namespace X86ISD {
// X86 Specific DAG Nodes
- enum NodeType {
+ enum NodeType : unsigned {
// Start the numbering where the builtin ops leave off.
FIRST_NUMBER = ISD::BUILTIN_OP_END,
- /// BSF - Bit scan forward.
- /// BSR - Bit scan reverse.
+ /// Bit scan forward.
BSF,
+ /// Bit scan reverse.
BSR,
- /// SHLD, SHRD - Double shift instructions. These correspond to
+ /// Double shift instructions. These correspond to
/// X86::SHLDxx and X86::SHRDxx instructions.
SHLD,
SHRD,
- /// FAND - Bitwise logical AND of floating point values. This corresponds
+ /// Bitwise logical AND of floating point values. This corresponds
/// to X86::ANDPS or X86::ANDPD.
FAND,
- /// FOR - Bitwise logical OR of floating point values. This corresponds
+ /// Bitwise logical OR of floating point values. This corresponds
/// to X86::ORPS or X86::ORPD.
FOR,
- /// FXOR - Bitwise logical XOR of floating point values. This corresponds
+ /// Bitwise logical XOR of floating point values. This corresponds
/// to X86::XORPS or X86::XORPD.
FXOR,
- /// FANDN - Bitwise logical ANDNOT of floating point values. This
+ /// Bitwise logical ANDNOT of floating point values. This
/// corresponds to X86::ANDNPS or X86::ANDNPD.
FANDN,
- /// FSRL - Bitwise logical right shift of floating point values. These
- /// corresponds to X86::PSRLDQ.
- FSRL,
-
- /// CALL - These operations represent an abstract X86 call
+ /// These operations represent an abstract X86 call
/// instruction, which includes a bunch of information. In particular the
/// operands of these node are:
///
///
CALL,
- /// RDTSC_DAG - This operation implements the lowering for
- /// readcyclecounter
+ /// This operation implements the lowering for readcyclecounter
RDTSC_DAG,
/// X86 Read Time-Stamp Counter and Processor ID.
/// 1 is the number of bytes of stack to pop.
RET_FLAG,
- /// REP_STOS - Repeat fill, corresponds to X86::REP_STOSx.
+ /// Repeat fill, corresponds to X86::REP_STOSx.
REP_STOS,
- /// REP_MOVS - Repeat move, corresponds to X86::REP_MOVSx.
+ /// Repeat move, corresponds to X86::REP_MOVSx.
REP_MOVS,
- /// GlobalBaseReg - On Darwin, this node represents the result of the popl
+ /// On Darwin, this node represents the result of the popl
/// at function entry, used for PIC code.
GlobalBaseReg,
- /// Wrapper - A wrapper node for TargetConstantPool,
+ /// A wrapper node for TargetConstantPool,
/// TargetExternalSymbol, and TargetGlobalAddress.
Wrapper,
- /// WrapperRIP - Special wrapper used under X86-64 PIC mode for RIP
+ /// Special wrapper used under X86-64 PIC mode for RIP
/// relative displacements.
WrapperRIP,
- /// MOVDQ2Q - Copies a 64-bit value from the low word of an XMM vector
+ /// 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,
- /// MMX_MOVD2W - Copies a 32-bit value from the low word of a MMX
+ /// Copies a 32-bit value from the low word of a MMX
/// vector to a GPR.
MMX_MOVD2W,
- /// MMX_MOVW2D - Copies a GPR into the low 32-bit word of a MMX vector
+ /// Copies a GPR into the low 32-bit word of a MMX vector
/// and zero out the high word.
MMX_MOVW2D,
- /// PEXTRB - Extract an 8-bit value from a vector and zero extend it to
+ /// Extract an 8-bit value from a vector and zero extend it to
/// i32, corresponds to X86::PEXTRB.
PEXTRB,
- /// PEXTRW - Extract a 16-bit value from a vector and zero extend it to
+ /// Extract a 16-bit value from a vector and zero extend it to
/// i32, corresponds to X86::PEXTRW.
PEXTRW,
- /// INSERTPS - Insert any element of a 4 x float vector into any element
+ /// Insert any element of a 4 x float vector into any element
/// of a destination 4 x floatvector.
INSERTPS,
- /// PINSRB - Insert the lower 8-bits of a 32-bit value to a vector,
+ /// Insert the lower 8-bits of a 32-bit value to a vector,
/// corresponds to X86::PINSRB.
PINSRB,
- /// PINSRW - Insert the lower 16-bits of a 32-bit value to a vector,
+ /// Insert the lower 16-bits of a 32-bit value to a vector,
/// corresponds to X86::PINSRW.
PINSRW, MMX_PINSRW,
- /// PSHUFB - Shuffle 16 8-bit values within a vector.
+ /// Shuffle 16 8-bit values within a vector.
PSHUFB,
- /// ANDNP - Bitwise Logical AND NOT of Packed FP values.
+ /// Compute Sum of Absolute Differences.
+ PSADBW,
+ /// Compute Double Block Packed Sum-Absolute-Differences
+ DBPSADBW,
+
+ /// Bitwise Logical AND NOT of Packed FP values.
ANDNP,
- /// PSIGN - Copy integer sign.
+ /// Copy integer sign.
PSIGN,
- /// BLENDI - Blend where the selector is an immediate.
+ /// Blend where the selector is an immediate.
BLENDI,
- /// SHRUNKBLEND - Blend where the condition has been shrunk.
+ /// Blend where the condition has been shrunk.
/// This is used to emphasize that the condition mask is
/// no more valid for generic VSELECT optimizations.
SHRUNKBLEND,
- /// ADDSUB - Combined add and sub on an FP vector.
+ /// Combined add and sub on an FP vector.
ADDSUB,
- // FADD, FSUB, FMUL, FDIV, FMIN, FMAX - FP vector ops with rounding mode.
+
+ // FP vector ops with rounding mode.
FADD_RND,
FSUB_RND,
FMUL_RND,
FDIV_RND,
-
- // SUBUS - Integer sub with unsigned saturation.
+ FMAX_RND,
+ FMIN_RND,
+ FSQRT_RND,
+
+ // FP vector get exponent
+ FGETEXP_RND,
+ // Extract Normalized Mantissas
+ VGETMANT,
+ // FP Scale
+ SCALEF,
+ // Integer add/sub with unsigned saturation.
+ ADDUS,
SUBUS,
-
- /// HADD - Integer horizontal add.
+ // Integer add/sub with signed saturation.
+ ADDS,
+ SUBS,
+ // Unsigned Integer average
+ AVG,
+ /// Integer horizontal add.
HADD,
- /// HSUB - Integer horizontal sub.
+ /// Integer horizontal sub.
HSUB,
- /// FHADD - Floating point horizontal add.
+ /// Floating point horizontal add.
FHADD,
- /// FHSUB - Floating point horizontal sub.
+ /// Floating point horizontal sub.
FHSUB,
- /// UMAX, UMIN - Unsigned integer max and min.
- UMAX, UMIN,
+ // Integer absolute value
+ ABS,
- /// SMAX, SMIN - Signed integer max and min.
- SMAX, SMIN,
+ // Detect Conflicts Within a Vector
+ CONFLICT,
- /// FMAX, FMIN - Floating point max and min.
- ///
+ /// Floating point max and min.
FMAX, FMIN,
- /// FMAXC, FMINC - Commutative FMIN and FMAX.
+ /// Commutative FMIN and FMAX.
FMAXC, FMINC,
- /// FRSQRT, FRCP - Floating point reciprocal-sqrt and reciprocal
- /// approximation. Note that these typically require refinement
+ /// Floating point reciprocal-sqrt and reciprocal approximation.
+ /// Note that these typically require refinement
/// in order to obtain suitable precision.
FRSQRT, FRCP,
- // TLSADDR - Thread Local Storage.
+ // Thread Local Storage.
TLSADDR,
- // TLSBASEADDR - Thread Local Storage. A call to get the start address
+ // Thread Local Storage. A call to get the start address
// of the TLS block for the current module.
TLSBASEADDR,
- // TLSCALL - Thread Local Storage. When calling to an OS provided
+ // Thread Local Storage. When calling to an OS provided
// thunk at the address from an earlier relocation.
TLSCALL,
- // EH_RETURN - Exception Handling helpers.
+ // Exception Handling helpers.
EH_RETURN,
- // EH_SJLJ_SETJMP - SjLj exception handling setjmp.
+ // SjLj exception handling setjmp.
EH_SJLJ_SETJMP,
- // EH_SJLJ_LONGJMP - SjLj exception handling longjmp.
+ // SjLj exception handling longjmp.
EH_SJLJ_LONGJMP,
- /// TC_RETURN - Tail call return. See X86TargetLowering::LowerCall for
+ /// Tail call return. See X86TargetLowering::LowerCall for
/// the list of operands.
TC_RETURN,
- // VZEXT_MOVL - Vector move to low scalar and zero higher vector elements.
+ // Vector move to low scalar and zero higher vector elements.
VZEXT_MOVL,
- // VZEXT - Vector integer zero-extend.
+ // Vector integer zero-extend.
VZEXT,
- // VSEXT - Vector integer signed-extend.
+ // Vector integer signed-extend.
VSEXT,
- // VTRUNC - Vector integer truncate.
+ // Vector integer truncate.
VTRUNC,
+ // Vector integer truncate with unsigned/signed saturation.
+ VTRUNCUS, VTRUNCS,
- // VTRUNC - Vector integer truncate with mask.
- VTRUNCM,
-
- // VFPEXT - Vector FP extend.
+ // Vector FP extend.
VFPEXT,
- // VFPROUND - Vector FP round.
+ // Vector FP round.
VFPROUND,
- // VSHL, VSRL - 128-bit vector logical left / right shift
+ // Vector signed/unsigned integer to double.
+ CVTDQ2PD, CVTUDQ2PD,
+
+ // 128-bit vector logical left / right shift
VSHLDQ, VSRLDQ,
- // VSHL, VSRL, VSRA - Vector shift elements
+ // Vector shift elements
VSHL, VSRL, VSRA,
- // VSHLI, VSRLI, VSRAI - Vector shift elements by immediate
+ // Vector shift elements by immediate
VSHLI, VSRLI, VSRAI,
- // CMPP - Vector packed double/float comparison.
+ // Vector packed double/float comparison.
CMPP,
- // PCMP* - Vector integer comparisons.
+ // Vector integer comparisons.
PCMPEQ, PCMPGT,
- // PCMP*M - Vector integer comparisons, the result is in a mask vector.
+ // Vector integer comparisons, the result is in a mask vector.
PCMPEQM, PCMPGTM,
- /// CMPM, CMPMU - Vector comparison generating mask bits for fp and
+ /// Vector comparison generating mask bits for fp and
/// integer signed and unsigned data types.
CMPM,
CMPMU,
+ // Vector comparison with rounding mode for FP values
+ CMPM_RND,
- // ADD, SUB, SMUL, etc. - Arithmetic operations with FLAGS results.
+ // Arithmetic operations with FLAGS results.
ADD, SUB, ADC, SBB, SMUL,
INC, DEC, OR, XOR, AND,
- BEXTR, // BEXTR - Bit field extract
+ BEXTR, // Bit field extract
UMUL, // LOW, HI, FLAGS = umul LHS, RHS
UDIVREM8_ZEXT_HREG,
SDIVREM8_SEXT_HREG,
- // MUL_IMM - X86 specific multiply by immediate.
+ // X86-specific multiply by immediate.
MUL_IMM,
- // PTEST - Vector bitwise comparisons.
+ // Vector bitwise comparisons.
PTEST,
- // TESTP - Vector packed fp sign bitwise comparisons.
+ // Vector packed fp sign bitwise comparisons.
TESTP,
- // TESTM, TESTNM - Vector "test" in AVX-512, the result is in a mask vector.
+ // Vector "test" in AVX-512, the result is in a mask vector.
TESTM,
TESTNM,
// OR/AND test for masks
KORTEST,
+ KTEST,
// Several flavors of instructions with vector shuffle behaviors.
PACKSS,
PSHUFHW,
PSHUFLW,
SHUFP,
+ //Shuffle Packed Values at 128-bit granularity
+ SHUF128,
MOVDDUP,
MOVSHDUP,
MOVSLDUP,
VPERMIV3,
VPERMI,
VPERM2X128,
+ //Fix Up Special Packed Float32/64 values
+ VFIXUPIMM,
+ //Range Restriction Calculation For Packed Pairs of Float32/64 values
+ VRANGE,
+ // Reduce - Perform Reduction Transformation on scalar\packed FP
+ VREDUCE,
+ // RndScale - Round FP Values To Include A Given Number Of Fraction Bits
+ VRNDSCALE,
+ // VFPCLASS - Tests Types Of a FP Values
+ VFPCLASS,
+ // Broadcast scalar to vector
VBROADCAST,
- // masked broadcast
- VBROADCASTM,
+ // Broadcast subvector to vector
+ SUBV_BROADCAST,
// Insert/Extract vector element
VINSERT,
VEXTRACT,
+ /// SSE4A Extraction and Insertion.
+ EXTRQI, INSERTQI,
+
+ // XOP arithmetic/logical shifts
+ VPSHA, VPSHL,
+ // XOP signed/unsigned integer comparisons
+ VPCOM, VPCOMU,
+
// Vector multiply packed unsigned doubleword integers
PMULUDQ,
// Vector multiply packed signed doubleword integers
PMULDQ,
-
+ // Vector Multiply Packed UnsignedIntegers with Round and Scale
+ MULHRS,
+ // Multiply and Add Packed Integers
+ VPMADDUBSW, VPMADDWD,
// FMA nodes
FMADD,
FNMADD,
FNMSUB_RND,
FMADDSUB_RND,
FMSUBADD_RND,
- RNDSCALE,
// Compress and expand
COMPRESS,
EXPAND,
+ //Convert Unsigned/Integer to Scalar Floating-Point Value
+ //with rounding mode
+ SINT_TO_FP_RND,
+ UINT_TO_FP_RND,
+
+ // Vector float/double to signed/unsigned integer.
+ FP_TO_SINT_RND, FP_TO_UINT_RND,
// 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,
// falls back to heap allocation if not.
SEG_ALLOCA,
- // Windows's _ftol2 runtime routine to do fptoui.
- WIN_FTOL,
-
// Memory barrier
MEMBARRIER,
MFENCE,
const X86Subtarget &STI);
unsigned getJumpTableEncoding() const override;
+ bool useSoftFloat() const override;
- MVT getScalarShiftAmountTy(EVT LHSTy) const override { return MVT::i8; }
+ MVT getScalarShiftAmountTy(const DataLayout &, EVT) const override {
+ return MVT::i8;
+ }
const MCExpr *
LowerCustomJumpTableEntry(const MachineJumpTableInfo *MJTI,
/// function arguments in the caller parameter area. For X86, aggregates
/// that contains are placed at 16-byte boundaries while the rest are at
/// 4-byte boundaries.
- unsigned getByValTypeAlignment(Type *Ty) const override;
+ unsigned getByValTypeAlignment(Type *Ty,
+ const DataLayout &DL) const override;
/// Returns the target specific optimal type for load
/// and store operations as a result of memset, memcpy, and memmove
/// legal as the hook is used before type legalization.
bool isSafeMemOpType(MVT VT) const override;
- /// Returns true if the target allows
- /// unaligned memory accesses. of the specified type. Returns whether it
- /// is "fast" by reference in the second argument.
+ /// Returns true if the target allows unaligned memory accesses of the
+ /// specified type. Returns whether it is "fast" in the last argument.
bool allowsMisalignedMemoryAccesses(EVT VT, unsigned AS, unsigned Align,
bool *Fast) const override;
bool isCheapToSpeculateCtlz() const override;
/// Return the value type to use for ISD::SETCC.
- EVT getSetCCResultType(LLVMContext &Context, EVT VT) const override;
+ EVT getSetCCResultType(const DataLayout &DL, LLVMContext &Context,
+ EVT VT) const override;
/// Determine which of the bits specified in Mask are known to be either
/// zero or one and return them in the KnownZero/KnownOne bitsets.
bool ExpandInlineAsm(CallInst *CI) const override;
- ConstraintType
- getConstraintType(const std::string &Constraint) const override;
+ ConstraintType getConstraintType(StringRef Constraint) const override;
/// Examine constraint string and operand type and determine a weight value.
/// The operand object must already have been set up with the operand type.
std::vector<SDValue> &Ops,
SelectionDAG &DAG) const override;
+ unsigned
+ getInlineAsmMemConstraint(StringRef ConstraintCode) const override {
+ if (ConstraintCode == "i")
+ return InlineAsm::Constraint_i;
+ else if (ConstraintCode == "o")
+ return InlineAsm::Constraint_o;
+ else if (ConstraintCode == "v")
+ return InlineAsm::Constraint_v;
+ else if (ConstraintCode == "X")
+ return InlineAsm::Constraint_X;
+ return TargetLowering::getInlineAsmMemConstraint(ConstraintCode);
+ }
+
/// 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<unsigned, const TargetRegisterClass*>
- getRegForInlineAsmConstraint(const std::string &Constraint,
- MVT VT) const override;
+ std::pair<unsigned, const TargetRegisterClass *>
+ getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI,
+ StringRef Constraint, MVT VT) const override;
/// Return true if the addressing mode represented
/// by AM is legal for this target, for a load/store of the specified type.
- bool isLegalAddressingMode(const AddrMode &AM, Type *Ty) const override;
+ bool isLegalAddressingMode(const DataLayout &DL, const AddrMode &AM,
+ Type *Ty, unsigned AS) const override;
/// Return true if the specified immediate is legal
/// icmp immediate, that is the target has icmp instructions which can
/// of the specified type.
/// If the AM is supported, the return value must be >= 0.
/// If the AM is not supported, it returns a negative value.
- int getScalingFactorCost(const AddrMode &AM, Type *Ty) const override;
+ int getScalingFactorCost(const DataLayout &DL, const AddrMode &AM, Type *Ty,
+ unsigned AS) const override;
bool isVectorShiftByScalarCheap(Type *Ty) const override;
(VT == MVT::f32 && X86ScalarSSEf32); // f32 is when SSE1
}
- /// Return true if the target uses the MSVC _ftol2 routine for fptoui.
- bool isTargetFTOL() const;
-
- /// Return true if the MSVC _ftol2 routine should be used for fptoui to the
- /// given type.
- bool isIntegerTypeFTOL(EVT VT) const {
- return isTargetFTOL() && VT == MVT::i64;
- }
-
/// \brief Returns true if it is beneficial to convert a load of a constant
/// to just the constant itself.
bool shouldConvertConstantLoadToIntImm(const APInt &Imm,
return nullptr; // nothing to do, move along.
}
- unsigned getRegisterByName(const char* RegName, EVT VT) const override;
+ unsigned getRegisterByName(const char* RegName, EVT VT,
+ SelectionDAG &DAG) const override;
/// This method returns a target specific FastISel object,
/// or null if the target does not support "fast" ISel.
bool getStackCookieLocation(unsigned &AddressSpace,
unsigned &Offset) const override;
+ /// Return true if the target stores SafeStack pointer at a fixed offset in
+ /// some non-standard address space, and populates the address space and
+ /// offset as appropriate.
+ bool getSafeStackPointerLocation(unsigned &AddressSpace,
+ unsigned &Offset) const override;
+
SDValue BuildFILD(SDValue Op, EVT SrcVT, SDValue Chain, SDValue StackSlot,
SelectionDAG &DAG) const;
/// \brief Customize the preferred legalization strategy for certain types.
LegalizeTypeAction getPreferredVectorAction(EVT VT) const override;
+ bool isIntDivCheap(EVT VT, AttributeSet Attr) const override;
+
protected:
- std::pair<const TargetRegisterClass*, uint8_t>
- findRepresentativeClass(MVT VT) const override;
+ std::pair<const TargetRegisterClass *, uint8_t>
+ findRepresentativeClass(const TargetRegisterInfo *TRI,
+ MVT VT) const override;
private:
/// Keep a pointer to the X86Subtarget around so that we can
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 LowerCLEANUPRET(SDValue Op, SelectionDAG &DAG) const;
SDValue lowerEH_SJLJ_SETJMP(SDValue Op, SelectionDAG &DAG) const;
SDValue lowerEH_SJLJ_LONGJMP(SDValue Op, SelectionDAG &DAG) const;
SDValue LowerINIT_TRAMPOLINE(SDValue Op, SelectionDAG &DAG) const;
SDValue LowerFLT_ROUNDS_(SDValue Op, SelectionDAG &DAG) const;
SDValue LowerWin64_i128OP(SDValue Op, SelectionDAG &DAG) const;
+ SDValue LowerGC_TRANSITION_START(SDValue Op, SelectionDAG &DAG) const;
+ SDValue LowerGC_TRANSITION_END(SDValue Op, SelectionDAG &DAG) const;
SDValue
LowerFormalArguments(SDValue Chain,
const MCPhysReg *getScratchRegisters(CallingConv::ID CC) const override;
- bool shouldExpandAtomicLoadInIR(LoadInst *SI) const override;
+ TargetLoweringBase::AtomicExpansionKind
+ shouldExpandAtomicLoadInIR(LoadInst *SI) const override;
bool shouldExpandAtomicStoreInIR(StoreInst *SI) const override;
- bool shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const override;
+ TargetLoweringBase::AtomicExpansionKind
+ shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const override;
LoadInst *
lowerIdempotentRMWIntoFencedLoad(AtomicRMWInst *AI) const override;
- bool needsCmpXchgNb(const Type *MemType) const;
+ bool needsCmpXchgNb(Type *MemType) const;
/// Utility function to emit atomic-load-arith operations (and, or, xor,
/// nand, max, min, umax, umin). It takes the corresponding instruction to
MachineBasicBlock *EmitLoweredSelect(MachineInstr *I,
MachineBasicBlock *BB) const;
+ MachineBasicBlock *EmitLoweredAtomicFP(MachineInstr *I,
+ MachineBasicBlock *BB) const;
+
MachineBasicBlock *EmitLoweredWinAlloca(MachineInstr *MI,
MachineBasicBlock *BB) const;
/// Use rcp* to speed up fdiv calculations.
SDValue getRecipEstimate(SDValue Operand, DAGCombinerInfo &DCI,
unsigned &RefinementSteps) const override;
+
+ /// Reassociate floating point divisions into multiply by reciprocal.
+ unsigned combineRepeatedFPDivisors() const override;
};
namespace X86 {