodeToFCC(CC), MVT::i32)); + DAG.getConstant(condCodeToFCC(CC), MVT::i32)); } // Creates and returns a CMovFPT/F node. static SDValue createCMovFP(SelectionDAG &DAG, SDValue Cond, SDValue True, - SDValue False, DebugLoc DL) { + SDValue False, SDLoc DL) { ConstantSDNode *CC = cast(Cond.getOperand(2)); bool invert = invertFPCondCodeUser((Mips::CondCode)CC->getSExtValue()); + SDValue FCC0 = DAG.getRegister(Mips::FCC0, MVT::i32); return DAG.getNode((invert ? MipsISD::CMovFP_F : MipsISD::CMovFP_T), DL, - True.getValueType(), True, False, Cond); + True.getValueType(), True, FCC0, False, Cond); } static SDValue performSELECTCombine(SDNode *N, SelectionDAG &DAG, @@ -539,19 +537,65 @@ static SDValue performSELECTCombine(SDNode *N, SelectionDAG &DAG, if (!FalseTy.isInteger()) return SDValue(); - ConstantSDNode *CN = dyn_cast(False); + ConstantSDNode *FalseC = dyn_cast(False); - if (!CN || CN->getZExtValue()) + // If the RHS (False) is 0, we swap the order of the operands + // of ISD::SELECT (obviously also inverting the condition) so that we can + // take advantage of conditional moves using the $0 register. + // Example: + // return (a != 0) ? x : 0; + // load $reg, x + // movz $reg, $0, a + if (!FalseC) return SDValue(); - const DebugLoc DL = N->getDebugLoc(); - ISD::CondCode CC = cast(SetCC.getOperand(2))->get(); + const SDLoc DL(N); + + if (!FalseC->getZExtValue()) { + ISD::CondCode CC = cast(SetCC.getOperand(2))->get(); + SDValue True = N->getOperand(1); + + SetCC = DAG.getSetCC(DL, SetCC.getValueType(), SetCC.getOperand(0), + SetCC.getOperand(1), ISD::getSetCCInverse(CC, true)); + + return DAG.getNode(ISD::SELECT, DL, FalseTy, SetCC, False, True); + } + + // If both operands are integer constants there's a possibility that we + // can do some interesting optimizations. SDValue True = N->getOperand(1); + ConstantSDNode *TrueC = dyn_cast(True); + + if (!TrueC || !True.getValueType().isInteger()) + return SDValue(); - SetCC = DAG.getSetCC(DL, SetCC.getValueType(), SetCC.getOperand(0), - SetCC.getOperand(1), ISD::getSetCCInverse(CC, true)); + // We'll also ignore MVT::i64 operands as this optimizations proves + // to be ineffective because of the required sign extensions as the result + // of a SETCC operator is always MVT::i32 for non-vector types. + if (True.getValueType() == MVT::i64) + return SDValue(); - return DAG.getNode(ISD::SELECT, DL, FalseTy, SetCC, False, True); + int64_t Diff = TrueC->getSExtValue() - FalseC->getSExtValue(); + + // 1) (a < x) ? y : y-1 + // slti $reg1, a, x + // addiu $reg2, $reg1, y-1 + if (Diff == 1) + return DAG.getNode(ISD::ADD, DL, SetCC.getValueType(), SetCC, False); + + // 2) (a < x) ? y-1 : y + // slti $reg1, a, x + // xor $reg1, $reg1, 1 + // addiu $reg2, $reg1, y-1 + if (Diff == -1) { + ISD::CondCode CC = cast(SetCC.getOperand(2))->get(); + SetCC = DAG.getSetCC(DL, SetCC.getValueType(), SetCC.getOperand(0), + SetCC.getOperand(1), ISD::getSetCCInverse(CC, true)); + return DAG.getNode(ISD::ADD, DL, SetCC.getValueType(), SetCC, True); + } + + // Couldn't optimize. + return SDValue(); } static SDValue performANDCombine(SDNode *N, SelectionDAG &DAG, @@ -560,7 +604,7 @@ static SDValue performANDCombine(SDNode *N, SelectionDAG &DAG, // Pattern match EXT. // $dst = and ((sra or srl) $src , pos), (2**size - 1) // => ext $dst, $src, size, pos - if (DCI.isBeforeLegalizeOps() || !Subtarget->hasMips32r2()) + if (DCI.isBeforeLegalizeOps() || !Subtarget->hasExtractInsert()) return SDValue(); SDValue ShiftRight = N->getOperand(0), Mask = N->getOperand(1); @@ -589,7 +633,7 @@ static SDValue performANDCombine(SDNode *N, SelectionDAG &DAG, if (SMPos != 0 || Pos + SMSize > ValTy.getSizeInBits()) return SDValue(); - return DAG.getNode(MipsISD::Ext, N->getDebugLoc(), ValTy, + return DAG.getNode(MipsISD::Ext, SDLoc(N), ValTy, ShiftRight.getOperand(0), DAG.getConstant(Pos, MVT::i32), DAG.getConstant(SMSize, MVT::i32)); } @@ -601,7 +645,7 @@ static SDValue performORCombine(SDNode *N, SelectionDAG &DAG, // $dst = or (and $src1 , mask0), (and (shl $src, pos), mask1), // where mask1 = (2**size - 1) << pos, mask0 = ~mask1 // => ins $dst, $src, size, pos, $src1 - if (DCI.isBeforeLegalizeOps() || !Subtarget->hasMips32r2()) + if (DCI.isBeforeLegalizeOps() || !Subtarget->hasExtractInsert()) return SDValue(); SDValue And0 = N->getOperand(0), And1 = N->getOperand(1); @@ -643,7 +687,7 @@ static SDValue performORCombine(SDNode *N, SelectionDAG &DAG, if ((Shamt != SMPos0) || (SMPos0 + SMSize0 > ValTy.getSizeInBits())) return SDValue(); - return DAG.getNode(MipsISD::Ins, N->getDebugLoc(), ValTy, Shl.getOperand(0), + return DAG.getNode(MipsISD::Ins, SDLoc(N), ValTy, Shl.getOperand(0), DAG.getConstant(SMPos0, MVT::i32), DAG.getConstant(SMSize0, MVT::i32), And0.getOperand(0)); } @@ -668,7 +712,7 @@ static SDValue performADDCombine(SDNode *N, SelectionDAG &DAG, return SDValue(); EVT ValTy = N->getValueType(0); - DebugLoc DL = N->getDebugLoc(); + SDLoc DL(N); SDValue Add1 = DAG.getNode(ISD::ADD, DL, ValTy, N->getOperand(0), Add.getOperand(0)); @@ -712,10 +756,7 @@ void MipsTargetLowering::ReplaceNodeResults(SDNode *N, SmallVectorImpl &Results, SelectionDAG &DAG) const { - SDValue Res = LowerOperation(SDValue(N, 0), DAG); - - for (unsigned I = 0, E = Res->getNumValues(); I != E; ++I) - Results.push_back(Res.getValue(I)); + return LowerOperationWrapper(N, Results, DAG); } SDValue MipsTargetLowering:: @@ -739,16 +780,14 @@ LowerOperation(SDValue Op, SelectionDAG &DAG) const case ISD::FRAMEADDR: return lowerFRAMEADDR(Op, DAG); case ISD::RETURNADDR: return lowerRETURNADDR(Op, DAG); case ISD::EH_RETURN: return lowerEH_RETURN(Op, DAG); - case ISD::MEMBARRIER: return lowerMEMBARRIER(Op, DAG); case ISD::ATOMIC_FENCE: return lowerATOMIC_FENCE(Op, DAG); case ISD::SHL_PARTS: return lowerShiftLeftParts(Op, DAG); case ISD::SRA_PARTS: return lowerShiftRightParts(Op, DAG, true); case ISD::SRL_PARTS: return lowerShiftRightParts(Op, DAG, false); case ISD::LOAD: return lowerLOAD(Op, DAG); case ISD::STORE: return lowerSTORE(Op, DAG); - case ISD::INTRINSIC_WO_CHAIN: return lowerINTRINSIC_WO_CHAIN(Op, DAG); - case ISD::INTRINSIC_W_CHAIN: return lowerINTRINSIC_W_CHAIN(Op, DAG); case ISD::ADD: return lowerADD(Op, DAG); + case ISD::FP_TO_SINT: return lowerFP_TO_SINT(Op, DAG); } return SDValue(); } @@ -768,6 +807,30 @@ addLiveIn(MachineFunction &MF, unsigned PReg, const TargetRegisterClass *RC) return VReg; } +static MachineBasicBlock *expandPseudoDIV(MachineInstr *MI, + MachineBasicBlock &MBB, + const TargetInstrInfo &TII, + bool Is64Bit) { + if (NoZeroDivCheck) + return &MBB; + + // Insert instruction "teq $divisor_reg, $zero, 7". + MachineBasicBlock::iterator I(MI); + MachineInstrBuilder MIB; + MachineOperand &Divisor = MI->getOperand(2); + MIB = BuildMI(MBB, std::next(I), MI->getDebugLoc(), TII.get(Mips::TEQ)) + .addReg(Divisor.getReg(), getKillRegState(Divisor.isKill())) + .addReg(Mips::ZERO).addImm(7); + + // Use the 32-bit sub-register if this is a 64-bit division. + if (Is64Bit) + MIB->getOperand(0).setSubReg(Mips::sub_32); + + // Clear Divisor's kill flag. + Divisor.setIsKill(false); + return &MBB; +} + MachineBasicBlock * MipsTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI, MachineBasicBlock *BB) const { @@ -775,108 +838,82 @@ MipsTargetLowering::EmitInstrWithCustomInserter(MachineInstr *MI, default: llvm_unreachable("Unexpected instr type to insert"); case Mips::ATOMIC_LOAD_ADD_I8: - case Mips::ATOMIC_LOAD_ADD_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, Mips::ADDu); case Mips::ATOMIC_LOAD_ADD_I16: - case Mips::ATOMIC_LOAD_ADD_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, Mips::ADDu); case Mips::ATOMIC_LOAD_ADD_I32: - case Mips::ATOMIC_LOAD_ADD_I32_P8: return emitAtomicBinary(MI, BB, 4, Mips::ADDu); case Mips::ATOMIC_LOAD_ADD_I64: - case Mips::ATOMIC_LOAD_ADD_I64_P8: return emitAtomicBinary(MI, BB, 8, Mips::DADDu); case Mips::ATOMIC_LOAD_AND_I8: - case Mips::ATOMIC_LOAD_AND_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, Mips::AND); case Mips::ATOMIC_LOAD_AND_I16: - case Mips::ATOMIC_LOAD_AND_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, Mips::AND); case Mips::ATOMIC_LOAD_AND_I32: - case Mips::ATOMIC_LOAD_AND_I32_P8: return emitAtomicBinary(MI, BB, 4, Mips::AND); case Mips::ATOMIC_LOAD_AND_I64: - case Mips::ATOMIC_LOAD_AND_I64_P8: return emitAtomicBinary(MI, BB, 8, Mips::AND64); case Mips::ATOMIC_LOAD_OR_I8: - case Mips::ATOMIC_LOAD_OR_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, Mips::OR); case Mips::ATOMIC_LOAD_OR_I16: - case Mips::ATOMIC_LOAD_OR_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, Mips::OR); case Mips::ATOMIC_LOAD_OR_I32: - case Mips::ATOMIC_LOAD_OR_I32_P8: return emitAtomicBinary(MI, BB, 4, Mips::OR); case Mips::ATOMIC_LOAD_OR_I64: - case Mips::ATOMIC_LOAD_OR_I64_P8: return emitAtomicBinary(MI, BB, 8, Mips::OR64); case Mips::ATOMIC_LOAD_XOR_I8: - case Mips::ATOMIC_LOAD_XOR_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, Mips::XOR); case Mips::ATOMIC_LOAD_XOR_I16: - case Mips::ATOMIC_LOAD_XOR_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, Mips::XOR); case Mips::ATOMIC_LOAD_XOR_I32: - case Mips::ATOMIC_LOAD_XOR_I32_P8: return emitAtomicBinary(MI, BB, 4, Mips::XOR); case Mips::ATOMIC_LOAD_XOR_I64: - case Mips::ATOMIC_LOAD_XOR_I64_P8: return emitAtomicBinary(MI, BB, 8, Mips::XOR64); case Mips::ATOMIC_LOAD_NAND_I8: - case Mips::ATOMIC_LOAD_NAND_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, 0, true); case Mips::ATOMIC_LOAD_NAND_I16: - case Mips::ATOMIC_LOAD_NAND_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, 0, true); case Mips::ATOMIC_LOAD_NAND_I32: - case Mips::ATOMIC_LOAD_NAND_I32_P8: return emitAtomicBinary(MI, BB, 4, 0, true); case Mips::ATOMIC_LOAD_NAND_I64: - case Mips::ATOMIC_LOAD_NAND_I64_P8: return emitAtomicBinary(MI, BB, 8, 0, true); case Mips::ATOMIC_LOAD_SUB_I8: - case Mips::ATOMIC_LOAD_SUB_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, Mips::SUBu); case Mips::ATOMIC_LOAD_SUB_I16: - case Mips::ATOMIC_LOAD_SUB_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, Mips::SUBu); case Mips::ATOMIC_LOAD_SUB_I32: - case Mips::ATOMIC_LOAD_SUB_I32_P8: return emitAtomicBinary(MI, BB, 4, Mips::SUBu); case Mips::ATOMIC_LOAD_SUB_I64: - case Mips::ATOMIC_LOAD_SUB_I64_P8: return emitAtomicBinary(MI, BB, 8, Mips::DSUBu); case Mips::ATOMIC_SWAP_I8: - case Mips::ATOMIC_SWAP_I8_P8: return emitAtomicBinaryPartword(MI, BB, 1, 0); case Mips::ATOMIC_SWAP_I16: - case Mips::ATOMIC_SWAP_I16_P8: return emitAtomicBinaryPartword(MI, BB, 2, 0); case Mips::ATOMIC_SWAP_I32: - case Mips::ATOMIC_SWAP_I32_P8: return emitAtomicBinary(MI, BB, 4, 0); case Mips::ATOMIC_SWAP_I64: - case Mips::ATOMIC_SWAP_I64_P8: return emitAtomicBinary(MI, BB, 8, 0); case Mips::ATOMIC_CMP_SWAP_I8: - case Mips::ATOMIC_CMP_SWAP_I8_P8: return emitAtomicCmpSwapPartword(MI, BB, 1); case Mips::ATOMIC_CMP_SWAP_I16: - case Mips::ATOMIC_CMP_SWAP_I16_P8: return emitAtomicCmpSwapPartword(MI, BB, 2); case Mips::ATOMIC_CMP_SWAP_I32: - case Mips::ATOMIC_CMP_SWAP_I32_P8: return emitAtomicCmpSwap(MI, BB, 4); case Mips::ATOMIC_CMP_SWAP_I64: - case Mips::ATOMIC_CMP_SWAP_I64_P8: return emitAtomicCmpSwap(MI, BB, 8); + case Mips::PseudoSDIV: + case Mips::PseudoUDIV: + return expandPseudoDIV(MI, *BB, *getTargetMachine().getInstrInfo(), false); + case Mips::PseudoDSDIV: + case Mips::PseudoDUDIV: + return expandPseudoDIV(MI, *BB, *getTargetMachine().getInstrInfo(), true); } } @@ -896,16 +933,16 @@ MipsTargetLowering::emitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB, unsigned LL, SC, AND, NOR, ZERO, BEQ; if (Size == 4) { - LL = IsN64 ? Mips::LL_P8 : Mips::LL; - SC = IsN64 ? Mips::SC_P8 : Mips::SC; + LL = isMicroMips ? Mips::LL_MM : Mips::LL; + SC = isMicroMips ? Mips::SC_MM : Mips::SC; AND = Mips::AND; NOR = Mips::NOR; ZERO = Mips::ZERO; BEQ = Mips::BEQ; } else { - LL = IsN64 ? Mips::LLD_P8 : Mips::LLD; - SC = IsN64 ? Mips::SCD_P8 : Mips::SCD; + LL = Mips::LLD; + SC = Mips::SCD; AND = Mips::AND64; NOR = Mips::NOR64; ZERO = Mips::ZERO_64; @@ -931,8 +968,7 @@ MipsTargetLowering::emitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB, // Transfer the remainder of BB and its successor edges to exitMBB. exitMBB->splice(exitMBB->begin(), BB, - llvm::next(MachineBasicBlock::iterator(MI)), - BB->end()); + std::next(MachineBasicBlock::iterator(MI)), BB->end()); exitMBB->transferSuccessorsAndUpdatePHIs(BB); // thisMBB: @@ -963,7 +999,7 @@ MipsTargetLowering::emitAtomicBinary(MachineInstr *MI, MachineBasicBlock *BB, BuildMI(BB, DL, TII->get(SC), Success).addReg(StoreVal).addReg(Ptr).addImm(0); BuildMI(BB, DL, TII->get(BEQ)).addReg(Success).addReg(ZERO).addMBB(loopMBB); - MI->eraseFromParent(); // The instruction is gone now. + MI->eraseFromParent(); // The instruction is gone now. return exitMBB; } @@ -974,15 +1010,13 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, unsigned Size, unsigned BinOpcode, bool Nand) const { assert((Size == 1 || Size == 2) && - "Unsupported size for EmitAtomicBinaryPartial."); + "Unsupported size for EmitAtomicBinaryPartial."); MachineFunction *MF = BB->getParent(); MachineRegisterInfo &RegInfo = MF->getRegInfo(); const TargetRegisterClass *RC = getRegClassFor(MVT::i32); const TargetInstrInfo *TII = getTargetMachine().getInstrInfo(); DebugLoc DL = MI->getDebugLoc(); - unsigned LL = IsN64 ? Mips::LL_P8 : Mips::LL; - unsigned SC = IsN64 ? Mips::SC_P8 : Mips::SC; unsigned Dest = MI->getOperand(0).getReg(); unsigned Ptr = MI->getOperand(1).getReg(); @@ -1020,7 +1054,7 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, // Transfer the remainder of BB and its successor edges to exitMBB. exitMBB->splice(exitMBB->begin(), BB, - llvm::next(MachineBasicBlock::iterator(MI)), BB->end()); + std::next(MachineBasicBlock::iterator(MI)), BB->end()); exitMBB->transferSuccessorsAndUpdatePHIs(BB); BB->addSuccessor(loopMBB); @@ -1044,13 +1078,20 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, BuildMI(BB, DL, TII->get(Mips::AND), AlignedAddr) .addReg(Ptr).addReg(MaskLSB2); BuildMI(BB, DL, TII->get(Mips::ANDi), PtrLSB2).addReg(Ptr).addImm(3); - BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(PtrLSB2).addImm(3); + if (Subtarget->isLittle()) { + BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(PtrLSB2).addImm(3); + } else { + unsigned Off = RegInfo.createVirtualRegister(RC); + BuildMI(BB, DL, TII->get(Mips::XORi), Off) + .addReg(PtrLSB2).addImm((Size == 1) ? 3 : 2); + BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(Off).addImm(3); + } BuildMI(BB, DL, TII->get(Mips::ORi), MaskUpper) .addReg(Mips::ZERO).addImm(MaskImm); BuildMI(BB, DL, TII->get(Mips::SLLV), Mask) - .addReg(ShiftAmt).addReg(MaskUpper); + .addReg(MaskUpper).addReg(ShiftAmt); BuildMI(BB, DL, TII->get(Mips::NOR), Mask2).addReg(Mips::ZERO).addReg(Mask); - BuildMI(BB, DL, TII->get(Mips::SLLV), Incr2).addReg(ShiftAmt).addReg(Incr); + BuildMI(BB, DL, TII->get(Mips::SLLV), Incr2).addReg(Incr).addReg(ShiftAmt); // atomic.load.binop // loopMBB: @@ -1072,7 +1113,7 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, // beq success,$0,loopMBB BB = loopMBB; - BuildMI(BB, DL, TII->get(LL), OldVal).addReg(AlignedAddr).addImm(0); + BuildMI(BB, DL, TII->get(Mips::LL), OldVal).addReg(AlignedAddr).addImm(0); if (Nand) { // and andres, oldval, incr2 // nor binopres, $0, andres @@ -1086,7 +1127,7 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, // and newval, binopres, mask BuildMI(BB, DL, TII->get(BinOpcode), BinOpRes).addReg(OldVal).addReg(Incr2); BuildMI(BB, DL, TII->get(Mips::AND), NewVal).addReg(BinOpRes).addReg(Mask); - } else {// atomic.swap + } else { // atomic.swap // and newval, incr2, mask BuildMI(BB, DL, TII->get(Mips::AND), NewVal).addReg(Incr2).addReg(Mask); } @@ -1095,7 +1136,7 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, .addReg(OldVal).addReg(Mask2); BuildMI(BB, DL, TII->get(Mips::OR), StoreVal) .addReg(MaskedOldVal0).addReg(NewVal); - BuildMI(BB, DL, TII->get(SC), Success) + BuildMI(BB, DL, TII->get(Mips::SC), Success) .addReg(StoreVal).addReg(AlignedAddr).addImm(0); BuildMI(BB, DL, TII->get(Mips::BEQ)) .addReg(Success).addReg(Mips::ZERO).addMBB(loopMBB); @@ -1111,21 +1152,20 @@ MipsTargetLowering::emitAtomicBinaryPartword(MachineInstr *MI, BuildMI(BB, DL, TII->get(Mips::AND), MaskedOldVal1) .addReg(OldVal).addReg(Mask); BuildMI(BB, DL, TII->get(Mips::SRLV), SrlRes) - .addReg(ShiftAmt).addReg(MaskedOldVal1); + .addReg(MaskedOldVal1).addReg(ShiftAmt); BuildMI(BB, DL, TII->get(Mips::SLL), SllRes) .addReg(SrlRes).addImm(ShiftImm); BuildMI(BB, DL, TII->get(Mips::SRA), Dest) .addReg(SllRes).addImm(ShiftImm); - MI->eraseFromParent(); // The instruction is gone now. + MI->eraseFromParent(); // The instruction is gone now. return exitMBB; } -MachineBasicBlock * -MipsTargetLowering::emitAtomicCmpSwap(MachineInstr *MI, - MachineBasicBlock *BB, - unsigned Size) const { +MachineBasicBlock * MipsTargetLowering::emitAtomicCmpSwap(MachineInstr *MI, + MachineBasicBlock *BB, + unsigned Size) const { assert((Size == 4 || Size == 8) && "Unsupported size for EmitAtomicCmpSwap."); MachineFunction *MF = BB->getParent(); @@ -1136,15 +1176,14 @@ MipsTargetLowering::emitAtomicCmpSwap(MachineInstr *MI, unsigned LL, SC, ZERO, BNE, BEQ; if (Size == 4) { - LL = IsN64 ? Mips::LL_P8 : Mips::LL; - SC = IsN64 ? Mips::SC_P8 : Mips::SC; + LL = isMicroMips ? Mips::LL_MM : Mips::LL; + SC = isMicroMips ? Mips::SC_MM : Mips::SC; ZERO = Mips::ZERO; BNE = Mips::BNE; BEQ = Mips::BEQ; - } - else { - LL = IsN64 ? Mips::LLD_P8 : Mips::LLD; - SC = IsN64 ? Mips::SCD_P8 : Mips::SCD; + } else { + LL = Mips::LLD; + SC = Mips::SCD; ZERO = Mips::ZERO_64; BNE = Mips::BNE64; BEQ = Mips::BEQ64; @@ -1170,7 +1209,7 @@ MipsTargetLowering::emitAtomicCmpSwap(MachineInstr *MI, // Transfer the remainder of BB and its successor edges to exitMBB. exitMBB->splice(exitMBB->begin(), BB, - llvm::next(MachineBasicBlock::iterator(MI)), BB->end()); + std::next(MachineBasicBlock::iterator(MI)), BB->end()); exitMBB->transferSuccessorsAndUpdatePHIs(BB); // thisMBB: @@ -1199,7 +1238,7 @@ MipsTargetLowering::emitAtomicCmpSwap(MachineInstr *MI, BuildMI(BB, DL, TII->get(BEQ)) .addReg(Success).addReg(ZERO).addMBB(loop1MBB); - MI->eraseFromParent(); // The instruction is gone now. + MI->eraseFromParent(); // The instruction is gone now. return exitMBB; } @@ -1216,8 +1255,6 @@ MipsTargetLowering::emitAtomicCmpSwapPartword(MachineInstr *MI, const TargetRegisterClass *RC = getRegClassFor(MVT::i32); const TargetInstrInfo *TII = getTargetMachine().getInstrInfo(); DebugLoc DL = MI->getDebugLoc(); - unsigned LL = IsN64 ? Mips::LL_P8 : Mips::LL; - unsigned SC = IsN64 ? Mips::SC_P8 : Mips::SC; unsigned Dest = MI->getOperand(0).getReg(); unsigned Ptr = MI->getOperand(1).getReg(); @@ -1258,7 +1295,7 @@ MipsTargetLowering::emitAtomicCmpSwapPartword(MachineInstr *MI, // Transfer the remainder of BB and its successor edges to exitMBB. exitMBB->splice(exitMBB->begin(), BB, - llvm::next(MachineBasicBlock::iterator(MI)), BB->end()); + std::next(MachineBasicBlock::iterator(MI)), BB->end()); exitMBB->transferSuccessorsAndUpdatePHIs(BB); BB->addSuccessor(loop1MBB); @@ -1287,27 +1324,34 @@ MipsTargetLowering::emitAtomicCmpSwapPartword(MachineInstr *MI, BuildMI(BB, DL, TII->get(Mips::AND), AlignedAddr) .addReg(Ptr).addReg(MaskLSB2); BuildMI(BB, DL, TII->get(Mips::ANDi), PtrLSB2).addReg(Ptr).addImm(3); - BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(PtrLSB2).addImm(3); + if (Subtarget->isLittle()) { + BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(PtrLSB2).addImm(3); + } else { + unsigned Off = RegInfo.createVirtualRegister(RC); + BuildMI(BB, DL, TII->get(Mips::XORi), Off) + .addReg(PtrLSB2).addImm((Size == 1) ? 3 : 2); + BuildMI(BB, DL, TII->get(Mips::SLL), ShiftAmt).addReg(Off).addImm(3); + } BuildMI(BB, DL, TII->get(Mips::ORi), MaskUpper) .addReg(Mips::ZERO).addImm(MaskImm); BuildMI(BB, DL, TII->get(Mips::SLLV), Mask) - .addReg(ShiftAmt).addReg(MaskUpper); + .addReg(MaskUpper).addReg(ShiftAmt); BuildMI(BB, DL, TII->get(Mips::NOR), Mask2).addReg(Mips::ZERO).addReg(Mask); BuildMI(BB, DL, TII->get(Mips::ANDi), MaskedCmpVal) .addReg(CmpVal).addImm(MaskImm); BuildMI(BB, DL, TII->get(Mips::SLLV), ShiftedCmpVal) - .addReg(ShiftAmt).addReg(MaskedCmpVal); + .addReg(MaskedCmpVal).addReg(ShiftAmt); BuildMI(BB, DL, TII->get(Mips::ANDi), MaskedNewVal) .addReg(NewVal).addImm(MaskImm); BuildMI(BB, DL, TII->get(Mips::SLLV), ShiftedNewVal) - .addReg(ShiftAmt).addReg(MaskedNewVal); + .addReg(MaskedNewVal).addReg(ShiftAmt); // loop1MBB: // ll oldval,0(alginedaddr) // and maskedoldval0,oldval,mask // bne maskedoldval0,shiftedcmpval,sinkMBB BB = loop1MBB; - BuildMI(BB, DL, TII->get(LL), OldVal).addReg(AlignedAddr).addImm(0); + BuildMI(BB, DL, TII->get(Mips::LL), OldVal).addReg(AlignedAddr).addImm(0); BuildMI(BB, DL, TII->get(Mips::AND), MaskedOldVal0) .addReg(OldVal).addReg(Mask); BuildMI(BB, DL, TII->get(Mips::BNE)) @@ -1323,7 +1367,7 @@ MipsTargetLowering::emitAtomicCmpSwapPartword(MachineInstr *MI, .addReg(OldVal).addReg(Mask2); BuildMI(BB, DL, TII->get(Mips::OR), StoreVal) .addReg(MaskedOldVal1).addReg(ShiftedNewVal); - BuildMI(BB, DL, TII->get(SC), Success) + BuildMI(BB, DL, TII->get(Mips::SC), Success) .addReg(StoreVal).addReg(AlignedAddr).addImm(0); BuildMI(BB, DL, TII->get(Mips::BEQ)) .addReg(Success).addReg(Mips::ZERO).addMBB(loop1MBB); @@ -1336,7 +1380,7 @@ MipsTargetLowering::emitAtomicCmpSwapPartword(MachineInstr *MI, int64_t ShiftImm = (Size == 1) ? 24 : 16; BuildMI(BB, DL, TII->get(Mips::SRLV), SrlRes) - .addReg(ShiftAmt).addReg(MaskedOldVal0); + .addReg(MaskedOldVal0).addReg(ShiftAmt); BuildMI(BB, DL, TII->get(Mips::SLL), SllRes) .addReg(SrlRes).addImm(ShiftImm); BuildMI(BB, DL, TII->get(Mips::SRA), Dest) @@ -1354,7 +1398,7 @@ SDValue MipsTargetLowering::lowerBR_JT(SDValue Op, SelectionDAG &DAG) const { SDValue Chain = Op.getOperand(0); SDValue Table = Op.getOperand(1); SDValue Index = Op.getOperand(2); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); EVT PTy = getPointerTy(); unsigned EntrySize = DAG.getMachineFunction().getJumpTableInfo()->getEntrySize(*getDataLayout()); @@ -1380,14 +1424,12 @@ SDValue MipsTargetLowering::lowerBR_JT(SDValue Op, SelectionDAG &DAG) const { return DAG.getNode(ISD::BRIND, DL, MVT::Other, Chain, Addr); } -SDValue MipsTargetLowering:: -lowerBRCOND(SDValue Op, SelectionDAG &DAG) const -{ +SDValue MipsTargetLowering::lowerBRCOND(SDValue Op, SelectionDAG &DAG) const { // The first operand is the chain, the second is the condition, the third is // the block to branch to if the condition is true. SDValue Chain = Op.getOperand(0); SDValue Dest = Op.getOperand(2); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue CondRes = createFPCmp(DAG, Op.getOperand(1)); @@ -1400,8 +1442,9 @@ lowerBRCOND(SDValue Op, SelectionDAG &DAG) const (Mips::CondCode)cast(CCNode)->getZExtValue(); unsigned Opc = invertFPCondCodeUser(CC) ? Mips::BRANCH_F : Mips::BRANCH_T; SDValue BrCode = DAG.getConstant(Opc, MVT::i32); + SDValue FCC0 = DAG.getRegister(Mips::FCC0, MVT::i32); return DAG.getNode(MipsISD::FPBrcond, DL, Op.getValueType(), Chain, BrCode, - Dest, CondRes); + FCC0, Dest, CondRes); } SDValue MipsTargetLowering:: @@ -1414,15 +1457,16 @@ lowerSELECT(SDValue Op, SelectionDAG &DAG) const return Op; return createCMovFP(DAG, Cond, Op.getOperand(1), Op.getOperand(2), - Op.getDebugLoc()); + SDLoc(Op)); } SDValue MipsTargetLowering:: lowerSELECT_CC(SDValue Op, SelectionDAG &DAG) const { - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); EVT Ty = Op.getOperand(0).getValueType(); - SDValue Cond = DAG.getNode(ISD::SETCC, DL, getSetCCResultType(Ty), + SDValue Cond = DAG.getNode(ISD::SETCC, DL, + getSetCCResultType(*DAG.getContext(), Ty), Op.getOperand(0), Op.getOperand(1), Op.getOperand(4)); @@ -1439,14 +1483,16 @@ SDValue MipsTargetLowering::lowerSETCC(SDValue Op, SelectionDAG &DAG) const { SDValue True = DAG.getConstant(1, MVT::i32); SDValue False = DAG.getConstant(0, MVT::i32); - return createCMovFP(DAG, Cond, True, False, Op.getDebugLoc()); + return createCMovFP(DAG, Cond, True, False, SDLoc(Op)); } SDValue MipsTargetLowering::lowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const { // FIXME there isn't actually debug info here - DebugLoc DL = Op.getDebugLoc(); - const GlobalValue *GV = cast(Op)->getGlobal(); + SDLoc DL(Op); + EVT Ty = Op.getValueType(); + GlobalAddressSDNode *N = cast(Op); + const GlobalValue *GV = N->getGlobal(); if (getTargetMachine().getRelocationModel() != Reloc::PIC_ && !IsN64) { const MipsTargetObjectFile &TLOF = @@ -1463,26 +1509,31 @@ SDValue MipsTargetLowering::lowerGlobalAddress(SDValue Op, } // %hi/%lo relocation - return getAddrNonPIC(Op, DAG); + return getAddrNonPIC(N, Ty, DAG); } if (GV->hasInternalLinkage() || (GV->hasLocalLinkage() && !isa(GV))) - return getAddrLocal(Op, DAG, HasMips64); + return getAddrLocal(N, Ty, DAG, HasMips64); if (LargeGOT) - return getAddrGlobalLargeGOT(Op, DAG, MipsII::MO_GOT_HI16, - MipsII::MO_GOT_LO16); + return getAddrGlobalLargeGOT(N, Ty, DAG, MipsII::MO_GOT_HI16, + MipsII::MO_GOT_LO16, DAG.getEntryNode(), + MachinePointerInfo::getGOT()); - return getAddrGlobal(Op, DAG, - HasMips64 ? MipsII::MO_GOT_DISP : MipsII::MO_GOT16); + return getAddrGlobal(N, Ty, DAG, + HasMips64 ? MipsII::MO_GOT_DISP : MipsII::MO_GOT16, + DAG.getEntryNode(), MachinePointerInfo::getGOT()); } SDValue MipsTargetLowering::lowerBlockAddress(SDValue Op, SelectionDAG &DAG) const { + BlockAddressSDNode *N = cast(Op); + EVT Ty = Op.getValueType(); + if (getTargetMachine().getRelocationModel() != Reloc::PIC_ && !IsN64) - return getAddrNonPIC(Op, DAG); + return getAddrNonPIC(N, Ty, DAG); - return getAddrLocal(Op, DAG, HasMips64); + return getAddrLocal(N, Ty, DAG, HasMips64); } SDValue MipsTargetLowering:: @@ -1493,7 +1544,7 @@ lowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const // Local Exec TLS Model. GlobalAddressSDNode *GA = cast(Op); - DebugLoc DL = GA->getDebugLoc(); + SDLoc DL(GA); const GlobalValue *GV = GA->getGlobal(); EVT PtrVT = getPointerTy(); @@ -1569,10 +1620,13 @@ lowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const SDValue MipsTargetLowering:: lowerJumpTable(SDValue Op, SelectionDAG &DAG) const { + JumpTableSDNode *N = cast(Op); + EVT Ty = Op.getValueType(); + if (getTargetMachine().getRelocationModel() != Reloc::PIC_ && !IsN64) - return getAddrNonPIC(Op, DAG); + return getAddrNonPIC(N, Ty, DAG); - return getAddrLocal(Op, DAG, HasMips64); + return getAddrLocal(N, Ty, DAG, HasMips64); } SDValue MipsTargetLowering:: @@ -1587,18 +1641,20 @@ lowerConstantPool(SDValue Op, SelectionDAG &DAG) const // SDValue GPRelNode = DAG.getNode(MipsISD::GPRel, MVT::i32, CP); // SDValue GOT = DAG.getGLOBAL_OFFSET_TABLE(MVT::i32); // ResNode = DAG.getNode(ISD::ADD, MVT::i32, GOT, GPRelNode); + ConstantPoolSDNode *N = cast(Op); + EVT Ty = Op.getValueType(); if (getTargetMachine().getRelocationModel() != Reloc::PIC_ && !IsN64) - return getAddrNonPIC(Op, DAG); + return getAddrNonPIC(N, Ty, DAG); - return getAddrLocal(Op, DAG, HasMips64); + return getAddrLocal(N, Ty, DAG, HasMips64); } SDValue MipsTargetLowering::lowerVASTART(SDValue Op, SelectionDAG &DAG) const { MachineFunction &MF = DAG.getMachineFunction(); MipsFunctionInfo *FuncInfo = MF.getInfo(); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue FI = DAG.getFrameIndex(FuncInfo->getVarArgsFrameIndex(), getPointerTy()); @@ -1609,12 +1665,13 @@ SDValue MipsTargetLowering::lowerVASTART(SDValue Op, SelectionDAG &DAG) const { MachinePointerInfo(SV), false, false, 0); } -static SDValue lowerFCOPYSIGN32(SDValue Op, SelectionDAG &DAG, bool HasR2) { +static SDValue lowerFCOPYSIGN32(SDValue Op, SelectionDAG &DAG, + bool HasExtractInsert) { EVT TyX = Op.getOperand(0).getValueType(); EVT TyY = Op.getOperand(1).getValueType(); SDValue Const1 = DAG.getConstant(1, MVT::i32); SDValue Const31 = DAG.getConstant(31, MVT::i32); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue Res; // If operand is of type f64, extract the upper 32-bit. Otherwise, bitcast it @@ -1628,7 +1685,7 @@ static SDValue lowerFCOPYSIGN32(SDValue Op, SelectionDAG &DAG, bool HasR2) { DAG.getNode(MipsISD::ExtractElementF64, DL, MVT::i32, Op.getOperand(1), Const1); - if (HasR2) { + if (HasExtractInsert) { // ext E, Y, 31, 1 ; extract bit31 of Y // ins X, E, 31, 1 ; insert extracted bit at bit31 of X SDValue E = DAG.getNode(MipsISD::Ext, DL, MVT::i32, Y, Const31, Const1); @@ -1654,18 +1711,19 @@ static SDValue lowerFCOPYSIGN32(SDValue Op, SelectionDAG &DAG, bool HasR2) { return DAG.getNode(MipsISD::BuildPairF64, DL, MVT::f64, LowX, Res); } -static SDValue lowerFCOPYSIGN64(SDValue Op, SelectionDAG &DAG, bool HasR2) { +static SDValue lowerFCOPYSIGN64(SDValue Op, SelectionDAG &DAG, + bool HasExtractInsert) { unsigned WidthX = Op.getOperand(0).getValueSizeInBits(); unsigned WidthY = Op.getOperand(1).getValueSizeInBits(); EVT TyX = MVT::getIntegerVT(WidthX), TyY = MVT::getIntegerVT(WidthY); SDValue Const1 = DAG.getConstant(1, MVT::i32); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); // Bitcast to integer nodes. SDValue X = DAG.getNode(ISD::BITCAST, DL, TyX, Op.getOperand(0)); SDValue Y = DAG.getNode(ISD::BITCAST, DL, TyY, Op.getOperand(1)); - if (HasR2) { + if (HasExtractInsert) { // ext E, Y, width(Y) - 1, 1 ; extract bit width(Y)-1 of Y // ins X, E, width(X) - 1, 1 ; insert extracted bit at bit width(X)-1 of X SDValue E = DAG.getNode(MipsISD::Ext, DL, TyY, Y, @@ -1705,14 +1763,15 @@ static SDValue lowerFCOPYSIGN64(SDValue Op, SelectionDAG &DAG, bool HasR2) { SDValue MipsTargetLowering::lowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG) const { if (Subtarget->hasMips64()) - return lowerFCOPYSIGN64(Op, DAG, Subtarget->hasMips32r2()); + return lowerFCOPYSIGN64(Op, DAG, Subtarget->hasExtractInsert()); - return lowerFCOPYSIGN32(Op, DAG, Subtarget->hasMips32r2()); + return lowerFCOPYSIGN32(Op, DAG, Subtarget->hasExtractInsert()); } -static SDValue lowerFABS32(SDValue Op, SelectionDAG &DAG, bool HasR2) { +static SDValue lowerFABS32(SDValue Op, SelectionDAG &DAG, + bool HasExtractInsert) { SDValue Res, Const1 = DAG.getConstant(1, MVT::i32); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); // If operand is of type f64, extract the upper 32-bit. Otherwise, bitcast it // to i32. @@ -1722,7 +1781,7 @@ static SDValue lowerFABS32(SDValue Op, SelectionDAG &DAG, bool HasR2) { Const1); // Clear MSB. - if (HasR2) + if (HasExtractInsert) Res = DAG.getNode(MipsISD::Ins, DL, MVT::i32, DAG.getRegister(Mips::ZERO, MVT::i32), DAG.getConstant(31, MVT::i32), Const1, X); @@ -1739,15 +1798,16 @@ static SDValue lowerFABS32(SDValue Op, SelectionDAG &DAG, bool HasR2) { return DAG.getNode(MipsISD::BuildPairF64, DL, MVT::f64, LowX, Res); } -static SDValue lowerFABS64(SDValue Op, SelectionDAG &DAG, bool HasR2) { +static SDValue lowerFABS64(SDValue Op, SelectionDAG &DAG, + bool HasExtractInsert) { SDValue Res, Const1 = DAG.getConstant(1, MVT::i32); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); // Bitcast to integer node. SDValue X = DAG.getNode(ISD::BITCAST, DL, MVT::i64, Op.getOperand(0)); // Clear MSB. - if (HasR2) + if (HasExtractInsert) Res = DAG.getNode(MipsISD::Ins, DL, MVT::i64, DAG.getRegister(Mips::ZERO_64, MVT::i64), DAG.getConstant(63, MVT::i32), Const1, X); @@ -1762,9 +1822,9 @@ static SDValue lowerFABS64(SDValue Op, SelectionDAG &DAG, bool HasR2) { SDValue MipsTargetLowering::lowerFABS(SDValue Op, SelectionDAG &DAG) const { if (Subtarget->hasMips64() && (Op.getValueType() == MVT::f64)) - return lowerFABS64(Op, DAG, Subtarget->hasMips32r2()); + return lowerFABS64(Op, DAG, Subtarget->hasExtractInsert()); - return lowerFABS32(Op, DAG, Subtarget->hasMips32r2()); + return lowerFABS32(Op, DAG, Subtarget->hasExtractInsert()); } SDValue MipsTargetLowering:: @@ -1776,7 +1836,7 @@ lowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const { MachineFrameInfo *MFI = DAG.getMachineFunction().getFrameInfo(); MFI->setFrameAddressIsTaken(true); EVT VT = Op.getValueType(); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue FrameAddr = DAG.getCopyFromReg(DAG.getEntryNode(), DL, IsN64 ? Mips::FP_64 : Mips::FP, VT); return FrameAddr; @@ -1784,6 +1844,9 @@ lowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const { SDValue MipsTargetLowering::lowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const { + if (verifyReturnAddressArgumentIsConstant(Op, DAG)) + return SDValue(); + // check the depth assert((cast(Op.getOperand(0))->getZExtValue() == 0) && "Return address can be determined only for current frame."); @@ -1796,7 +1859,7 @@ SDValue MipsTargetLowering::lowerRETURNADDR(SDValue Op, // Return RA, which contains the return address. Mark it an implicit live-in. unsigned Reg = MF.addLiveIn(RA, getRegClassFor(VT)); - return DAG.getCopyFromReg(DAG.getEntryNode(), Op.getDebugLoc(), Reg, VT); + return DAG.getCopyFromReg(DAG.getEntryNode(), SDLoc(Op), Reg, VT); } // An EH_RETURN is the result of lowering llvm.eh.return which in turn is @@ -1812,7 +1875,7 @@ SDValue MipsTargetLowering::lowerEH_RETURN(SDValue Op, SelectionDAG &DAG) SDValue Chain = Op.getOperand(0); SDValue Offset = Op.getOperand(1); SDValue Handler = Op.getOperand(2); - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); EVT Ty = IsN64 ? MVT::i64 : MVT::i32; // Store stack offset in V1, store jump target in V0. Glue CopyToReg and @@ -1827,28 +1890,19 @@ SDValue MipsTargetLowering::lowerEH_RETURN(SDValue Op, SelectionDAG &DAG) Chain.getValue(1)); } -// TODO: set SType according to the desired memory barrier behavior. -SDValue -MipsTargetLowering::lowerMEMBARRIER(SDValue Op, SelectionDAG &DAG) const { - unsigned SType = 0; - DebugLoc DL = Op.getDebugLoc(); - return DAG.getNode(MipsISD::Sync, DL, MVT::Other, Op.getOperand(0), - DAG.getConstant(SType, MVT::i32)); -} - SDValue MipsTargetLowering::lowerATOMIC_FENCE(SDValue Op, SelectionDAG &DAG) const { // FIXME: Need pseudo-fence for 'singlethread' fences // FIXME: Set SType for weaker fences where supported/appropriate. unsigned SType = 0; - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); return DAG.getNode(MipsISD::Sync, DL, MVT::Other, Op.getOperand(0), DAG.getConstant(SType, MVT::i32)); } SDValue MipsTargetLowering::lowerShiftLeftParts(SDValue Op, SelectionDAG &DAG) const { - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue Lo = Op.getOperand(0), Hi = Op.getOperand(1); SDValue Shamt = Op.getOperand(2); @@ -1879,7 +1933,7 @@ SDValue MipsTargetLowering::lowerShiftLeftParts(SDValue Op, SDValue MipsTargetLowering::lowerShiftRightParts(SDValue Op, SelectionDAG &DAG, bool IsSRA) const { - DebugLoc DL = Op.getDebugLoc(); + SDLoc DL(Op); SDValue Lo = Op.getOperand(0), Hi = Op.getOperand(1); SDValue Shamt = Op.getOperand(2); @@ -1918,12 +1972,12 @@ SDValue MipsTargetLowering::lowerShiftRightParts(SDValue Op, SelectionDAG &DAG, return DAG.getMergeValues(Ops, 2, DL); } -static SDValue CreateLoadLR(unsigned Opc, SelectionDAG &DAG, LoadSDNode *LD, +static SDValue createLoadLR(unsigned Opc, SelectionDAG &DAG, LoadSDNode *LD, SDValue Chain, SDValue Src, unsigned Offset) { SDValue Ptr = LD->getBasePtr(); EVT VT = LD->getValueType(0), MemVT = LD->getMemoryVT(); EVT BasePtrVT = Ptr.getValueType(); - DebugLoc DL = LD->getDebugLoc(); + SDLoc DL(LD); SDVTList VTList = DAG.getVTList(VT, MVT::Other); if (Offset) @@ -1958,15 +2012,15 @@ SDValue MipsTargetLowering::lowerLOAD(SDValue Op, SelectionDAG &DAG) const { // (set tmp, (ldl (add baseptr, 7), undef)) // (set dst, (ldr baseptr, tmp)) if ((VT == MVT::i64) && (ExtType == ISD::NON_EXTLOAD)) { - SDValue LDL = CreateLoadLR(MipsISD::LDL, DAG, LD, Chain, Undef, + SDValue LDL = createLoadLR(MipsISD::LDL, DAG, LD, Chain, Undef, IsLittle ? 7 : 0); - return CreateLoadLR(MipsISD::LDR, DAG, LD, LDL.getValue(1), LDL, + return createLoadLR(MipsISD::LDR, DAG, LD, LDL.getValue(1), LDL, IsLittle ? 0 : 7); } - SDValue LWL = CreateLoadLR(MipsISD::LWL, DAG, LD, Chain, Undef, + SDValue LWL = createLoadLR(MipsISD::LWL, DAG, LD, Chain, Undef, IsLittle ? 3 : 0); - SDValue LWR = CreateLoadLR(MipsISD::LWR, DAG, LD, LWL.getValue(1), LWL, + SDValue LWR = createLoadLR(MipsISD::LWR, DAG, LD, LWL.getValue(1), LWL, IsLittle ? 0 : 3); // Expand @@ -1989,7 +2043,7 @@ SDValue MipsTargetLowering::lowerLOAD(SDValue Op, SelectionDAG &DAG) const { // (set tmp1, (lwr baseptr, tmp0)) // (set tmp2, (shl tmp1, 32)) // (set dst, (srl tmp2, 32)) - DebugLoc DL = LD->getDebugLoc(); + SDLoc DL(LD); SDValue Const32 = DAG.getConstant(32, MVT::i32); SDValue SLL = DAG.getNode(ISD::SHL, DL, MVT::i64, LWR, Const32); SDValue SRL = DAG.getNode(ISD::SRL, DL, MVT::i64, SLL, Const32); @@ -1997,11 +2051,11 @@ SDValue MipsTargetLowering::lowerLOAD(SDValue Op, SelectionDAG &DAG) const { return DAG.getMergeValues(Ops, 2, DL); } -static SDValue CreateStoreLR(unsigned Opc, SelectionDAG &DAG, StoreSDNode *SD, +static SDValue createStoreLR(unsigned Opc, SelectionDAG &DAG, StoreSDNode *SD, SDValue Chain, unsigned Offset) { SDValue Ptr = SD->getBasePtr(), Value = SD->getValue(); EVT MemVT = SD->getMemoryVT(), BasePtrVT = Ptr.getValueType(); - DebugLoc DL = SD->getDebugLoc(); + SDLoc DL(SD); SDVTList VTList = DAG.getVTList(MVT::Other); if (Offset) @@ -2014,16 +2068,8 @@ static SDValue CreateStoreLR(unsigned Opc, SelectionDAG &DAG, StoreSDNode *SD, } // Expand an unaligned 32 or 64-bit integer store node. -SDValue MipsTargetLowering::lowerSTORE(SDValue Op, SelectionDAG &DAG) const { - StoreSDNode *SD = cast(Op); - EVT MemVT = SD->getMemoryVT(); - - // Return if store is aligned or if MemVT is neither i32 nor i64. - if ((SD->getAlignment() >= MemVT.getSizeInBits() / 8) || - ((MemVT != MVT::i32) && (MemVT != MVT::i64))) - return SDValue(); - - bool IsLittle = Subtarget->isLittle(); +static SDValue lowerUnalignedIntStore(StoreSDNode *SD, SelectionDAG &DAG, + bool IsLittle) { SDValue Value = SD->getValue(), Chain = SD->getChain(); EVT VT = Value.getValueType(); @@ -2034,9 +2080,9 @@ SDValue MipsTargetLowering::lowerSTORE(SDValue Op, SelectionDAG &DAG) const { // (swl val, (add baseptr, 3)) // (swr val, baseptr) if ((VT == MVT::i32) || SD->isTruncatingStore()) { - SDValue SWL = CreateStoreLR(MipsISD::SWL, DAG, SD, Chain, + SDValue SWL = createStoreLR(MipsISD::SWL, DAG, SD, Chain, IsLittle ? 3 : 0); - return CreateStoreLR(MipsISD::SWR, DAG, SD, SWL, IsLittle ? 0 : 3); + return createStoreLR(MipsISD::SWR, DAG, SD, SWL, IsLittle ? 0 : 3); } assert(VT == MVT::i64); @@ -2046,172 +2092,36 @@ SDValue MipsTargetLowering::lowerSTORE(SDValue Op, SelectionDAG &DAG) const { // to // (sdl val, (add baseptr, 7)) // (sdr val, baseptr) - SDValue SDL = CreateStoreLR(MipsISD::SDL, DAG, SD, Chain, IsLittle ? 7 : 0); - return CreateStoreLR(MipsISD::SDR, DAG, SD, SDL, IsLittle ? 0 : 7); + SDValue SDL = createStoreLR(MipsISD::SDL, DAG, SD, Chain, IsLittle ? 7 : 0); + return createStoreLR(MipsISD::SDR, DAG, SD, SDL, IsLittle ? 0 : 7); } -static SDValue initAccumulator(SDValue In, DebugLoc DL, SelectionDAG &DAG) { - SDValue InLo = DAG.getNode(ISD::EXTRACT_ELEMENT, DL, MVT::i32, In, - DAG.getConstant(0, MVT::i32)); - SDValue InHi = DAG.getNode(ISD::EXTRACT_ELEMENT, DL, MVT::i32, In, - DAG.getConstant(1, MVT::i32)); - return DAG.getNode(MipsISD::InsertLOHI, DL, MVT::Untyped, InLo, InHi); -} +// Lower (store (fp_to_sint $fp) $ptr) to (store (TruncIntFP $fp), $ptr). +static SDValue lowerFP_TO_SINT_STORE(StoreSDNode *SD, SelectionDAG &DAG) { + SDValue Val = SD->getValue(); -static SDValue extractLOHI(SDValue Op, DebugLoc DL, SelectionDAG &DAG) { - SDValue Lo = DAG.getNode(MipsISD::ExtractLOHI, DL, MVT::i32, Op, - DAG.getConstant(Mips::sub_lo, MVT::i32)); - SDValue Hi = DAG.getNode(MipsISD::ExtractLOHI, DL, MVT::i32, Op, - DAG.getConstant(Mips::sub_hi, MVT::i32)); - return DAG.getNode(ISD::BUILD_PAIR, DL, MVT::i64, Lo, Hi); -} - -// This function expands mips intrinsic nodes which have 64-bit input operands -// or output values. -// -// out64 = intrinsic-node in64 -// => -// lo = copy (extract-element (in64, 0)) -// hi = copy (extract-element (in64, 1)) -// mips-specific-node -// v0 = copy lo -// v1 = copy hi -// out64 = merge-values (v0, v1) -// -static SDValue lowerDSPIntr(SDValue Op, SelectionDAG &DAG, unsigned Opc) { - DebugLoc DL = Op.getDebugLoc(); - bool HasChainIn = Op->getOperand(0).getValueType() == MVT::Other; - SmallVector Ops; - unsigned OpNo = 0; - - // See if Op has a chain input. - if (HasChainIn) - Ops.push_back(Op->getOperand(OpNo++)); - - // The next operand is the intrinsic opcode. - assert(Op->getOperand(OpNo).getOpcode() == ISD::TargetConstant); - - // See if the next operand has type i64. - SDValue Opnd = Op->getOperand(++OpNo), In64; - - if (Opnd.getValueType() == MVT::i64) - In64 = initAccumulator(Opnd, DL, DAG); - else - Ops.push_back(Opnd); - - // Push the remaining operands. - for (++OpNo ; OpNo < Op->getNumOperands(); ++OpNo) - Ops.push_back(Op->getOperand(OpNo)); - - // Add In64 to the end of the list. - if (In64.getNode()) - Ops.push_back(In64); - - // Scan output. - SmallVector ResTys; + if (Val.getOpcode() != ISD::FP_TO_SINT) + return SDValue(); - for (SDNode::value_iterator I = Op->value_begin(), E = Op->value_end(); - I != E; ++I) - ResTys.push_back((*I == MVT::i64) ? MVT::Untyped : *I); + EVT FPTy = EVT::getFloatingPointVT(Val.getValueSizeInBits()); + SDValue Tr = DAG.getNode(MipsISD::TruncIntFP, SDLoc(Val), FPTy, + Val.getOperand(0)); - // Create node. - SDValue Val = DAG.getNode(Opc, DL, ResTys, &Ops[0], Ops.size()); - SDValue Out = (ResTys[0] == MVT::Untyped) ? extractLOHI(Val, DL, DAG) : Val; + return DAG.getStore(SD->getChain(), SDLoc(SD), Tr, SD->getBasePtr(), + SD->getPointerInfo(), SD->isVolatile(), + SD->isNonTemporal(), SD->getAlignment()); +} - if (!HasChainIn) - return Out; +SDValue MipsTargetLowering::lowerSTORE(SDValue Op, SelectionDAG &DAG) const { + StoreSDNode *SD = cast(Op); + EVT MemVT = SD->getMemoryVT(); - assert(Val->getValueType(1) == MVT::Other); - SDValue Vals[] = { Out, SDValue(Val.getNode(), 1) }; - return DAG.getMergeValues(Vals, 2, DL); -} + // Lower unaligned integer stores. + if ((SD->getAlignment() < MemVT.getSizeInBits() / 8) && + ((MemVT == MVT::i32) || (MemVT == MVT::i64))) + return lowerUnalignedIntStore(SD, DAG, Subtarget->isLittle()); -SDValue MipsTargetLowering::lowerINTRINSIC_WO_CHAIN(SDValue Op, - SelectionDAG &DAG) const { - switch (cast(Op->getOperand(0))->getZExtValue()) { - default: - return SDValue(); - case Intrinsic::mips_shilo: - return lowerDSPIntr(Op, DAG, MipsISD::SHILO); - case Intrinsic::mips_dpau_h_qbl: - return lowerDSPIntr(Op, DAG, MipsISD::DPAU_H_QBL); - case Intrinsic::mips_dpau_h_qbr: - return lowerDSPIntr(Op, DAG, MipsISD::DPAU_H_QBR); - case Intrinsic::mips_dpsu_h_qbl: - return lowerDSPIntr(Op, DAG, MipsISD::DPSU_H_QBL); - case Intrinsic::mips_dpsu_h_qbr: - return lowerDSPIntr(Op, DAG, MipsISD::DPSU_H_QBR); - case Intrinsic::mips_dpa_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPA_W_PH); - case Intrinsic::mips_dps_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPS_W_PH); - case Intrinsic::mips_dpax_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPAX_W_PH); - case Intrinsic::mips_dpsx_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPSX_W_PH); - case Intrinsic::mips_mulsa_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::MULSA_W_PH); - case Intrinsic::mips_mult: - return lowerDSPIntr(Op, DAG, MipsISD::Mult); - case Intrinsic::mips_multu: - return lowerDSPIntr(Op, DAG, MipsISD::Multu); - case Intrinsic::mips_madd: - return lowerDSPIntr(Op, DAG, MipsISD::MAdd); - case Intrinsic::mips_maddu: - return lowerDSPIntr(Op, DAG, MipsISD::MAddu); - case Intrinsic::mips_msub: - return lowerDSPIntr(Op, DAG, MipsISD::MSub); - case Intrinsic::mips_msubu: - return lowerDSPIntr(Op, DAG, MipsISD::MSubu); - } -} - -SDValue MipsTargetLowering::lowerINTRINSIC_W_CHAIN(SDValue Op, - SelectionDAG &DAG) const { - switch (cast(Op->getOperand(1))->getZExtValue()) { - default: - return SDValue(); - case Intrinsic::mips_extp: - return lowerDSPIntr(Op, DAG, MipsISD::EXTP); - case Intrinsic::mips_extpdp: - return lowerDSPIntr(Op, DAG, MipsISD::EXTPDP); - case Intrinsic::mips_extr_w: - return lowerDSPIntr(Op, DAG, MipsISD::EXTR_W); - case Intrinsic::mips_extr_r_w: - return lowerDSPIntr(Op, DAG, MipsISD::EXTR_R_W); - case Intrinsic::mips_extr_rs_w: - return lowerDSPIntr(Op, DAG, MipsISD::EXTR_RS_W); - case Intrinsic::mips_extr_s_h: - return lowerDSPIntr(Op, DAG, MipsISD::EXTR_S_H); - case Intrinsic::mips_mthlip: - return lowerDSPIntr(Op, DAG, MipsISD::MTHLIP); - case Intrinsic::mips_mulsaq_s_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::MULSAQ_S_W_PH); - case Intrinsic::mips_maq_s_w_phl: - return lowerDSPIntr(Op, DAG, MipsISD::MAQ_S_W_PHL); - case Intrinsic::mips_maq_s_w_phr: - return lowerDSPIntr(Op, DAG, MipsISD::MAQ_S_W_PHR); - case Intrinsic::mips_maq_sa_w_phl: - return lowerDSPIntr(Op, DAG, MipsISD::MAQ_SA_W_PHL); - case Intrinsic::mips_maq_sa_w_phr: - return lowerDSPIntr(Op, DAG, MipsISD::MAQ_SA_W_PHR); - case Intrinsic::mips_dpaq_s_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPAQ_S_W_PH); - case Intrinsic::mips_dpsq_s_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPSQ_S_W_PH); - case Intrinsic::mips_dpaq_sa_l_w: - return lowerDSPIntr(Op, DAG, MipsISD::DPAQ_SA_L_W); - case Intrinsic::mips_dpsq_sa_l_w: - return lowerDSPIntr(Op, DAG, MipsISD::DPSQ_SA_L_W); - case Intrinsic::mips_dpaqx_s_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPAQX_S_W_PH); - case Intrinsic::mips_dpaqx_sa_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPAQX_SA_W_PH); - case Intrinsic::mips_dpsqx_s_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPSQX_S_W_PH); - case Intrinsic::mips_dpsqx_sa_w_ph: - return lowerDSPIntr(Op, DAG, MipsISD::DPSQX_SA_W_PH); - } + return lowerFP_TO_SINT_STORE(SD, DAG); } SDValue MipsTargetLowering::lowerADD(SDValue Op, SelectionDAG &DAG) const { @@ -2231,10 +2141,18 @@ SDValue MipsTargetLowering::lowerADD(SDValue Op, SelectionDAG &DAG) const { EVT ValTy = Op->getValueType(0); int FI = MFI->CreateFixedObject(Op.getValueSizeInBits() / 8, 0, false); SDValue InArgsAddr = DAG.getFrameIndex(FI, ValTy); - return DAG.getNode(ISD::ADD, Op->getDebugLoc(), ValTy, InArgsAddr, + return DAG.getNode(ISD::ADD, SDLoc(Op), ValTy, InArgsAddr, DAG.getConstant(0, ValTy)); } +SDValue MipsTargetLowering::lowerFP_TO_SINT(SDValue Op, + SelectionDAG &DAG) const { + EVT FPTy = EVT::getFloatingPointVT(Op.getValueSizeInBits()); + SDValue Trunc = DAG.getNode(MipsISD::TruncIntFP, SDLoc(Op), FPTy, + Op.getOperand(0)); + return DAG.getNode(ISD::BITCAST, SDLoc(Op), Op.getValueType(), Trunc); +} + //===----------------------------------------------------------------------===// // Calling Convention Implementation //===----------------------------------------------------------------------===// @@ -2254,21 +2172,14 @@ SDValue MipsTargetLowering::lowerADD(SDValue Op, SelectionDAG &DAG) const { // For vararg functions, all arguments are passed in A0, A1, A2, A3 and stack. //===----------------------------------------------------------------------===// -static bool CC_MipsO32(unsigned ValNo, MVT ValVT, - MVT LocVT, CCValAssign::LocInfo LocInfo, - ISD::ArgFlagsTy ArgFlags, CCState &State) { +static bool CC_MipsO32(unsigned ValNo, MVT ValVT, MVT LocVT, + CCValAssign::LocInfo LocInfo, ISD::ArgFlagsTy ArgFlags, + CCState &State, const uint16_t *F64Regs) { - static const unsigned IntRegsSize=4, FloatRegsSize=2; + static const unsigned IntRegsSize = 4, FloatRegsSize = 2; - static const uint16_t IntRegs[] = { - Mips::A0, Mips::A1, Mips::A2, Mips::A3 - }; - static const uint16_t F32Regs[] = { - Mips::F12, Mips::F14 - }; - static const uint16_t F64Regs[] = { - Mips::D6, Mips::D7 - }; + static const uint16_t IntRegs[] = { Mips::A0, Mips::A1, Mips::A2, Mips::A3 }; + static const uint16_t F32Regs[] = { Mips::F12, Mips::F14 }; // Do not process byval args here. if (ArgFlags.isByVal()) @@ -2337,14 +2248,28 @@ static bool CC_MipsO32(unsigned ValNo, MVT ValVT, return false; } +static bool CC_MipsO32_FP32(unsigned ValNo, MVT ValVT, + MVT LocVT, CCValAssign::LocInfo LocInfo, + ISD::ArgFlagsTy ArgFlags, CCState &State) { + static const uint16_t F64Regs[] = { Mips::D6, Mips::D7 }; + + return CC_MipsO32(ValNo, ValVT, LocVT, LocInfo, ArgFlags, State, F64Regs); +} + +static bool CC_MipsO32_FP64(unsigned ValNo, MVT ValVT, + MVT LocVT, CCValAssign::LocInfo LocInfo, + ISD::ArgFlagsTy ArgFlags, CCState &State) { + static const uint16_t F64Regs[] = { Mips::D12_64, Mips::D14_64 }; + + return CC_MipsO32(ValNo, ValVT, LocVT, LocInfo, ArgFlags, State, F64Regs); +} + #include "MipsGenCallingConv.inc" //===----------------------------------------------------------------------===// // Call Calling Convention Implementation //===----------------------------------------------------------------------===// -static const unsigned O32IntRegsSize = 4; - // Return next O32 integer argument register. static unsigned getNextIntArgReg(unsigned Reg) { assert((Reg == Mips::A0) || (Reg == Mips::A2)); @@ -2353,7 +2278,7 @@ static unsigned getNextIntArgReg(unsigned Reg) { SDValue MipsTargetLowering::passArgOnStack(SDValue StackPtr, unsigned Offset, - SDValue Chain, SDValue Arg, DebugLoc DL, + SDValue Chain, SDValue Arg, SDLoc DL, bool IsTailCall, SelectionDAG &DAG) const { if (!IsTailCall) { SDValue PtrOff = DAG.getNode(ISD::ADD, DL, getPointerTy(), StackPtr, @@ -2407,6 +2332,15 @@ getOpndList(SmallVectorImpl &Ops, const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo(); const uint32_t *Mask = TRI->getCallPreservedMask(CLI.CallConv); assert(Mask && "Missing call preserved mask for calling convention"); + if (Subtarget->inMips16HardFloat()) { + if (GlobalAddressSDNode *G = dyn_cast(CLI.Callee)) { + llvm::StringRef Sym = G->getGlobal()->getName(); + Function *F = G->getGlobal()->getParent()->getFunction(Sym); + if (F && F->hasFnAttribute("__Mips16RetHelper")) { + Mask = MipsRegisterInfo::getMips16RetHelperMask(); + } + } + } Ops.push_back(CLI.DAG.getRegisterMask(Mask)); if (InFlag.getNode()) @@ -2419,10 +2353,10 @@ SDValue MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, SmallVectorImpl &InVals) const { SelectionDAG &DAG = CLI.DAG; - DebugLoc &DL = CLI.DL; - SmallVector &Outs = CLI.Outs; - SmallVector &OutVals = CLI.OutVals; - SmallVector &Ins = CLI.Ins; + SDLoc DL = CLI.DL; + SmallVectorImpl &Outs = CLI.Outs; + SmallVectorImpl &OutVals = CLI.OutVals; + SmallVectorImpl &Ins = CLI.Ins; SDValue Chain = CLI.Chain; SDValue Callee = CLI.Callee; bool &IsTailCall = CLI.IsTailCall; @@ -2432,16 +2366,20 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, MachineFunction &MF = DAG.getMachineFunction(); MachineFrameInfo *MFI = MF.getFrameInfo(); const TargetFrameLowering *TFL = MF.getTarget().getFrameLowering(); + MipsFunctionInfo *FuncInfo = MF.getInfo(); bool IsPIC = getTargetMachine().getRelocationModel() == Reloc::PIC_; // Analyze operands of the call, assigning locations to each operand. SmallVector ArgLocs; CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), getTargetMachine(), ArgLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, IsO32, CCInfo); + MipsCC::SpecialCallingConvType SpecialCallingConv = + getSpecialCallingConv(Callee); + MipsCC MipsCCInfo(CallConv, IsO32, Subtarget->isFP64bit(), CCInfo, + SpecialCallingConv); MipsCCInfo.analyzeCallOperands(Outs, IsVarArg, - getTargetMachine().Options.UseSoftFloat, + Subtarget->mipsSEUsesSoftFloat(), Callee.getNode(), CLI.Args); // Get a count of how many bytes are to be pushed on the stack. @@ -2464,7 +2402,7 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, SDValue NextStackOffsetVal = DAG.getIntPtrConstant(NextStackOffset, true); if (!IsTailCall) - Chain = DAG.getCALLSEQ_START(Chain, NextStackOffsetVal); + Chain = DAG.getCALLSEQ_START(Chain, NextStackOffsetVal, DL); SDValue StackPtr = DAG.getCopyFromReg(Chain, DL, IsN64 ? Mips::SP_64 : Mips::SP, @@ -2558,32 +2496,40 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, bool IsPICCall = (IsN64 || IsPIC); // true if calls are translated to jalr $25 bool GlobalOrExternal = false, InternalLinkage = false; SDValue CalleeLo; + EVT Ty = Callee.getValueType(); if (GlobalAddressSDNode *G = dyn_cast(Callee)) { if (IsPICCall) { - InternalLinkage = G->getGlobal()->hasInternalLinkage(); + const GlobalValue *Val = G->getGlobal(); + InternalLinkage = Val->hasInternalLinkage(); if (InternalLinkage) - Callee = getAddrLocal(Callee, DAG, HasMips64); + Callee = getAddrLocal(G, Ty, DAG, HasMips64); else if (LargeGOT) - Callee = getAddrGlobalLargeGOT(Callee, DAG, MipsII::MO_CALL_HI16, - MipsII::MO_CALL_LO16); + Callee = getAddrGlobalLargeGOT(G, Ty, DAG, MipsII::MO_CALL_HI16, + MipsII::MO_CALL_LO16, Chain, + FuncInfo->callPtrInfo(Val)); else - Callee = getAddrGlobal(Callee, DAG, MipsII::MO_GOT_CALL); + Callee = getAddrGlobal(G, Ty, DAG, MipsII::MO_GOT_CALL, Chain, + FuncInfo->callPtrInfo(Val)); } else Callee = DAG.getTargetGlobalAddress(G->getGlobal(), DL, getPointerTy(), 0, MipsII::MO_NO_FLAG); GlobalOrExternal = true; } else if (ExternalSymbolSDNode *S = dyn_cast(Callee)) { + const char *Sym = S->getSymbol(); + if (!IsN64 && !IsPIC) // !N64 && static - Callee = DAG.getTargetExternalSymbol(S->getSymbol(), getPointerTy(), + Callee = DAG.getTargetExternalSymbol(Sym, getPointerTy(), MipsII::MO_NO_FLAG); else if (LargeGOT) - Callee = getAddrGlobalLargeGOT(Callee, DAG, MipsII::MO_CALL_HI16, - MipsII::MO_CALL_LO16); + Callee = getAddrGlobalLargeGOT(S, Ty, DAG, MipsII::MO_CALL_HI16, + MipsII::MO_CALL_LO16, Chain, + FuncInfo->callPtrInfo(Sym)); else // N64 || PIC - Callee = getAddrGlobal(Callee, DAG, MipsII::MO_GOT_CALL); + Callee = getAddrGlobal(S, Ty, DAG, MipsII::MO_GOT_CALL, Chain, + FuncInfo->callPtrInfo(Sym)); GlobalOrExternal = true; } @@ -2602,7 +2548,7 @@ MipsTargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI, // Create the CALLSEQ_END node. Chain = DAG.getCALLSEQ_END(Chain, NextStackOffsetVal, - DAG.getIntPtrConstant(0, true), InFlag); + DAG.getIntPtrConstant(0, true), InFlag, DL); InFlag = Chain.getValue(1); // Handle result values, copying them out of physregs into vregs that we @@ -2617,7 +2563,7 @@ SDValue MipsTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag, CallingConv::ID CallConv, bool IsVarArg, const SmallVectorImpl &Ins, - DebugLoc DL, SelectionDAG &DAG, + SDLoc DL, SelectionDAG &DAG, SmallVectorImpl &InVals, const SDNode *CallNode, const Type *RetTy) const { @@ -2625,9 +2571,9 @@ MipsTargetLowering::LowerCallResult(SDValue Chain, SDValue InFlag, SmallVector RVLocs; CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), getTargetMachine(), RVLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, IsO32, CCInfo); + MipsCC MipsCCInfo(CallConv, IsO32, Subtarget->isFP64bit(), CCInfo); - MipsCCInfo.analyzeCallResult(Ins, getTargetMachine().Options.UseSoftFloat, + MipsCCInfo.analyzeCallResult(Ins, Subtarget->mipsSEUsesSoftFloat(), CallNode, RetTy); // Copy all of the result registers out of their specified physreg. @@ -2656,7 +2602,7 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, CallingConv::ID CallConv, bool IsVarArg, const SmallVectorImpl &Ins, - DebugLoc DL, SelectionDAG &DAG, + SDLoc DL, SelectionDAG &DAG, SmallVectorImpl &InVals) const { MachineFunction &MF = DAG.getMachineFunction(); @@ -2672,10 +2618,10 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, SmallVector ArgLocs; CCState CCInfo(CallConv, IsVarArg, DAG.getMachineFunction(), getTargetMachine(), ArgLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, IsO32, CCInfo); + MipsCC MipsCCInfo(CallConv, IsO32, Subtarget->isFP64bit(), CCInfo); Function::const_arg_iterator FuncArg = DAG.getMachineFunction().getFunction()->arg_begin(); - bool UseSoftFloat = getTargetMachine().Options.UseSoftFloat; + bool UseSoftFloat = Subtarget->mipsSEUsesSoftFloat(); MipsCCInfo.analyzeFormalArguments(Ins, UseSoftFloat, FuncArg); MipsFI->setFormalArgInfo(CCInfo.getNextStackOffset(), @@ -2704,21 +2650,9 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, // Arguments stored on registers if (IsRegLoc) { - EVT RegVT = VA.getLocVT(); + MVT RegVT = VA.getLocVT(); unsigned ArgReg = VA.getLocReg(); - const TargetRegisterClass *RC; - - if (RegVT == MVT::i32) - RC = Subtarget->inMips16Mode()? &Mips::CPU16RegsRegClass : - &Mips::CPURegsRegClass; - else if (RegVT == MVT::i64) - RC = &Mips::CPU64RegsRegClass; - else if (RegVT == MVT::f32) - RC = &Mips::FGR32RegClass; - else if (RegVT == MVT::f64) - RC = HasMips64 ? &Mips::FGR64RegClass : &Mips::AFGR64RegClass; - else - llvm_unreachable("RegVT not supported by FormalArguments Lowering"); + const TargetRegisterClass *RC = getRegClassFor(RegVT); // Transform the arguments stored on // physical registers into virtual ones @@ -2768,9 +2702,11 @@ MipsTargetLowering::LowerFormalArguments(SDValue Chain, // Create load nodes to retrieve arguments from the stack SDValue FIN = DAG.getFrameIndex(FI, getPointerTy()); - InVals.push_back(DAG.getLoad(ValVT, DL, Chain, FIN, - MachinePointerInfo::getFixedStack(FI), - false, false, false, 0)); + SDValue Load = DAG.getLoad(ValVT, DL, Chain, FIN, + MachinePointerInfo::getFixedStack(FI), + false, false, false, 0); + InVals.push_back(Load); + OutChains.push_back(Load.getValue(1)); } } @@ -2822,7 +2758,7 @@ MipsTargetLowering::LowerReturn(SDValue Chain, CallingConv::ID CallConv, bool IsVarArg, const SmallVectorImpl &Outs, const SmallVectorImpl &OutVals, - DebugLoc DL, SelectionDAG &DAG) const { + SDLoc DL, SelectionDAG &DAG) const { // CCValAssign - represent the assignment of // the return value to a location SmallVector RVLocs; @@ -2831,10 +2767,10 @@ MipsTargetLowering::LowerReturn(SDValue Chain, // CCState - Info about the registers and stack slot. CCState CCInfo(CallConv, IsVarArg, MF, getTargetMachine(), RVLocs, *DAG.getContext()); - MipsCC MipsCCInfo(CallConv, IsO32, CCInfo); + MipsCC MipsCCInfo(CallConv, IsO32, Subtarget->isFP64bit(), CCInfo); // Analyze return values. - MipsCCInfo.analyzeReturn(Outs, getTargetMachine().Options.UseSoftFloat, + MipsCCInfo.analyzeReturn(Outs, Subtarget->mipsSEUsesSoftFloat(), MF.getFunction()->getReturnType()); SDValue Flag; @@ -2893,7 +2829,7 @@ MipsTargetLowering::LowerReturn(SDValue Chain, MipsTargetLowering::ConstraintType MipsTargetLowering:: getConstraintType(const std::string &Constraint) const { - // Mips specific constrainy + // Mips specific constraints // GCC config/mips/constraints.md // // 'd' : An address register. Equivalent to r @@ -2944,16 +2880,19 @@ MipsTargetLowering::getSingleConstraintMatchWeight( if (type->isIntegerTy()) weight = CW_Register; break; - case 'f': - if (type->isFloatTy()) + case 'f': // FPU or MSA register + if (Subtarget->hasMSA() && type->isVectorTy() && + cast(type)->getBitWidth() == 128) + weight = CW_Register; + else if (type->isFloatTy()) weight = CW_Register; break; case 'c': // $25 for indirect jumps case 'l': // lo register case 'x': // hilo register pair - if (type->isIntegerTy()) + if (type->isIntegerTy()) weight = CW_SpecificReg; - break; + break; case 'I': // signed 16 bit immediate case 'J': // integer zero case 'K': // unsigned 16 bit immediate @@ -2971,11 +2910,109 @@ MipsTargetLowering::getSingleConstraintMatchWeight( return weight; } +/// This is a helper function to parse a physical register string and split it +/// into non-numeric and numeric parts (Prefix and Reg). The first boolean flag +/// that is returned indicates whether parsing was successful. The second flag +/// is true if the numeric part exists. +static std::pair +parsePhysicalReg(const StringRef &C, std::string &Prefix, + unsigned long long &Reg) { + if (C.front() != '{' || C.back() != '}') + return std::make_pair(false, false); + + // Search for the first numeric character. + StringRef::const_iterator I, B = C.begin() + 1, E = C.end() - 1; + I = std::find_if(B, E, std::ptr_fun(isdigit)); + + Prefix.assign(B, I - B); + + // The second flag is set to false if no numeric characters were found. + if (I == E) + return std::make_pair(true, false); + + // Parse the numeric characters. + return std::make_pair(!getAsUnsignedInteger(StringRef(I, E - I), 10, Reg), + true); +} + +std::pair MipsTargetLowering:: +parseRegForInlineAsmConstraint(const StringRef &C, MVT VT) const { + const TargetRegisterInfo *TRI = getTargetMachine().getRegisterInfo(); + const TargetRegisterClass *RC; + std::string Prefix; + unsigned long long Reg; + + std::pair R = parsePhysicalReg(C, Prefix, Reg); + + if (!R.first) + return std::make_pair((unsigned)0, (const TargetRegisterClass*)0); + + if ((Prefix == "hi" || Prefix == "lo")) { // Parse hi/lo. + // No numeric characters follow "hi" or "lo". + if (R.second) + return std::make_pair((unsigned)0, (const TargetRegisterClass*)0); + + RC = TRI->getRegClass(Prefix == "hi" ? + Mips::HI32RegClassID : Mips::LO32RegClassID); + return std::make_pair(*(RC->begin()), RC); + } else if (Prefix.compare(0, 4, "$msa") == 0) { + // Parse $msa(ir|csr|access|save|modify|request|map|unmap) + + // No numeric characters follow the name. + if (R.second) + return std::make_pair((unsigned)0, (const TargetRegisterClass *)0); + + Reg = StringSwitch(Prefix) + .Case("$msair", Mips::MSAIR) + .Case("$msacsr", Mips::MSACSR) + .Case("$msaaccess", Mips::MSAAccess) + .Case("$msasave", Mips::MSASave) + .Case("$msamodify", Mips::MSAModify) + .Case("$msarequest", Mips::MSARequest) + .Case("$msamap", Mips::MSAMap) + .Case("$msaunmap", Mips::MSAUnmap) + .Default(0); + + if (!Reg) + return std::make_pair((unsigned)0, (const TargetRegisterClass *)0); + + RC = TRI->getRegClass(Mips::MSACtrlRegClassID); + return std::make_pair(Reg, RC); + } + + if (!R.second) + return std::make_pair((unsigned)0, (const TargetRegisterClass*)0); + + if (Prefix == "$f") { // Parse $f0-$f31. + // If the size of FP registers is 64-bit or Reg is an even number, select + // the 64-bit register class. Otherwise, select the 32-bit register class. + if (VT == MVT::Other) + VT = (Subtarget->isFP64bit() || !(Reg % 2)) ? MVT::f64 : MVT::f32; + + RC = getRegClassFor(VT); + + if (RC == &Mips::AFGR64RegClass) { + assert(Reg % 2 == 0); + Reg >>= 1; + } + } else if (Prefix == "$fcc") // Parse $fcc0-$fcc7. + RC = TRI->getRegClass(Mips::FCCRegClassID); + else if (Prefix == "$w") { // Parse $w0-$w31. + RC = getRegClassFor((VT == MVT::Other) ? MVT::v16i8 : VT); + } else { // Parse $0-$31. + assert(Prefix == "$"); + RC = getRegClassFor((VT == MVT::Other) ? MVT::i32 : VT); + } + + assert(Reg < RC->getNumRegs()); + return std::make_pair(*(RC->begin() + Reg), RC); +} + /// Given a register class constraint, like 'r', if this corresponds directly /// to an LLVM register class, return a register of 0 and the register class /// pointer. std::pair MipsTargetLowering:: -getRegForInlineAsmConstraint(const std::string &Constraint, EVT VT) const +getRegForInlineAsmConstraint(const std::string &Constraint, MVT VT) const { if (Constraint.size() == 1) { switch (Constraint[0]) { @@ -2985,18 +3022,26 @@ getRegForInlineAsmConstraint(const std::string &Constraint, EVT VT) const if (VT == MVT::i32 || VT == MVT::i16 || VT == MVT::i8) { if (Subtarget->inMips16Mode()) return std::make_pair(0U, &Mips::CPU16RegsRegClass); - return std::make_pair(0U, &Mips::CPURegsRegClass); + return std::make_pair(0U, &Mips::GPR32RegClass); } if (VT == MVT::i64 && !HasMips64) - return std::make_pair(0U, &Mips::CPURegsRegClass); + return std::make_pair(0U, &Mips::GPR32RegClass); if (VT == MVT::i64 && HasMips64) - return std::make_pair(0U, &Mips::CPU64RegsRegClass); + return std::make_pair(0U, &Mips::GPR64RegClass); // This will generate an error message return std::make_pair(0u, static_cast(0)); - case 'f': - if (VT == MVT::f32) + case 'f': // FPU or MSA register + if (VT == MVT::v16i8) + return std::make_pair(0U, &Mips::MSA128BRegClass); + else if (VT == MVT::v8i16 || VT == MVT::v8f16) + return std::make_pair(0U, &Mips::MSA128HRegClass); + else if (VT == MVT::v4i32 || VT == MVT::v4f32) + return std::make_pair(0U, &Mips::MSA128WRegClass); + else if (VT == MVT::v2i64 || VT == MVT::v2f64) + return std::make_pair(0U, &Mips::MSA128DRegClass); + else if (VT == MVT::f32) return std::make_pair(0U, &Mips::FGR32RegClass); - if ((VT == MVT::f64) && (!Subtarget->isSingleFloat())) { + else if ((VT == MVT::f64) && (!Subtarget->isSingleFloat())) { if (Subtarget->isFP64bit()) return std::make_pair(0U, &Mips::FGR64RegClass); return std::make_pair(0U, &Mips::AFGR64RegClass); @@ -3004,19 +3049,26 @@ getRegForInlineAsmConstraint(const std::string &Constraint, EVT VT) const break; case 'c': // register suitable for indirect jump if (VT == MVT::i32) - return std::make_pair((unsigned)Mips::T9, &Mips::CPURegsRegClass); + return std::make_pair((unsigned)Mips::T9, &Mips::GPR32RegClass); assert(VT == MVT::i64 && "Unexpected type."); - return std::make_pair((unsigned)Mips::T9_64, &Mips::CPU64RegsRegClass); + return std::make_pair((unsigned)Mips::T9_64, &Mips::GPR64RegClass); case 'l': // register suitable for indirect jump if (VT == MVT::i32) - return std::make_pair((unsigned)Mips::LO, &Mips::HILORegClass); - return std::make_pair((unsigned)Mips::LO64, &Mips::HILO64RegClass); + return std::make_pair((unsigned)Mips::LO0, &Mips::LO32RegClass); + return std::make_pair((unsigned)Mips::LO0_64, &Mips::LO64RegClass); case 'x': // register suitable for indirect jump // Fixme: Not triggering the use of both hi and low // This will generate an error message return std::make_pair(0u, static_cast(0)); } } + + std::pair R; + R = parseRegForInlineAsmConstraint(Constraint, VT); + + if (R.second) + return R; + return TargetLowering::getRegForInlineAsmConstraint(Constraint, VT); } @@ -3115,8 +3167,8 @@ void MipsTargetLowering::LowerAsmOperandForConstraint(SDValue Op, TargetLowering::LowerAsmOperandForConstraint(Op, Constraint, Ops, DAG); } -bool -MipsTargetLowering::isLegalAddressingMode(const AddrMode &AM, Type *Ty) const { +bool MipsTargetLowering::isLegalAddressingMode(const AddrMode &AM, + Type *Ty) const { // No global is ever allowed as a base. if (AM.BaseGV) return false; @@ -3180,13 +3232,13 @@ static bool isF128SoftLibCall(const char *CallSym) { "log10l", "log2l", "logl", "nearbyintl", "powl", "rintl", "sinl", "sqrtl", "truncl"}; - const char * const *End = LibCalls + array_lengthof(LibCalls); + const char *const *End = LibCalls + array_lengthof(LibCalls); // Check that LibCalls is sorted alphabetically. MipsTargetLowering::LTStr Comp; #ifndef NDEBUG - for (const char * const *I = LibCalls; I < End - 1; ++I) + for (const char *const *I = LibCalls; I < End - 1; ++I) assert(Comp(*I, *(I + 1))); #endif @@ -3207,13 +3259,32 @@ static bool originalTypeIsF128(const Type *Ty, const SDNode *CallNode) { return (ES && Ty->isIntegerTy(128) && isF128SoftLibCall(ES->getSymbol())); } -MipsTargetLowering::MipsCC::MipsCC(CallingConv::ID CC, bool IsO32_, - CCState &Info) - : CCInfo(Info), CallConv(CC), IsO32(IsO32_) { +MipsTargetLowering::MipsCC::SpecialCallingConvType + MipsTargetLowering::getSpecialCallingConv(SDValue Callee) const { + MipsCC::SpecialCallingConvType SpecialCallingConv = + MipsCC::NoSpecialCallingConv;; + if (Subtarget->inMips16HardFloat()) { + if (GlobalAddressSDNode *G = dyn_cast(Callee)) { + llvm::StringRef Sym = G->getGlobal()->getName(); + Function *F = G->getGlobal()->getParent()->getFunction(Sym); + if (F && F->hasFnAttribute("__Mips16RetHelper")) { + SpecialCallingConv = MipsCC::Mips16RetHelperConv; + } + } + } + return SpecialCallingConv; +} + +MipsTargetLowering::MipsCC::MipsCC( + CallingConv::ID CC, bool IsO32_, bool IsFP64_, CCState &Info, + MipsCC::SpecialCallingConvType SpecialCallingConv_) + : CCInfo(Info), CallConv(CC), IsO32(IsO32_), IsFP64(IsFP64_), + SpecialCallingConv(SpecialCallingConv_){ // Pre-allocate reserved argument area. CCInfo.AllocateStack(reservedArgArea(), 1); } + void MipsTargetLowering::MipsCC:: analyzeCallOperands(const SmallVectorImpl &Args, bool IsVarArg, bool IsSoftFloat, const SDNode *CallNode, @@ -3321,11 +3392,10 @@ analyzeReturn(const SmallVectorImpl &Outs, bool IsSoftFloat, analyzeReturn(Outs, IsSoftFloat, 0, RetTy); } -void -MipsTargetLowering::MipsCC::handleByValArg(unsigned ValNo, MVT ValVT, - MVT LocVT, - CCValAssign::LocInfo LocInfo, - ISD::ArgFlagsTy ArgFlags) { +void MipsTargetLowering::MipsCC::handleByValArg(unsigned ValNo, MVT ValVT, + MVT LocVT, + CCValAssign::LocInfo LocInfo, + ISD::ArgFlagsTy ArgFlags) { assert(ArgFlags.getByValSize() && "Byval argument's size shouldn't be 0."); struct ByValArgInfo ByVal; @@ -3361,11 +3431,13 @@ llvm::CCAssignFn *MipsTargetLowering::MipsCC::fixedArgFn() const { if (CallConv == CallingConv::Fast) return CC_Mips_FastCC; - return IsO32 ? CC_MipsO32 : CC_MipsN; + if (SpecialCallingConv == Mips16RetHelperConv) + return CC_Mips16RetHelper; + return IsO32 ? (IsFP64 ? CC_MipsO32_FP64 : CC_MipsO32_FP32) : CC_MipsN; } llvm::CCAssignFn *MipsTargetLowering::MipsCC::varArgFn() const { - return IsO32 ? CC_MipsO32 : CC_MipsN_VarArg; + return IsO32 ? (IsFP64 ? CC_MipsO32_FP64 : CC_MipsO32_FP32) : CC_MipsN_VarArg; } const uint16_t *MipsTargetLowering::MipsCC::shadowRegs() const { @@ -3411,7 +3483,7 @@ MVT MipsTargetLowering::MipsCC::getRegVT(MVT VT, const Type *OrigTy, } void MipsTargetLowering:: -copyByValRegs(SDValue Chain, DebugLoc DL, std::vector &OutChains, +copyByValRegs(SDValue Chain, SDLoc DL, std::vector &OutChains, SelectionDAG &DAG, const ISD::ArgFlagsTy &Flags, SmallVectorImpl &InVals, const Argument *FuncArg, const MipsCC &CC, const ByValArgInfo &ByVal) const { @@ -3455,9 +3527,9 @@ copyByValRegs(SDValue Chain, DebugLoc DL, std::vector &OutChains, // Copy byVal arg to registers and stack. void MipsTargetLowering:: -passByValArg(SDValue Chain, DebugLoc DL, +passByValArg(SDValue Chain, SDLoc DL, std::deque< std::pair > &RegsToPass, - SmallVector &MemOpChains, SDValue StackPtr, + SmallVectorImpl &MemOpChains, SDValue StackPtr, MachineFrameInfo *MFI, SelectionDAG &DAG, SDValue Arg, const MipsCC &CC, const ByValArgInfo &ByVal, const ISD::ArgFlagsTy &Flags, bool isLittle) const { @@ -3543,17 +3615,15 @@ passByValArg(SDValue Chain, DebugLoc DL, DAG.getConstant(Offset, PtrTy)); SDValue Dst = DAG.getNode(ISD::ADD, DL, PtrTy, StackPtr, DAG.getIntPtrConstant(ByVal.Address)); - Chain = DAG.getMemcpy(Chain, DL, Dst, Src, - DAG.getConstant(MemCpySize, PtrTy), Alignment, - /*isVolatile=*/false, /*AlwaysInline=*/false, + Chain = DAG.getMemcpy(Chain, DL, Dst, Src, DAG.getConstant(MemCpySize, PtrTy), + Alignment, /*isVolatile=*/false, /*AlwaysInline=*/false, MachinePointerInfo(0), MachinePointerInfo(0)); MemOpChains.push_back(Chain); } -void -MipsTargetLowering::writeVarArgRegs(std::vector &OutChains, - const MipsCC &CC, SDValue Chain, - DebugLoc DL, SelectionDAG &DAG) const { +void MipsTargetLowering::writeVarArgRegs(std::vector &OutChains, + const MipsCC &CC, SDValue Chain, + SDLoc DL, SelectionDAG &DAG) const { unsigned NumRegs = CC.numIntArgRegs(); const uint16_t *ArgRegs = CC.intArgRegs(); const CCState &CCInfo = CC.getCCInfo(); @@ -3571,8 +3641,7 @@ MipsTargetLowering::writeVarArgRegs(std::vector &OutChains, if (NumRegs == Idx) VaArgOffset = RoundUpToAlignment(CCInfo.getNextStackOffset(), RegSize); else - VaArgOffset = - (int)CC.reservedArgArea() - (int)(RegSize * (NumRegs - Idx)); + VaArgOffset = (int)CC.reservedArgArea() - (int)(RegSize * (NumRegs - Idx)); // Record the frame index of the first variable argument // which is a value necessary to VASTART.