X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FX86%2FX86ISelLowering.h;h=edcdceb39a592c353af441aded70b2e20225051d;hb=1292c226458b68a119d3a387a0527f453b2065c2;hp=46fa3cefdc34cb8e2d789278cc87976bc1dc79ec;hpb=86afec7730e719600952bc9019f25e85289032a9;p=oota-llvm.git diff --git a/lib/Target/X86/X86ISelLowering.h b/lib/Target/X86/X86ISelLowering.h index 46fa3cefdc3..edcdceb39a5 100644 --- a/lib/Target/X86/X86ISelLowering.h +++ b/lib/Target/X86/X86ISelLowering.h @@ -57,35 +57,6 @@ namespace llvm { /// corresponds to X86::PSRLDQ. FSRL, - /// FILD, FILD_FLAG - This instruction implements SINT_TO_FP with the - /// integer source in memory and FP reg result. This corresponds to the - /// X86::FILD*m instructions. It has three inputs (token chain, address, - /// and source type) and two outputs (FP value and token chain). FILD_FLAG - /// also produces a flag). - FILD, - FILD_FLAG, - - /// FP_TO_INT*_IN_MEM - This instruction implements FP_TO_SINT with the - /// integer destination in memory and a FP reg source. This corresponds - /// to the X86::FIST*m instructions and the rounding mode change stuff. It - /// has two inputs (token chain and address) and two outputs (int value - /// and token chain). - FP_TO_INT16_IN_MEM, - FP_TO_INT32_IN_MEM, - FP_TO_INT64_IN_MEM, - - /// FLD - This instruction implements an extending load to FP stack slots. - /// This corresponds to the X86::FLD32m / X86::FLD64m. It takes a chain - /// operand, ptr to load from, and a ValueType node indicating the type - /// to load to. - FLD, - - /// FST - This instruction implements a truncating store to FP stack - /// slots. This corresponds to the X86::FST32m / X86::FST64m. It takes a - /// chain operand, value to store, address, and a ValueType to store it - /// as. - FST, - /// CALL - These operations represent an abstract X86 call /// instruction, which includes a bunch of information. In particular the /// operands of these node are: @@ -105,7 +76,7 @@ namespace llvm { /// CALL, - /// RDTSC_DAG - This operation implements the lowering for + /// RDTSC_DAG - This operation implements the lowering for /// readcyclecounter RDTSC_DAG, @@ -157,11 +128,15 @@ namespace llvm { /// relative displacements. WrapperRIP, - /// MOVQ2DQ - Copies a 64-bit value from a vector to another vector. - /// Can be used to move a vector value from a MMX register to a XMM - /// register. + /// MOVQ2DQ - Copies a 64-bit value from an MMX vector to the low word + /// of an XMM vector, with the high word zero filled. MOVQ2DQ, + /// MOVDQ2Q - Copies a 64-bit value from the low word of an XMM vector + /// to an MMX vector. If you think this is too close to the previous + /// mnemonic, so do I; blame Intel. + MOVDQ2Q, + /// PEXTRB - Extract an 8-bit value from a vector and zero extend it to /// i32, corresponds to X86::PEXTRB. PEXTRB, @@ -197,12 +172,13 @@ namespace llvm { // TLSADDR - Thread Local Storage. TLSADDR, - // SegmentBaseAddress - The address segment:0 - SegmentBaseAddress, + // TLSCALL - Thread Local Storage. When calling to an OS provided + // thunk at the address from an earlier relocation. + TLSCALL, // EH_RETURN - Exception Handling helpers. EH_RETURN, - + /// TC_RETURN - Tail call return. /// operand #0 chain /// operand #1 callee (register or absolute) @@ -210,53 +186,84 @@ namespace llvm { /// operand #3 optional in flag TC_RETURN, - // LCMPXCHG_DAG, LCMPXCHG8_DAG - Compare and swap. - LCMPXCHG_DAG, - LCMPXCHG8_DAG, - - // FNSTCW16m - Store FP control world into i16 memory. - FNSTCW16m, - // VZEXT_MOVL - Vector move low and zero extend. VZEXT_MOVL, - // VZEXT_LOAD - Load, scalar_to_vector, and zero extend. - VZEXT_LOAD, - // VSHL, VSRL - Vector logical left / right shift. VSHL, VSRL, // CMPPD, CMPPS - Vector double/float comparison. // CMPPD, CMPPS - Vector double/float comparison. CMPPD, CMPPS, - + // PCMP* - Vector integer comparisons. PCMPEQB, PCMPEQW, PCMPEQD, PCMPEQQ, PCMPGTB, PCMPGTW, PCMPGTD, PCMPGTQ, - // Advanced Encryption Standard (AES) Instructions - AESIMC, AESENC, AESENCLAST, AESDEC, AESDECLAST, - // ADD, SUB, SMUL, UMUL, etc. - Arithmetic operations with FLAGS results. ADD, SUB, SMUL, UMUL, INC, DEC, OR, XOR, AND, // MUL_IMM - X86 specific multiply by immediate. MUL_IMM, - + // PTEST - Vector bitwise comparisons PTEST, + // TESTP - Vector packed fp sign bitwise comparisons + TESTP, + + // Several flavors of instructions with vector shuffle behaviors. + PALIGN, + PSHUFD, + PSHUFHW, + PSHUFLW, + PSHUFHW_LD, + PSHUFLW_LD, + SHUFPD, + SHUFPS, + MOVDDUP, + MOVSHDUP, + MOVSLDUP, + MOVSHDUP_LD, + MOVSLDUP_LD, + MOVLHPS, + MOVLHPD, + MOVHLPS, + MOVHLPD, + MOVLPS, + MOVLPD, + MOVSD, + MOVSS, + UNPCKLPS, + UNPCKLPD, + UNPCKHPS, + UNPCKHPD, + PUNPCKLBW, + PUNPCKLWD, + PUNPCKLDQ, + PUNPCKLQDQ, + PUNPCKHBW, + PUNPCKHWD, + PUNPCKHDQ, + PUNPCKHQDQ, + // VASTART_SAVE_XMM_REGS - Save xmm argument registers to the stack, // according to %al. An operator is needed so that this can be expanded // with control flow. VASTART_SAVE_XMM_REGS, - // MINGW_ALLOCA - MingW's __alloca call to do stack probing. - MINGW_ALLOCA, + // WIN_ALLOCA - Windows's _chkstk call to do stack probing. + WIN_ALLOCA, - // ATOMADD64_DAG, ATOMSUB64_DAG, ATOMOR64_DAG, ATOMAND64_DAG, - // ATOMXOR64_DAG, ATOMNAND64_DAG, ATOMSWAP64_DAG - + // Memory barrier + MEMBARRIER, + MFENCE, + SFENCE, + LFENCE, + + // ATOMADD64_DAG, ATOMSUB64_DAG, ATOMOR64_DAG, ATOMAND64_DAG, + // ATOMXOR64_DAG, ATOMNAND64_DAG, ATOMSWAP64_DAG - // Atomic 64-bit binary operations. ATOMADD64_DAG = ISD::FIRST_TARGET_MEMORY_OPCODE, ATOMSUB64_DAG, @@ -264,7 +271,50 @@ namespace llvm { ATOMXOR64_DAG, ATOMAND64_DAG, ATOMNAND64_DAG, - ATOMSWAP64_DAG + ATOMSWAP64_DAG, + + // LCMPXCHG_DAG, LCMPXCHG8_DAG - Compare and swap. + LCMPXCHG_DAG, + LCMPXCHG8_DAG, + + // VZEXT_LOAD - Load, scalar_to_vector, and zero extend. + VZEXT_LOAD, + + // FNSTCW16m - Store FP control world into i16 memory. + FNSTCW16m, + + /// FP_TO_INT*_IN_MEM - This instruction implements FP_TO_SINT with the + /// integer destination in memory and a FP reg source. This corresponds + /// to the X86::FIST*m instructions and the rounding mode change stuff. It + /// has two inputs (token chain and address) and two outputs (int value + /// and token chain). + FP_TO_INT16_IN_MEM, + FP_TO_INT32_IN_MEM, + FP_TO_INT64_IN_MEM, + + /// FILD, FILD_FLAG - This instruction implements SINT_TO_FP with the + /// integer source in memory and FP reg result. This corresponds to the + /// X86::FILD*m instructions. It has three inputs (token chain, address, + /// and source type) and two outputs (FP value and token chain). FILD_FLAG + /// also produces a flag). + FILD, + FILD_FLAG, + + /// FLD - This instruction implements an extending load to FP stack slots. + /// This corresponds to the X86::FLD32m / X86::FLD64m. It takes a chain + /// operand, ptr to load from, and a ValueType node indicating the type + /// to load to. + FLD, + + /// FST - This instruction implements a truncating store to FP stack + /// slots. This corresponds to the X86::FST32m / X86::FST64m. It takes a + /// chain operand, value to store, address, and a ValueType to store it + /// as. + FST, + + /// VAARG_64 - This instruction grabs the address of the next argument + /// from a va_list. (reads and modifies the va_list in memory) + VAARG_64 // WARNING: Do not add anything in the end unless you want the node to // have memop! In fact, starting from ATOMADD64_DAG all opcodes will be @@ -377,25 +427,16 @@ namespace llvm { //===--------------------------------------------------------------------===// // X86TargetLowering - X86 Implementation of the TargetLowering interface class X86TargetLowering : public TargetLowering { - int VarArgsFrameIndex; // FrameIndex for start of varargs area. - int RegSaveFrameIndex; // X86-64 vararg func register save area. - unsigned VarArgsGPOffset; // X86-64 vararg func int reg offset. - unsigned VarArgsFPOffset; // X86-64 vararg func fp reg offset. - int BytesToPopOnReturn; // Number of arg bytes ret should pop. - public: explicit X86TargetLowering(X86TargetMachine &TM); - /// getPICBaseSymbol - Return the X86-32 PIC base. - MCSymbol *getPICBaseSymbol(const MachineFunction *MF, MCContext &Ctx) const; - virtual unsigned getJumpTableEncoding() const; virtual const MCExpr * LowerCustomJumpTableEntry(const MachineJumpTableInfo *MJTI, const MachineBasicBlock *MBB, unsigned uid, MCContext &Ctx) const; - + /// getPICJumpTableRelocaBase - Returns relocation base for the given PIC /// jumptable. virtual SDValue getPICJumpTableRelocBase(SDValue Table, @@ -403,11 +444,6 @@ namespace llvm { virtual const MCExpr * getPICJumpTableRelocBaseExpr(const MachineFunction *MF, unsigned JTI, MCContext &Ctx) const; - - // Return the number of bytes that a function should pop when it returns (in - // addition to the space used by the return address). - // - unsigned getBytesToPopOnReturn() const { return BytesToPopOnReturn; } /// getStackPtrReg - Return the stack pointer register we are using: either /// ESP or RSP. @@ -421,11 +457,20 @@ namespace llvm { /// getOptimalMemOpType - Returns the target specific optimal type for load /// and store operations as a result of memset, memcpy, and memmove - /// lowering. It returns EVT::iAny if SelectionDAG should be responsible for - /// determining it. - virtual EVT getOptimalMemOpType(uint64_t Size, unsigned Align, - bool isSrcConst, bool isSrcStr, - SelectionDAG &DAG) const; + /// lowering. If DstAlign is zero that means it's safe to destination + /// alignment can satisfy any constraint. Similarly if SrcAlign is zero it + /// means there isn't a need to check it against alignment requirement, + /// probably because the source does not need to be loaded. If + /// 'NonScalarIntSafe' is true, that means it's safe to return a + /// non-scalar-integer type, e.g. empty string source, constant, or loaded + /// from memory. 'MemcpyStrSrc' indicates whether the memcpy source is + /// constant so it does not need to be loaded. + /// It returns EVT::Other if the type should be determined using generic + /// target-independent logic. + virtual EVT + getOptimalMemOpType(uint64_t Size, unsigned DstAlign, unsigned SrcAlign, + bool NonScalarIntSafe, bool MemcpyStrSrc, + MachineFunction &MF) const; /// allowsUnalignedMemoryAccesses - Returns true if the target allows /// unaligned memory accesses. of the specified type. @@ -435,22 +480,34 @@ namespace llvm { /// LowerOperation - Provide custom lowering hooks for some operations. /// - virtual SDValue LowerOperation(SDValue Op, SelectionDAG &DAG); + virtual SDValue LowerOperation(SDValue Op, SelectionDAG &DAG) const; /// ReplaceNodeResults - Replace the results of node with an illegal result /// type with new values built out of custom code. /// virtual void ReplaceNodeResults(SDNode *N, SmallVectorImpl&Results, - SelectionDAG &DAG); + SelectionDAG &DAG) const; + - virtual SDValue PerformDAGCombine(SDNode *N, DAGCombinerInfo &DCI) const; - virtual MachineBasicBlock *EmitInstrWithCustomInserter(MachineInstr *MI, - MachineBasicBlock *MBB, - DenseMap *EM) const; + /// isTypeDesirableForOp - Return true if the target has native support for + /// the specified value type and it is 'desirable' to use the type for the + /// given node type. e.g. On x86 i16 is legal, but undesirable since i16 + /// instruction encodings are longer and some i16 instructions are slow. + virtual bool isTypeDesirableForOp(unsigned Opc, EVT VT) const; + + /// isTypeDesirable - Return true if the target has native support for the + /// specified value type and it is 'desirable' to use the type. e.g. On x86 + /// i16 is legal, but undesirable since i16 instruction encodings are longer + /// and some i16 instructions are slow. + virtual bool IsDesirableToPromoteOp(SDValue Op, EVT &PVT) const; + + virtual MachineBasicBlock * + EmitInstrWithCustomInserter(MachineInstr *MI, + MachineBasicBlock *MBB) const; + - /// getTargetNodeName - This method returns the name of a target specific /// DAG node. virtual const char *getTargetNodeName(unsigned Opcode) const; @@ -458,26 +515,36 @@ namespace llvm { /// getSetCCResultType - Return the ISD::SETCC ValueType virtual MVT::SimpleValueType getSetCCResultType(EVT VT) const; - /// computeMaskedBitsForTargetNode - Determine which of the bits specified - /// in Mask are known to be either zero or one and return them in the + /// computeMaskedBitsForTargetNode - Determine which of the bits specified + /// in Mask are known to be either zero or one and return them in the /// KnownZero/KnownOne bitsets. virtual void computeMaskedBitsForTargetNode(const SDValue Op, const APInt &Mask, - APInt &KnownZero, + APInt &KnownZero, APInt &KnownOne, const SelectionDAG &DAG, unsigned Depth = 0) const; + // ComputeNumSignBitsForTargetNode - Determine the number of bits in the + // operation that are sign bits. + virtual unsigned ComputeNumSignBitsForTargetNode(SDValue Op, + unsigned Depth) const; + virtual bool - isGAPlusOffset(SDNode *N, GlobalValue* &GA, int64_t &Offset) const; - - SDValue getReturnAddressFrameIndex(SelectionDAG &DAG); + isGAPlusOffset(SDNode *N, const GlobalValue* &GA, int64_t &Offset) const; + + SDValue getReturnAddressFrameIndex(SelectionDAG &DAG) const; virtual bool ExpandInlineAsm(CallInst *CI) const; - + ConstraintType getConstraintType(const std::string &Constraint) const; - - std::vector + + /// Examine constraint string and operand type and determine a weight value. + /// The operand object must already have been set up with the operand type. + virtual ConstraintWeight getSingleConstraintMatchWeight( + AsmOperandInfo &info, const char *constraint) const; + + std::vector getRegClassForInlineAsmConstraint(const std::string &Constraint, EVT VT) const; @@ -489,18 +556,17 @@ namespace llvm { /// being processed is 'm'. virtual void LowerAsmOperandForConstraint(SDValue Op, char ConstraintLetter, - bool hasMemory, std::vector &Ops, SelectionDAG &DAG) const; - + /// getRegForInlineAsmConstraint - Given a physical register constraint /// (e.g. {edx}), return the register number and the register class for the /// register. This should only be used for C_Register constraints. On /// error, this returns a register number of 0. - std::pair + std::pair getRegForInlineAsmConstraint(const std::string &Constraint, EVT VT) const; - + /// isLegalAddressingMode - Return true if the addressing mode represented /// by AM is legal for this target, for a load/store of the specified type. virtual bool isLegalAddressingMode(const AddrMode &AM, const Type *Ty)const; @@ -555,8 +621,8 @@ namespace llvm { // shrink long double fp constant since fldt is very slow. return !X86ScalarSSEf64 || VT == MVT::f80; } - - virtual const X86Subtarget* getSubtarget() { + + const X86Subtarget* getSubtarget() const { return Subtarget; } @@ -569,20 +635,24 @@ namespace llvm { /// createFastISel - This method returns a target specific FastISel object, /// or null if the target does not support "fast" ISel. - virtual FastISel * - createFastISel(MachineFunction &mf, - MachineModuleInfo *mmi, DwarfWriter *dw, - DenseMap &, - DenseMap &, - DenseMap & -#ifndef NDEBUG - , SmallSet & -#endif - ); + virtual FastISel *createFastISel(FunctionLoweringInfo &funcInfo) const; /// getFunctionAlignment - Return the Log2 alignment of this function. virtual unsigned getFunctionAlignment(const Function *F) const; + unsigned getRegPressureLimit(const TargetRegisterClass *RC, + MachineFunction &MF) const; + + /// getStackCookieLocation - Return true if the target stores stack + /// protector cookies at a fixed offset in some non-standard address + /// space, and populates the address space and offset as + /// appropriate. + virtual bool getStackCookieLocation(unsigned &AddressSpace, unsigned &Offset) const; + + protected: + std::pair + findRepresentativeClass(EVT VT) const; + private: /// Subtarget - Keep a pointer to the X86Subtarget around so that we can /// make the right decision when generating code for different targets. @@ -592,8 +662,8 @@ namespace llvm { /// X86StackPtr - X86 physical register used as stack ptr. unsigned X86StackPtr; - - /// X86ScalarSSEf32, X86ScalarSSEf64 - Select between SSE or x87 + + /// X86ScalarSSEf32, X86ScalarSSEf64 - Select between SSE or x87 /// floating point ops. /// When SSE is available, use it for f32 operations. /// When SSE2 is available, use it for f64 operations. @@ -613,17 +683,17 @@ namespace llvm { CallingConv::ID CallConv, bool isVarArg, const SmallVectorImpl &Ins, DebugLoc dl, SelectionDAG &DAG, - SmallVectorImpl &InVals); + SmallVectorImpl &InVals) const; SDValue LowerMemArgument(SDValue Chain, CallingConv::ID CallConv, const SmallVectorImpl &ArgInfo, DebugLoc dl, SelectionDAG &DAG, const CCValAssign &VA, MachineFrameInfo *MFI, - unsigned i); + unsigned i) const; SDValue LowerMemOpCallTo(SDValue Chain, SDValue StackPtr, SDValue Arg, DebugLoc dl, SelectionDAG &DAG, const CCValAssign &VA, - ISD::ArgFlagsTy Flags); + ISD::ArgFlagsTy Flags) const; // Call lowering helpers. @@ -636,123 +706,129 @@ namespace llvm { bool isCalleeStructRet, bool isCallerStructRet, const SmallVectorImpl &Outs, + const SmallVectorImpl &OutVals, const SmallVectorImpl &Ins, SelectionDAG& DAG) const; - bool IsCalleePop(bool isVarArg, CallingConv::ID CallConv); + bool IsCalleePop(bool isVarArg, CallingConv::ID CallConv) const; SDValue EmitTailCallLoadRetAddr(SelectionDAG &DAG, SDValue &OutRetAddr, SDValue Chain, bool IsTailCall, bool Is64Bit, - int FPDiff, DebugLoc dl); + int FPDiff, DebugLoc dl) const; - CCAssignFn *CCAssignFnForNode(CallingConv::ID CallConv) const; - unsigned GetAlignedArgumentStackSize(unsigned StackSize, SelectionDAG &DAG); + unsigned GetAlignedArgumentStackSize(unsigned StackSize, + SelectionDAG &DAG) const; std::pair FP_TO_INTHelper(SDValue Op, SelectionDAG &DAG, - bool isSigned); + bool isSigned) const; SDValue LowerAsSplatVectorLoad(SDValue SrcOp, EVT VT, DebugLoc dl, - SelectionDAG &DAG); - SDValue LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG); - SDValue LowerCONCAT_VECTORS(SDValue Op, SelectionDAG &DAG); - SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG); - SDValue LowerEXTRACT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG); - SDValue LowerEXTRACT_VECTOR_ELT_SSE4(SDValue Op, SelectionDAG &DAG); - SDValue LowerINSERT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG); - SDValue LowerINSERT_VECTOR_ELT_SSE4(SDValue Op, SelectionDAG &DAG); - SDValue LowerSCALAR_TO_VECTOR(SDValue Op, SelectionDAG &DAG); - SDValue LowerConstantPool(SDValue Op, SelectionDAG &DAG); - SDValue LowerBlockAddress(SDValue Op, SelectionDAG &DAG); + SelectionDAG &DAG) const; + SDValue LowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerCONCAT_VECTORS(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerVECTOR_SHUFFLE(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerEXTRACT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerEXTRACT_VECTOR_ELT_SSE4(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerINSERT_VECTOR_ELT(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerINSERT_VECTOR_ELT_SSE4(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerSCALAR_TO_VECTOR(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerConstantPool(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerBlockAddress(SDValue Op, SelectionDAG &DAG) const; SDValue LowerGlobalAddress(const GlobalValue *GV, DebugLoc dl, int64_t Offset, SelectionDAG &DAG) const; - SDValue LowerGlobalAddress(SDValue Op, SelectionDAG &DAG); - SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG); - SDValue LowerExternalSymbol(SDValue Op, SelectionDAG &DAG); - SDValue LowerShift(SDValue Op, SelectionDAG &DAG); + SDValue LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerExternalSymbol(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerShift(SDValue Op, SelectionDAG &DAG) const; SDValue BuildFILD(SDValue Op, EVT SrcVT, SDValue Chain, SDValue StackSlot, - SelectionDAG &DAG); - SDValue LowerSINT_TO_FP(SDValue Op, SelectionDAG &DAG); - SDValue LowerUINT_TO_FP(SDValue Op, SelectionDAG &DAG); - SDValue LowerUINT_TO_FP_i64(SDValue Op, SelectionDAG &DAG); - SDValue LowerUINT_TO_FP_i32(SDValue Op, SelectionDAG &DAG); - SDValue LowerFP_TO_SINT(SDValue Op, SelectionDAG &DAG); - SDValue LowerFP_TO_UINT(SDValue Op, SelectionDAG &DAG); - SDValue LowerFABS(SDValue Op, SelectionDAG &DAG); - SDValue LowerFNEG(SDValue Op, SelectionDAG &DAG); - SDValue LowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG); - SDValue LowerSETCC(SDValue Op, SelectionDAG &DAG); - SDValue LowerVSETCC(SDValue Op, SelectionDAG &DAG); - SDValue LowerSELECT(SDValue Op, SelectionDAG &DAG); - SDValue LowerBRCOND(SDValue Op, SelectionDAG &DAG); - SDValue LowerMEMSET(SDValue Op, SelectionDAG &DAG); - SDValue LowerJumpTable(SDValue Op, SelectionDAG &DAG); - SDValue LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG); - SDValue LowerVASTART(SDValue Op, SelectionDAG &DAG); - SDValue LowerVAARG(SDValue Op, SelectionDAG &DAG); - SDValue LowerVACOPY(SDValue Op, SelectionDAG &DAG); - SDValue LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG); - SDValue LowerRETURNADDR(SDValue Op, SelectionDAG &DAG); - SDValue LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG); - SDValue LowerFRAME_TO_ARGS_OFFSET(SDValue Op, SelectionDAG &DAG); - SDValue LowerEH_RETURN(SDValue Op, SelectionDAG &DAG); - SDValue LowerTRAMPOLINE(SDValue Op, SelectionDAG &DAG); - SDValue LowerFLT_ROUNDS_(SDValue Op, SelectionDAG &DAG); - SDValue LowerCTLZ(SDValue Op, SelectionDAG &DAG); - SDValue LowerCTTZ(SDValue Op, SelectionDAG &DAG); - SDValue LowerMUL_V2I64(SDValue Op, SelectionDAG &DAG); - SDValue LowerXALUO(SDValue Op, SelectionDAG &DAG); - - SDValue LowerCMP_SWAP(SDValue Op, SelectionDAG &DAG); - SDValue LowerLOAD_SUB(SDValue Op, SelectionDAG &DAG); - SDValue LowerREADCYCLECOUNTER(SDValue Op, SelectionDAG &DAG); + SelectionDAG &DAG) const; + SDValue LowerBITCAST(SDValue op, SelectionDAG &DAG) const; + SDValue LowerSINT_TO_FP(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerUINT_TO_FP(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerUINT_TO_FP_i64(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerUINT_TO_FP_i32(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFP_TO_SINT(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFP_TO_UINT(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFABS(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFNEG(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFCOPYSIGN(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerToBT(SDValue And, ISD::CondCode CC, + DebugLoc dl, SelectionDAG &DAG) const; + SDValue LowerSETCC(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerVSETCC(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerSELECT(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerBRCOND(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerMEMSET(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerJumpTable(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerDYNAMIC_STACKALLOC(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerVASTART(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerVAARG(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerVACOPY(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerINTRINSIC_WO_CHAIN(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerRETURNADDR(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFRAMEADDR(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFRAME_TO_ARGS_OFFSET(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerEH_RETURN(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerTRAMPOLINE(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerFLT_ROUNDS_(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerCTLZ(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerCTTZ(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerMUL_V2I64(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerSHL(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerXALUO(SDValue Op, SelectionDAG &DAG) const; + + SDValue LowerCMP_SWAP(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerLOAD_SUB(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerREADCYCLECOUNTER(SDValue Op, SelectionDAG &DAG) const; + SDValue LowerMEMBARRIER(SDValue Op, SelectionDAG &DAG) const; + + // Utility functions to help LowerVECTOR_SHUFFLE + SDValue LowerVECTOR_SHUFFLEv8i16(SDValue Op, SelectionDAG &DAG) const; virtual SDValue LowerFormalArguments(SDValue Chain, CallingConv::ID CallConv, bool isVarArg, const SmallVectorImpl &Ins, DebugLoc dl, SelectionDAG &DAG, - SmallVectorImpl &InVals); + SmallVectorImpl &InVals) const; virtual SDValue LowerCall(SDValue Chain, SDValue Callee, CallingConv::ID CallConv, bool isVarArg, bool &isTailCall, const SmallVectorImpl &Outs, + const SmallVectorImpl &OutVals, const SmallVectorImpl &Ins, DebugLoc dl, SelectionDAG &DAG, - SmallVectorImpl &InVals); + SmallVectorImpl &InVals) const; virtual SDValue LowerReturn(SDValue Chain, CallingConv::ID CallConv, bool isVarArg, const SmallVectorImpl &Outs, - DebugLoc dl, SelectionDAG &DAG); + const SmallVectorImpl &OutVals, + DebugLoc dl, SelectionDAG &DAG) const; + + virtual bool isUsedByReturnOnly(SDNode *N) const; virtual bool CanLowerReturn(CallingConv::ID CallConv, bool isVarArg, - const SmallVectorImpl &OutTys, - const SmallVectorImpl &ArgsFlags, - SelectionDAG &DAG); + const SmallVectorImpl &Outs, + LLVMContext &Context) const; void ReplaceATOMIC_BINARY_64(SDNode *N, SmallVectorImpl &Results, - SelectionDAG &DAG, unsigned NewOp); - - SDValue EmitTargetCodeForMemset(SelectionDAG &DAG, DebugLoc dl, - SDValue Chain, - SDValue Dst, SDValue Src, - SDValue Size, unsigned Align, - const Value *DstSV, uint64_t DstSVOff); - SDValue EmitTargetCodeForMemcpy(SelectionDAG &DAG, DebugLoc dl, - SDValue Chain, - SDValue Dst, SDValue Src, - SDValue Size, unsigned Align, - bool AlwaysInline, - const Value *DstSV, uint64_t DstSVOff, - const Value *SrcSV, uint64_t SrcSVOff); - + SelectionDAG &DAG, unsigned NewOp) const; + /// Utility function to emit string processing sse4.2 instructions /// that return in xmm0. /// This takes the instruction to expand, the associated machine basic /// block, the number of args, and whether or not the second arg is /// in memory or not. MachineBasicBlock *EmitPCMP(MachineInstr *BInstr, MachineBasicBlock *BB, - unsigned argNum, bool inMem) const; + unsigned argNum, bool inMem) const; + + /// Utility functions to emit monitor and mwait instructions. These + /// need to make sure that the arguments to the intrinsic are in the + /// correct registers. + MachineBasicBlock *EmitMonitor(MachineInstr *MI, + MachineBasicBlock *BB) const; + MachineBasicBlock *EmitMwait(MachineInstr *MI, MachineBasicBlock *BB) const; /// Utility function to emit atomic bitwise operations (and, or, xor). /// It takes the bitwise instruction to expand, the associated machine basic @@ -764,7 +840,6 @@ namespace llvm { unsigned immOpc, unsigned loadOpc, unsigned cxchgOpc, - unsigned copyOpc, unsigned notOpc, unsigned EAXreg, TargetRegisterClass *RC, @@ -778,7 +853,7 @@ namespace llvm { unsigned immOpcL, unsigned immOpcH, bool invSrc = false) const; - + /// Utility function to emit atomic min and max. It takes the min/max /// instruction to expand, the associated basic block, and the associated /// cmov opcode for moving the min or max value. @@ -786,39 +861,40 @@ namespace llvm { MachineBasicBlock *BB, unsigned cmovOpc) const; + // Utility function to emit the low-level va_arg code for X86-64. + MachineBasicBlock *EmitVAARG64WithCustomInserter( + MachineInstr *MI, + MachineBasicBlock *MBB) const; + /// Utility function to emit the xmm reg save portion of va_start. MachineBasicBlock *EmitVAStartSaveXMMRegsWithCustomInserter( MachineInstr *BInstr, MachineBasicBlock *BB) const; MachineBasicBlock *EmitLoweredSelect(MachineInstr *I, - MachineBasicBlock *BB, - DenseMap *EM) const; + MachineBasicBlock *BB) const; + + MachineBasicBlock *EmitLoweredWinAlloca(MachineInstr *MI, + MachineBasicBlock *BB) const; + + MachineBasicBlock *EmitLoweredTLSCall(MachineInstr *MI, + MachineBasicBlock *BB) const; - MachineBasicBlock *EmitLoweredMingwAlloca(MachineInstr *MI, - MachineBasicBlock *BB, - DenseMap *EM) const; + MachineBasicBlock *emitLoweredTLSAddr(MachineInstr *MI, + MachineBasicBlock *BB) const; /// Emit nodes that will be selected as "test Op0,Op0", or something /// equivalent, for use with the given x86 condition code. - SDValue EmitTest(SDValue Op0, unsigned X86CC, SelectionDAG &DAG); + SDValue EmitTest(SDValue Op0, unsigned X86CC, SelectionDAG &DAG) const; /// Emit nodes that will be selected as "cmp Op0,Op1", or something /// equivalent, for use with the given x86 condition code. SDValue EmitCmp(SDValue Op0, SDValue Op1, unsigned X86CC, - SelectionDAG &DAG); + SelectionDAG &DAG) const; }; namespace X86 { - FastISel *createFastISel(MachineFunction &mf, - MachineModuleInfo *mmi, DwarfWriter *dw, - DenseMap &, - DenseMap &, - DenseMap & -#ifndef NDEBUG - , SmallSet & -#endif - ); + FastISel *createFastISel(FunctionLoweringInfo &funcInfo); } }