Refactor X86 target to separate MC code from Target code.

llvm-svn: 135930

lib/MC/ELFObjectWriter.cpp

@@ -28,7 +28,7 @@
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/StringSwitch.h"
 
-#include "../Target/X86/X86FixupKinds.h"
+#include "../Target/X86/MCTargetDesc/X86FixupKinds.h"
 #include "../Target/ARM/MCTargetDesc/ARMFixupKinds.h"
 
 #include <vector>

lib/MC/WinCOFFObjectWriter.cpp

@@ -33,7 +33,7 @@
 
 #include "llvm/Support/TimeValue.h"
 
-#include "../Target/X86/X86FixupKinds.h"
+#include "../Target/X86/MCTargetDesc/X86FixupKinds.h"
 
 #include <cstdio>
 

lib/Target/X86/MCTargetDesc/X86BaseInfo.h

@@ -0,0 +1,544 @@
+//===-- X86BaseInfo.h - Top level definitions for X86 -------- --*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file contains small standalone helper functions and enum definitions for
+// the X86 target useful for the compiler back-end and the MC libraries.
+// As such, it deliberately does not include references to LLVM core
+// code gen types, passes, etc..
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef X86BASEINFO_H
+#define X86BASEINFO_H
+
+#include "X86MCTargetDesc.h"
+#include "llvm/Support/DataTypes.h"
+#include <cassert>
+
+namespace llvm {
+
+namespace X86 {
+  // Enums for memory operand decoding.  Each memory operand is represented with
+  // a 5 operand sequence in the form:
+  //   [BaseReg, ScaleAmt, IndexReg, Disp, Segment]
+  // These enums help decode this.
+  enum {
+    AddrBaseReg = 0,
+    AddrScaleAmt = 1,
+    AddrIndexReg = 2,
+    AddrDisp = 3,
+
+    /// AddrSegmentReg - The operand # of the segment in the memory operand.
+    AddrSegmentReg = 4,
+
+    /// AddrNumOperands - Total number of operands in a memory reference.
+    AddrNumOperands = 5
+  };
+} // end namespace X86;
+ 
+
+/// X86II - This namespace holds all of the target specific flags that
+/// instruction info tracks.
+///
+namespace X86II {
+  /// Target Operand Flag enum.
+  enum TOF {
+    //===------------------------------------------------------------------===//
+    // X86 Specific MachineOperand flags.
+
+    MO_NO_FLAG,
+
+    /// MO_GOT_ABSOLUTE_ADDRESS - On a symbol operand, this represents a
+    /// relocation of:
+    ///    SYMBOL_LABEL + [. - PICBASELABEL]
+    MO_GOT_ABSOLUTE_ADDRESS,
+
+    /// MO_PIC_BASE_OFFSET - On a symbol operand this indicates that the
+    /// immediate should get the value of the symbol minus the PIC base label:
+    ///    SYMBOL_LABEL - PICBASELABEL
+    MO_PIC_BASE_OFFSET,
+
+    /// MO_GOT - On a symbol operand this indicates that the immediate is the
+    /// offset to the GOT entry for the symbol name from the base of the GOT.
+    ///
+    /// See the X86-64 ELF ABI supplement for more details.
+    ///    SYMBOL_LABEL @GOT
+    MO_GOT,
+
+    /// MO_GOTOFF - On a symbol operand this indicates that the immediate is
+    /// the offset to the location of the symbol name from the base of the GOT.
+    ///
+    /// See the X86-64 ELF ABI supplement for more details.
+    ///    SYMBOL_LABEL @GOTOFF
+    MO_GOTOFF,
+
+    /// MO_GOTPCREL - On a symbol operand this indicates that the immediate is
+    /// offset to the GOT entry for the symbol name from the current code
+    /// location.
+    ///
+    /// See the X86-64 ELF ABI supplement for more details.
+    ///    SYMBOL_LABEL @GOTPCREL
+    MO_GOTPCREL,
+
+    /// MO_PLT - On a symbol operand this indicates that the immediate is
+    /// offset to the PLT entry of symbol name from the current code location.
+    ///
+    /// See the X86-64 ELF ABI supplement for more details.
+    ///    SYMBOL_LABEL @PLT
+    MO_PLT,
+
+    /// MO_TLSGD - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @TLSGD
+    MO_TLSGD,
+
+    /// MO_GOTTPOFF - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @GOTTPOFF
+    MO_GOTTPOFF,
+
+    /// MO_INDNTPOFF - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @INDNTPOFF
+    MO_INDNTPOFF,
+
+    /// MO_TPOFF - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @TPOFF
+    MO_TPOFF,
+
+    /// MO_NTPOFF - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @NTPOFF
+    MO_NTPOFF,
+
+    /// MO_DLLIMPORT - On a symbol operand "FOO", this indicates that the
+    /// reference is actually to the "__imp_FOO" symbol.  This is used for
+    /// dllimport linkage on windows.
+    MO_DLLIMPORT,
+
+    /// MO_DARWIN_STUB - On a symbol operand "FOO", this indicates that the
+    /// reference is actually to the "FOO$stub" symbol.  This is used for calls
+    /// and jumps to external functions on Tiger and earlier.
+    MO_DARWIN_STUB,
+
+    /// MO_DARWIN_NONLAZY - On a symbol operand "FOO", this indicates that the
+    /// reference is actually to the "FOO$non_lazy_ptr" symbol, which is a
+    /// non-PIC-base-relative reference to a non-hidden dyld lazy pointer stub.
+    MO_DARWIN_NONLAZY,
+
+    /// MO_DARWIN_NONLAZY_PIC_BASE - On a symbol operand "FOO", this indicates
+    /// that the reference is actually to "FOO$non_lazy_ptr - PICBASE", which is
+    /// a PIC-base-relative reference to a non-hidden dyld lazy pointer stub.
+    MO_DARWIN_NONLAZY_PIC_BASE,
+
+    /// MO_DARWIN_HIDDEN_NONLAZY_PIC_BASE - On a symbol operand "FOO", this
+    /// indicates that the reference is actually to "FOO$non_lazy_ptr -PICBASE",
+    /// which is a PIC-base-relative reference to a hidden dyld lazy pointer
+    /// stub.
+    MO_DARWIN_HIDDEN_NONLAZY_PIC_BASE,
+
+    /// MO_TLVP - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// This is the TLS offset for the Darwin TLS mechanism.
+    MO_TLVP,
+
+    /// MO_TLVP_PIC_BASE - On a symbol operand this indicates that the immediate
+    /// is some TLS offset from the picbase.
+    ///
+    /// This is the 32-bit TLS offset for Darwin TLS in PIC mode.
+    MO_TLVP_PIC_BASE
+  };
+
+  enum {
+    //===------------------------------------------------------------------===//
+    // Instruction encodings.  These are the standard/most common forms for X86
+    // instructions.
+    //
+
+    // PseudoFrm - This represents an instruction that is a pseudo instruction
+    // or one that has not been implemented yet.  It is illegal to code generate
+    // it, but tolerated for intermediate implementation stages.
+    Pseudo         = 0,
+
+    /// Raw - This form is for instructions that don't have any operands, so
+    /// they are just a fixed opcode value, like 'leave'.
+    RawFrm         = 1,
+
+    /// AddRegFrm - This form is used for instructions like 'push r32' that have
+    /// their one register operand added to their opcode.
+    AddRegFrm      = 2,
+
+    /// MRMDestReg - This form is used for instructions that use the Mod/RM byte
+    /// to specify a destination, which in this case is a register.
+    ///
+    MRMDestReg     = 3,
+
+    /// MRMDestMem - This form is used for instructions that use the Mod/RM byte
+    /// to specify a destination, which in this case is memory.
+    ///
+    MRMDestMem     = 4,
+
+    /// MRMSrcReg - This form is used for instructions that use the Mod/RM byte
+    /// to specify a source, which in this case is a register.
+    ///
+    MRMSrcReg      = 5,
+
+    /// MRMSrcMem - This form is used for instructions that use the Mod/RM byte
+    /// to specify a source, which in this case is memory.
+    ///
+    MRMSrcMem      = 6,
+
+    /// MRM[0-7][rm] - These forms are used to represent instructions that use
+    /// a Mod/RM byte, and use the middle field to hold extended opcode
+    /// information.  In the intel manual these are represented as /0, /1, ...
+    ///
+
+    // First, instructions that operate on a register r/m operand...
+    MRM0r = 16,  MRM1r = 17,  MRM2r = 18,  MRM3r = 19, // Format /0 /1 /2 /3
+    MRM4r = 20,  MRM5r = 21,  MRM6r = 22,  MRM7r = 23, // Format /4 /5 /6 /7
+
+    // Next, instructions that operate on a memory r/m operand...
+    MRM0m = 24,  MRM1m = 25,  MRM2m = 26,  MRM3m = 27, // Format /0 /1 /2 /3
+    MRM4m = 28,  MRM5m = 29,  MRM6m = 30,  MRM7m = 31, // Format /4 /5 /6 /7
+
+    // MRMInitReg - This form is used for instructions whose source and
+    // destinations are the same register.
+    MRMInitReg = 32,
+
+    //// MRM_C1 - A mod/rm byte of exactly 0xC1.
+    MRM_C1 = 33,
+    MRM_C2 = 34,
+    MRM_C3 = 35,
+    MRM_C4 = 36,
+    MRM_C8 = 37,
+    MRM_C9 = 38,
+    MRM_E8 = 39,
+    MRM_F0 = 40,
+    MRM_F8 = 41,
+    MRM_F9 = 42,
+    MRM_D0 = 45,
+    MRM_D1 = 46,
+
+    /// RawFrmImm8 - This is used for the ENTER instruction, which has two
+    /// immediates, the first of which is a 16-bit immediate (specified by
+    /// the imm encoding) and the second is a 8-bit fixed value.
+    RawFrmImm8 = 43,
+
+    /// RawFrmImm16 - This is used for CALL FAR instructions, which have two
+    /// immediates, the first of which is a 16 or 32-bit immediate (specified by
+    /// the imm encoding) and the second is a 16-bit fixed value.  In the AMD
+    /// manual, this operand is described as pntr16:32 and pntr16:16
+    RawFrmImm16 = 44,
+
+    FormMask       = 63,
+
+    //===------------------------------------------------------------------===//
+    // Actual flags...
+
+    // OpSize - Set if this instruction requires an operand size prefix (0x66),
+    // which most often indicates that the instruction operates on 16 bit data
+    // instead of 32 bit data.
+    OpSize      = 1 << 6,
+
+    // AsSize - Set if this instruction requires an operand size prefix (0x67),
+    // which most often indicates that the instruction address 16 bit address
+    // instead of 32 bit address (or 32 bit address in 64 bit mode).
+    AdSize      = 1 << 7,
+
+    //===------------------------------------------------------------------===//
+    // Op0Mask - There are several prefix bytes that are used to form two byte
+    // opcodes.  These are currently 0x0F, 0xF3, and 0xD8-0xDF.  This mask is
+    // used to obtain the setting of this field.  If no bits in this field is
+    // set, there is no prefix byte for obtaining a multibyte opcode.
+    //
+    Op0Shift    = 8,
+    Op0Mask     = 0x1F << Op0Shift,
+
+    // TB - TwoByte - Set if this instruction has a two byte opcode, which
+    // starts with a 0x0F byte before the real opcode.
+    TB          = 1 << Op0Shift,
+
+    // REP - The 0xF3 prefix byte indicating repetition of the following
+    // instruction.
+    REP         = 2 << Op0Shift,
+
+    // D8-DF - These escape opcodes are used by the floating point unit.  These
+    // values must remain sequential.
+    D8 = 3 << Op0Shift,   D9 = 4 << Op0Shift,
+    DA = 5 << Op0Shift,   DB = 6 << Op0Shift,
+    DC = 7 << Op0Shift,   DD = 8 << Op0Shift,
+    DE = 9 << Op0Shift,   DF = 10 << Op0Shift,
+
+    // XS, XD - These prefix codes are for single and double precision scalar
+    // floating point operations performed in the SSE registers.
+    XD = 11 << Op0Shift,  XS = 12 << Op0Shift,
+
+    // T8, TA, A6, A7 - Prefix after the 0x0F prefix.
+    T8 = 13 << Op0Shift,  TA = 14 << Op0Shift,
+    A6 = 15 << Op0Shift,  A7 = 16 << Op0Shift,
+
+    // TF - Prefix before and after 0x0F
+    TF = 17 << Op0Shift,
+
+    //===------------------------------------------------------------------===//
+    // REX_W - REX prefixes are instruction prefixes used in 64-bit mode.
+    // They are used to specify GPRs and SSE registers, 64-bit operand size,
+    // etc. We only cares about REX.W and REX.R bits and only the former is
+    // statically determined.
+    //
+    REXShift    = Op0Shift + 5,
+    REX_W       = 1 << REXShift,
+
+    //===------------------------------------------------------------------===//
+    // This three-bit field describes the size of an immediate operand.  Zero is
+    // unused so that we can tell if we forgot to set a value.
+    ImmShift = REXShift + 1,
+    ImmMask    = 7 << ImmShift,
+    Imm8       = 1 << ImmShift,
+    Imm8PCRel  = 2 << ImmShift,
+    Imm16      = 3 << ImmShift,
+    Imm16PCRel = 4 << ImmShift,
+    Imm32      = 5 << ImmShift,
+    Imm32PCRel = 6 << ImmShift,
+    Imm64      = 7 << ImmShift,
+
+    //===------------------------------------------------------------------===//
+    // FP Instruction Classification...  Zero is non-fp instruction.
+
+    // FPTypeMask - Mask for all of the FP types...
+    FPTypeShift = ImmShift + 3,
+    FPTypeMask  = 7 << FPTypeShift,
+
+    // NotFP - The default, set for instructions that do not use FP registers.
+    NotFP      = 0 << FPTypeShift,
+
+    // ZeroArgFP - 0 arg FP instruction which implicitly pushes ST(0), f.e. fld0
+    ZeroArgFP  = 1 << FPTypeShift,
+
+    // OneArgFP - 1 arg FP instructions which implicitly read ST(0), such as fst
+    OneArgFP   = 2 << FPTypeShift,
+
+    // OneArgFPRW - 1 arg FP instruction which implicitly read ST(0) and write a
+    // result back to ST(0).  For example, fcos, fsqrt, etc.
+    //
+    OneArgFPRW = 3 << FPTypeShift,
+
+    // TwoArgFP - 2 arg FP instructions which implicitly read ST(0), and an
+    // explicit argument, storing the result to either ST(0) or the implicit
+    // argument.  For example: fadd, fsub, fmul, etc...
+    TwoArgFP   = 4 << FPTypeShift,
+
+    // CompareFP - 2 arg FP instructions which implicitly read ST(0) and an
+    // explicit argument, but have no destination.  Example: fucom, fucomi, ...
+    CompareFP  = 5 << FPTypeShift,
+
+    // CondMovFP - "2 operand" floating point conditional move instructions.
+    CondMovFP  = 6 << FPTypeShift,
+
+    // SpecialFP - Special instruction forms.  Dispatch by opcode explicitly.
+    SpecialFP  = 7 << FPTypeShift,
+
+    // Lock prefix
+    LOCKShift = FPTypeShift + 3,
+    LOCK = 1 << LOCKShift,
+
+    // Segment override prefixes. Currently we just need ability to address
+    // stuff in gs and fs segments.
+    SegOvrShift = LOCKShift + 1,
+    SegOvrMask  = 3 << SegOvrShift,
+    FS          = 1 << SegOvrShift,
+    GS          = 2 << SegOvrShift,
+
+    // Execution domain for SSE instructions in bits 23, 24.
+    // 0 in bits 23-24 means normal, non-SSE instruction.
+    SSEDomainShift = SegOvrShift + 2,
+
+    OpcodeShift   = SSEDomainShift + 2,
+
+    //===------------------------------------------------------------------===//
+    /// VEX - The opcode prefix used by AVX instructions
+    VEXShift = OpcodeShift + 8,
+    VEX         = 1U << 0,
+
+    /// VEX_W - Has a opcode specific functionality, but is used in the same
+    /// way as REX_W is for regular SSE instructions.
+    VEX_W       = 1U << 1,
+
+    /// VEX_4V - Used to specify an additional AVX/SSE register. Several 2
+    /// address instructions in SSE are represented as 3 address ones in AVX
+    /// and the additional register is encoded in VEX_VVVV prefix.
+    VEX_4V      = 1U << 2,
+
+    /// VEX_I8IMM - Specifies that the last register used in a AVX instruction,
+    /// must be encoded in the i8 immediate field. This usually happens in
+    /// instructions with 4 operands.
+    VEX_I8IMM   = 1U << 3,
+
+    /// VEX_L - Stands for a bit in the VEX opcode prefix meaning the current
+    /// instruction uses 256-bit wide registers. This is usually auto detected
+    /// if a VR256 register is used, but some AVX instructions also have this
+    /// field marked when using a f256 memory references.
+    VEX_L       = 1U << 4,
+
+    /// Has3DNow0F0FOpcode - This flag indicates that the instruction uses the
+    /// wacky 0x0F 0x0F prefix for 3DNow! instructions.  The manual documents
+    /// this as having a 0x0F prefix with a 0x0F opcode, and each instruction
+    /// storing a classifier in the imm8 field.  To simplify our implementation,
+    /// we handle this by storeing the classifier in the opcode field and using
+    /// this flag to indicate that the encoder should do the wacky 3DNow! thing.
+    Has3DNow0F0FOpcode = 1U << 5
+  };
+
+  // getBaseOpcodeFor - This function returns the "base" X86 opcode for the
+  // specified machine instruction.
+  //
+  static inline unsigned char getBaseOpcodeFor(uint64_t TSFlags) {
+    return TSFlags >> X86II::OpcodeShift;
+  }
+
+  static inline bool hasImm(uint64_t TSFlags) {
+    return (TSFlags & X86II::ImmMask) != 0;
+  }
+
+  /// getSizeOfImm - Decode the "size of immediate" field from the TSFlags field
+  /// of the specified instruction.
+  static inline unsigned getSizeOfImm(uint64_t TSFlags) {
+    switch (TSFlags & X86II::ImmMask) {
+    default: assert(0 && "Unknown immediate size");
+    case X86II::Imm8:
+    case X86II::Imm8PCRel:  return 1;
+    case X86II::Imm16:
+    case X86II::Imm16PCRel: return 2;
+    case X86II::Imm32:
+    case X86II::Imm32PCRel: return 4;
+    case X86II::Imm64:      return 8;
+    }
+  }
+
+  /// isImmPCRel - Return true if the immediate of the specified instruction's
+  /// TSFlags indicates that it is pc relative.
+  static inline unsigned isImmPCRel(uint64_t TSFlags) {
+    switch (TSFlags & X86II::ImmMask) {
+    default: assert(0 && "Unknown immediate size");
+    case X86II::Imm8PCRel:
+    case X86II::Imm16PCRel:
+    case X86II::Imm32PCRel:
+      return true;
+    case X86II::Imm8:
+    case X86II::Imm16:
+    case X86II::Imm32:
+    case X86II::Imm64:
+      return false;
+    }
+  }
+
+  /// getMemoryOperandNo - The function returns the MCInst operand # for the
+  /// first field of the memory operand.  If the instruction doesn't have a
+  /// memory operand, this returns -1.
+  ///
+  /// Note that this ignores tied operands.  If there is a tied register which
+  /// is duplicated in the MCInst (e.g. "EAX = addl EAX, [mem]") it is only
+  /// counted as one operand.
+  ///
+  static inline int getMemoryOperandNo(uint64_t TSFlags) {
+    switch (TSFlags & X86II::FormMask) {
+    case X86II::MRMInitReg:  assert(0 && "FIXME: Remove this form");
+    default: assert(0 && "Unknown FormMask value in getMemoryOperandNo!");
+    case X86II::Pseudo:
+    case X86II::RawFrm:
+    case X86II::AddRegFrm:
+    case X86II::MRMDestReg:
+    case X86II::MRMSrcReg:
+    case X86II::RawFrmImm8:
+    case X86II::RawFrmImm16:
+       return -1;
+    case X86II::MRMDestMem:
+      return 0;
+    case X86II::MRMSrcMem: {
+      bool HasVEX_4V = (TSFlags >> X86II::VEXShift) & X86II::VEX_4V;
+      unsigned FirstMemOp = 1;
+      if (HasVEX_4V)
+        ++FirstMemOp;// Skip the register source (which is encoded in VEX_VVVV).
+
+      // FIXME: Maybe lea should have its own form?  This is a horrible hack.
+      //if (Opcode == X86::LEA64r || Opcode == X86::LEA64_32r ||
+      //    Opcode == X86::LEA16r || Opcode == X86::LEA32r)
+      return FirstMemOp;
+    }
+    case X86II::MRM0r: case X86II::MRM1r:
+    case X86II::MRM2r: case X86II::MRM3r:
+    case X86II::MRM4r: case X86II::MRM5r:
+    case X86II::MRM6r: case X86II::MRM7r:
+      return -1;
+    case X86II::MRM0m: case X86II::MRM1m:
+    case X86II::MRM2m: case X86II::MRM3m:
+    case X86II::MRM4m: case X86II::MRM5m:
+    case X86II::MRM6m: case X86II::MRM7m:
+      return 0;
+    case X86II::MRM_C1:
+    case X86II::MRM_C2:
+    case X86II::MRM_C3:
+    case X86II::MRM_C4:
+    case X86II::MRM_C8:
+    case X86II::MRM_C9:
+    case X86II::MRM_E8:
+    case X86II::MRM_F0:
+    case X86II::MRM_F8:
+    case X86II::MRM_F9:
+    case X86II::MRM_D0:
+    case X86II::MRM_D1:
+      return -1;
+    }
+  }
+
+  /// isX86_64ExtendedReg - Is the MachineOperand a x86-64 extended (r8 or
+  /// higher) register?  e.g. r8, xmm8, xmm13, etc.
+  static inline bool isX86_64ExtendedReg(unsigned RegNo) {
+    switch (RegNo) {
+    default: break;
+    case X86::R8:    case X86::R9:    case X86::R10:   case X86::R11:
+    case X86::R12:   case X86::R13:   case X86::R14:   case X86::R15:
+    case X86::R8D:   case X86::R9D:   case X86::R10D:  case X86::R11D:
+    case X86::R12D:  case X86::R13D:  case X86::R14D:  case X86::R15D:
+    case X86::R8W:   case X86::R9W:   case X86::R10W:  case X86::R11W:
+    case X86::R12W:  case X86::R13W:  case X86::R14W:  case X86::R15W:
+    case X86::R8B:   case X86::R9B:   case X86::R10B:  case X86::R11B:
+    case X86::R12B:  case X86::R13B:  case X86::R14B:  case X86::R15B:
+    case X86::XMM8:  case X86::XMM9:  case X86::XMM10: case X86::XMM11:
+    case X86::XMM12: case X86::XMM13: case X86::XMM14: case X86::XMM15:
+    case X86::YMM8:  case X86::YMM9:  case X86::YMM10: case X86::YMM11:
+    case X86::YMM12: case X86::YMM13: case X86::YMM14: case X86::YMM15:
+    case X86::CR8:   case X86::CR9:   case X86::CR10:  case X86::CR11:
+    case X86::CR12:  case X86::CR13:  case X86::CR14:  case X86::CR15:
+        return true;
+    }
+    return false;
+  }
+  
+  static inline bool isX86_64NonExtLowByteReg(unsigned reg) {
+    return (reg == X86::SPL || reg == X86::BPL ||
+            reg == X86::SIL || reg == X86::DIL);
+  }
+}
+
+} // end namespace llvm;
+
+#endif

lib/Target/X86/MCTargetDesc/X86MCTargetDesc.h

@@ -17,6 +17,9 @@
 #include <string>
 
 namespace llvm {
+class MCCodeEmitter;
+class MCContext;
+class MCInstrInfo;
 class MCRegisterInfo;
 class MCSubtargetInfo;
 class Target;
@@ -63,6 +66,11 @@ namespace X86_MC {
                                             StringRef FS);
 }
 
+MCCodeEmitter *createX86MCCodeEmitter(const MCInstrInfo &MCII,
+                                      const MCSubtargetInfo &STI,
+                                      MCContext &Ctx);
+
+
 } // End llvm namespace
 
 

lib/Target/X86/X86.h

@@ -15,6 +15,7 @@
 #ifndef TARGET_X86_H
 #define TARGET_X86_H
 
+#include "MCTargetDesc/X86BaseInfo.h"
 #include "MCTargetDesc/X86MCTargetDesc.h"
 #include "llvm/Support/DataTypes.h"
 #include "llvm/Target/TargetMachine.h"
@@ -60,10 +61,6 @@ FunctionPass *createSSEDomainFixPass();
 FunctionPass *createX86JITCodeEmitterPass(X86TargetMachine &TM,
                                           JITCodeEmitter &JCE);
 
-MCCodeEmitter *createX86MCCodeEmitter(const MCInstrInfo &MCII,
-                                      const MCSubtargetInfo &STI,
-                                      MCContext &Ctx);
-
 TargetAsmBackend *createX86_32AsmBackend(const Target &, const std::string &);
 TargetAsmBackend *createX86_64AsmBackend(const Target &, const std::string &);
 

lib/Target/X86/X86AsmBackend.cpp

@@ -9,7 +9,7 @@
 
 #include "llvm/MC/TargetAsmBackend.h"
 #include "X86.h"
-#include "X86FixupKinds.h"
+#include "MCTargetDesc/X86FixupKinds.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCELFObjectWriter.h"

lib/Target/X86/X86CodeEmitter.cpp

@@ -167,7 +167,7 @@ static unsigned determineREX(const MachineInstr &MI) {
       const MachineOperand& MO = MI.getOperand(i);
       if (MO.isReg()) {
         unsigned Reg = MO.getReg();
-        if (X86InstrInfo::isX86_64NonExtLowByteReg(Reg))
+        if (X86II::isX86_64NonExtLowByteReg(Reg))
           REX |= 0x40;
       }
     }

lib/Target/X86/X86InstrInfo.cpp

@@ -3013,31 +3013,6 @@ isSafeToMoveRegClassDefs(const TargetRegisterClass *RC) const {
            RC == &X86::RFP64RegClass || RC == &X86::RFP80RegClass);
 }
 
-
-/// isX86_64ExtendedReg - Is the MachineOperand a x86-64 extended (r8 or higher)
-/// register?  e.g. r8, xmm8, xmm13, etc.
-bool X86InstrInfo::isX86_64ExtendedReg(unsigned RegNo) {
-  switch (RegNo) {
-  default: break;
-  case X86::R8:    case X86::R9:    case X86::R10:   case X86::R11:
-  case X86::R12:   case X86::R13:   case X86::R14:   case X86::R15:
-  case X86::R8D:   case X86::R9D:   case X86::R10D:  case X86::R11D:
-  case X86::R12D:  case X86::R13D:  case X86::R14D:  case X86::R15D:
-  case X86::R8W:   case X86::R9W:   case X86::R10W:  case X86::R11W:
-  case X86::R12W:  case X86::R13W:  case X86::R14W:  case X86::R15W:
-  case X86::R8B:   case X86::R9B:   case X86::R10B:  case X86::R11B:
-  case X86::R12B:  case X86::R13B:  case X86::R14B:  case X86::R15B:
-  case X86::XMM8:  case X86::XMM9:  case X86::XMM10: case X86::XMM11:
-  case X86::XMM12: case X86::XMM13: case X86::XMM14: case X86::XMM15:
-  case X86::YMM8:  case X86::YMM9:  case X86::YMM10: case X86::YMM11:
-  case X86::YMM12: case X86::YMM13: case X86::YMM14: case X86::YMM15:
-  case X86::CR8:   case X86::CR9:   case X86::CR10:  case X86::CR11:
-  case X86::CR12:  case X86::CR13:  case X86::CR14:  case X86::CR15:
-    return true;
-  }
-  return false;
-}
-
 /// getGlobalBaseReg - Return a virtual register initialized with the
 /// the global base register value. Output instructions required to
 /// initialize the register in the function entry block, if necessary.

lib/Target/X86/X86InstrInfo.h

@@ -27,24 +27,6 @@ namespace llvm {
   class X86TargetMachine;
 
 namespace X86 {
-  // Enums for memory operand decoding.  Each memory operand is represented with
-  // a 5 operand sequence in the form:
-  //   [BaseReg, ScaleAmt, IndexReg, Disp, Segment]
-  // These enums help decode this.
-  enum {
-    AddrBaseReg = 0,
-    AddrScaleAmt = 1,
-    AddrIndexReg = 2,
-    AddrDisp = 3,
-
-    /// AddrSegmentReg - The operand # of the segment in the memory operand.
-    AddrSegmentReg = 4,
-
-    /// AddrNumOperands - Total number of operands in a memory reference.
-    AddrNumOperands = 5
-  };
-
-
   // X86 specific condition code. These correspond to X86_*_COND in
   // X86InstrInfo.td. They must be kept in synch.
   enum CondCode {
@@ -82,133 +64,8 @@ namespace X86 {
   /// GetOppositeBranchCondition - Return the inverse of the specified cond,
   /// e.g. turning COND_E to COND_NE.
   CondCode GetOppositeBranchCondition(X86::CondCode CC);
+}  // end namespace X86;
 
-}
-
-/// X86II - This namespace holds all of the target specific flags that
-/// instruction info tracks.
-///
-namespace X86II {
-  /// Target Operand Flag enum.
-  enum TOF {
-    //===------------------------------------------------------------------===//
-    // X86 Specific MachineOperand flags.
-
-    MO_NO_FLAG,
-
-    /// MO_GOT_ABSOLUTE_ADDRESS - On a symbol operand, this represents a
-    /// relocation of:
-    ///    SYMBOL_LABEL + [. - PICBASELABEL]
-    MO_GOT_ABSOLUTE_ADDRESS,
-
-    /// MO_PIC_BASE_OFFSET - On a symbol operand this indicates that the
-    /// immediate should get the value of the symbol minus the PIC base label:
-    ///    SYMBOL_LABEL - PICBASELABEL
-    MO_PIC_BASE_OFFSET,
-
-    /// MO_GOT - On a symbol operand this indicates that the immediate is the
-    /// offset to the GOT entry for the symbol name from the base of the GOT.
-    ///
-    /// See the X86-64 ELF ABI supplement for more details.
-    ///    SYMBOL_LABEL @GOT
-    MO_GOT,
-
-    /// MO_GOTOFF - On a symbol operand this indicates that the immediate is
-    /// the offset to the location of the symbol name from the base of the GOT.
-    ///
-    /// See the X86-64 ELF ABI supplement for more details.
-    ///    SYMBOL_LABEL @GOTOFF
-    MO_GOTOFF,
-
-    /// MO_GOTPCREL - On a symbol operand this indicates that the immediate is
-    /// offset to the GOT entry for the symbol name from the current code
-    /// location.
-    ///
-    /// See the X86-64 ELF ABI supplement for more details.
-    ///    SYMBOL_LABEL @GOTPCREL
-    MO_GOTPCREL,
-
-    /// MO_PLT - On a symbol operand this indicates that the immediate is
-    /// offset to the PLT entry of symbol name from the current code location.
-    ///
-    /// See the X86-64 ELF ABI supplement for more details.
-    ///    SYMBOL_LABEL @PLT
-    MO_PLT,
-
-    /// MO_TLSGD - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// See 'ELF Handling for Thread-Local Storage' for more details.
-    ///    SYMBOL_LABEL @TLSGD
-    MO_TLSGD,
-
-    /// MO_GOTTPOFF - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// See 'ELF Handling for Thread-Local Storage' for more details.
-    ///    SYMBOL_LABEL @GOTTPOFF
-    MO_GOTTPOFF,
-
-    /// MO_INDNTPOFF - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// See 'ELF Handling for Thread-Local Storage' for more details.
-    ///    SYMBOL_LABEL @INDNTPOFF
-    MO_INDNTPOFF,
-
-    /// MO_TPOFF - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// See 'ELF Handling for Thread-Local Storage' for more details.
-    ///    SYMBOL_LABEL @TPOFF
-    MO_TPOFF,
-
-    /// MO_NTPOFF - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// See 'ELF Handling for Thread-Local Storage' for more details.
-    ///    SYMBOL_LABEL @NTPOFF
-    MO_NTPOFF,
-
-    /// MO_DLLIMPORT - On a symbol operand "FOO", this indicates that the
-    /// reference is actually to the "__imp_FOO" symbol.  This is used for
-    /// dllimport linkage on windows.
-    MO_DLLIMPORT,
-
-    /// MO_DARWIN_STUB - On a symbol operand "FOO", this indicates that the
-    /// reference is actually to the "FOO$stub" symbol.  This is used for calls
-    /// and jumps to external functions on Tiger and earlier.
-    MO_DARWIN_STUB,
-
-    /// MO_DARWIN_NONLAZY - On a symbol operand "FOO", this indicates that the
-    /// reference is actually to the "FOO$non_lazy_ptr" symbol, which is a
-    /// non-PIC-base-relative reference to a non-hidden dyld lazy pointer stub.
-    MO_DARWIN_NONLAZY,
-
-    /// MO_DARWIN_NONLAZY_PIC_BASE - On a symbol operand "FOO", this indicates
-    /// that the reference is actually to "FOO$non_lazy_ptr - PICBASE", which is
-    /// a PIC-base-relative reference to a non-hidden dyld lazy pointer stub.
-    MO_DARWIN_NONLAZY_PIC_BASE,
-
-    /// MO_DARWIN_HIDDEN_NONLAZY_PIC_BASE - On a symbol operand "FOO", this
-    /// indicates that the reference is actually to "FOO$non_lazy_ptr -PICBASE",
-    /// which is a PIC-base-relative reference to a hidden dyld lazy pointer
-    /// stub.
-    MO_DARWIN_HIDDEN_NONLAZY_PIC_BASE,
-
-    /// MO_TLVP - On a symbol operand this indicates that the immediate is
-    /// some TLS offset.
-    ///
-    /// This is the TLS offset for the Darwin TLS mechanism.
-    MO_TLVP,
-
-    /// MO_TLVP_PIC_BASE - On a symbol operand this indicates that the immediate
-    /// is some TLS offset from the picbase.
-    ///
-    /// This is the 32-bit TLS offset for Darwin TLS in PIC mode.
-    MO_TLVP_PIC_BASE
-  };
-}
 
 /// isGlobalStubReference - Return true if the specified TargetFlag operand is
 /// a reference to a stub for a global, not the global itself.
@@ -243,353 +100,6 @@ inline static bool isGlobalRelativeToPICBase(unsigned char TargetFlag) {
   }
 }
 
-/// X86II - This namespace holds all of the target specific flags that
-/// instruction info tracks.
-///
-namespace X86II {
-  enum {
-    //===------------------------------------------------------------------===//
-    // Instruction encodings.  These are the standard/most common forms for X86
-    // instructions.
-    //
-
-    // PseudoFrm - This represents an instruction that is a pseudo instruction
-    // or one that has not been implemented yet.  It is illegal to code generate
-    // it, but tolerated for intermediate implementation stages.
-    Pseudo         = 0,
-
-    /// Raw - This form is for instructions that don't have any operands, so
-    /// they are just a fixed opcode value, like 'leave'.
-    RawFrm         = 1,
-
-    /// AddRegFrm - This form is used for instructions like 'push r32' that have
-    /// their one register operand added to their opcode.
-    AddRegFrm      = 2,
-
-    /// MRMDestReg - This form is used for instructions that use the Mod/RM byte
-    /// to specify a destination, which in this case is a register.
-    ///
-    MRMDestReg     = 3,
-
-    /// MRMDestMem - This form is used for instructions that use the Mod/RM byte
-    /// to specify a destination, which in this case is memory.
-    ///
-    MRMDestMem     = 4,
-
-    /// MRMSrcReg - This form is used for instructions that use the Mod/RM byte
-    /// to specify a source, which in this case is a register.
-    ///
-    MRMSrcReg      = 5,
-
-    /// MRMSrcMem - This form is used for instructions that use the Mod/RM byte
-    /// to specify a source, which in this case is memory.
-    ///
-    MRMSrcMem      = 6,
-
-    /// MRM[0-7][rm] - These forms are used to represent instructions that use
-    /// a Mod/RM byte, and use the middle field to hold extended opcode
-    /// information.  In the intel manual these are represented as /0, /1, ...
-    ///
-
-    // First, instructions that operate on a register r/m operand...
-    MRM0r = 16,  MRM1r = 17,  MRM2r = 18,  MRM3r = 19, // Format /0 /1 /2 /3
-    MRM4r = 20,  MRM5r = 21,  MRM6r = 22,  MRM7r = 23, // Format /4 /5 /6 /7
-
-    // Next, instructions that operate on a memory r/m operand...
-    MRM0m = 24,  MRM1m = 25,  MRM2m = 26,  MRM3m = 27, // Format /0 /1 /2 /3
-    MRM4m = 28,  MRM5m = 29,  MRM6m = 30,  MRM7m = 31, // Format /4 /5 /6 /7
-
-    // MRMInitReg - This form is used for instructions whose source and
-    // destinations are the same register.
-    MRMInitReg = 32,
-
-    //// MRM_C1 - A mod/rm byte of exactly 0xC1.
-    MRM_C1 = 33,
-    MRM_C2 = 34,
-    MRM_C3 = 35,
-    MRM_C4 = 36,
-    MRM_C8 = 37,
-    MRM_C9 = 38,
-    MRM_E8 = 39,
-    MRM_F0 = 40,
-    MRM_F8 = 41,
-    MRM_F9 = 42,
-    MRM_D0 = 45,
-    MRM_D1 = 46,
-
-    /// RawFrmImm8 - This is used for the ENTER instruction, which has two
-    /// immediates, the first of which is a 16-bit immediate (specified by
-    /// the imm encoding) and the second is a 8-bit fixed value.
-    RawFrmImm8 = 43,
-
-    /// RawFrmImm16 - This is used for CALL FAR instructions, which have two
-    /// immediates, the first of which is a 16 or 32-bit immediate (specified by
-    /// the imm encoding) and the second is a 16-bit fixed value.  In the AMD
-    /// manual, this operand is described as pntr16:32 and pntr16:16
-    RawFrmImm16 = 44,
-
-    FormMask       = 63,
-
-    //===------------------------------------------------------------------===//
-    // Actual flags...
-
-    // OpSize - Set if this instruction requires an operand size prefix (0x66),
-    // which most often indicates that the instruction operates on 16 bit data
-    // instead of 32 bit data.
-    OpSize      = 1 << 6,
-
-    // AsSize - Set if this instruction requires an operand size prefix (0x67),
-    // which most often indicates that the instruction address 16 bit address
-    // instead of 32 bit address (or 32 bit address in 64 bit mode).
-    AdSize      = 1 << 7,
-
-    //===------------------------------------------------------------------===//
-    // Op0Mask - There are several prefix bytes that are used to form two byte
-    // opcodes.  These are currently 0x0F, 0xF3, and 0xD8-0xDF.  This mask is
-    // used to obtain the setting of this field.  If no bits in this field is
-    // set, there is no prefix byte for obtaining a multibyte opcode.
-    //
-    Op0Shift    = 8,
-    Op0Mask     = 0x1F << Op0Shift,
-
-    // TB - TwoByte - Set if this instruction has a two byte opcode, which
-    // starts with a 0x0F byte before the real opcode.
-    TB          = 1 << Op0Shift,
-
-    // REP - The 0xF3 prefix byte indicating repetition of the following
-    // instruction.
-    REP         = 2 << Op0Shift,
-
-    // D8-DF - These escape opcodes are used by the floating point unit.  These
-    // values must remain sequential.
-    D8 = 3 << Op0Shift,   D9 = 4 << Op0Shift,
-    DA = 5 << Op0Shift,   DB = 6 << Op0Shift,
-    DC = 7 << Op0Shift,   DD = 8 << Op0Shift,
-    DE = 9 << Op0Shift,   DF = 10 << Op0Shift,
-
-    // XS, XD - These prefix codes are for single and double precision scalar
-    // floating point operations performed in the SSE registers.
-    XD = 11 << Op0Shift,  XS = 12 << Op0Shift,
-
-    // T8, TA, A6, A7 - Prefix after the 0x0F prefix.
-    T8 = 13 << Op0Shift,  TA = 14 << Op0Shift,
-    A6 = 15 << Op0Shift,  A7 = 16 << Op0Shift,
-
-    // TF - Prefix before and after 0x0F
-    TF = 17 << Op0Shift,
-
-    //===------------------------------------------------------------------===//
-    // REX_W - REX prefixes are instruction prefixes used in 64-bit mode.
-    // They are used to specify GPRs and SSE registers, 64-bit operand size,
-    // etc. We only cares about REX.W and REX.R bits and only the former is
-    // statically determined.
-    //
-    REXShift    = Op0Shift + 5,
-    REX_W       = 1 << REXShift,
-
-    //===------------------------------------------------------------------===//
-    // This three-bit field describes the size of an immediate operand.  Zero is
-    // unused so that we can tell if we forgot to set a value.
-    ImmShift = REXShift + 1,
-    ImmMask    = 7 << ImmShift,
-    Imm8       = 1 << ImmShift,
-    Imm8PCRel  = 2 << ImmShift,
-    Imm16      = 3 << ImmShift,
-    Imm16PCRel = 4 << ImmShift,
-    Imm32      = 5 << ImmShift,
-    Imm32PCRel = 6 << ImmShift,
-    Imm64      = 7 << ImmShift,
-
-    //===------------------------------------------------------------------===//
-    // FP Instruction Classification...  Zero is non-fp instruction.
-
-    // FPTypeMask - Mask for all of the FP types...
-    FPTypeShift = ImmShift + 3,
-    FPTypeMask  = 7 << FPTypeShift,
-
-    // NotFP - The default, set for instructions that do not use FP registers.
-    NotFP      = 0 << FPTypeShift,
-
-    // ZeroArgFP - 0 arg FP instruction which implicitly pushes ST(0), f.e. fld0
-    ZeroArgFP  = 1 << FPTypeShift,
-
-    // OneArgFP - 1 arg FP instructions which implicitly read ST(0), such as fst
-    OneArgFP   = 2 << FPTypeShift,
-
-    // OneArgFPRW - 1 arg FP instruction which implicitly read ST(0) and write a
-    // result back to ST(0).  For example, fcos, fsqrt, etc.
-    //
-    OneArgFPRW = 3 << FPTypeShift,
-
-    // TwoArgFP - 2 arg FP instructions which implicitly read ST(0), and an
-    // explicit argument, storing the result to either ST(0) or the implicit
-    // argument.  For example: fadd, fsub, fmul, etc...
-    TwoArgFP   = 4 << FPTypeShift,
-
-    // CompareFP - 2 arg FP instructions which implicitly read ST(0) and an
-    // explicit argument, but have no destination.  Example: fucom, fucomi, ...
-    CompareFP  = 5 << FPTypeShift,
-
-    // CondMovFP - "2 operand" floating point conditional move instructions.
-    CondMovFP  = 6 << FPTypeShift,
-
-    // SpecialFP - Special instruction forms.  Dispatch by opcode explicitly.
-    SpecialFP  = 7 << FPTypeShift,
-
-    // Lock prefix
-    LOCKShift = FPTypeShift + 3,
-    LOCK = 1 << LOCKShift,
-
-    // Segment override prefixes. Currently we just need ability to address
-    // stuff in gs and fs segments.
-    SegOvrShift = LOCKShift + 1,
-    SegOvrMask  = 3 << SegOvrShift,
-    FS          = 1 << SegOvrShift,
-    GS          = 2 << SegOvrShift,
-
-    // Execution domain for SSE instructions in bits 23, 24.
-    // 0 in bits 23-24 means normal, non-SSE instruction.
-    SSEDomainShift = SegOvrShift + 2,
-
-    OpcodeShift   = SSEDomainShift + 2,
-
-    //===------------------------------------------------------------------===//
-    /// VEX - The opcode prefix used by AVX instructions
-    VEXShift = OpcodeShift + 8,
-    VEX         = 1U << 0,
-
-    /// VEX_W - Has a opcode specific functionality, but is used in the same
-    /// way as REX_W is for regular SSE instructions.
-    VEX_W       = 1U << 1,
-
-    /// VEX_4V - Used to specify an additional AVX/SSE register. Several 2
-    /// address instructions in SSE are represented as 3 address ones in AVX
-    /// and the additional register is encoded in VEX_VVVV prefix.
-    VEX_4V      = 1U << 2,
-
-    /// VEX_I8IMM - Specifies that the last register used in a AVX instruction,
-    /// must be encoded in the i8 immediate field. This usually happens in
-    /// instructions with 4 operands.
-    VEX_I8IMM   = 1U << 3,
-
-    /// VEX_L - Stands for a bit in the VEX opcode prefix meaning the current
-    /// instruction uses 256-bit wide registers. This is usually auto detected
-    /// if a VR256 register is used, but some AVX instructions also have this
-    /// field marked when using a f256 memory references.
-    VEX_L       = 1U << 4,
-
-    /// Has3DNow0F0FOpcode - This flag indicates that the instruction uses the
-    /// wacky 0x0F 0x0F prefix for 3DNow! instructions.  The manual documents
-    /// this as having a 0x0F prefix with a 0x0F opcode, and each instruction
-    /// storing a classifier in the imm8 field.  To simplify our implementation,
-    /// we handle this by storeing the classifier in the opcode field and using
-    /// this flag to indicate that the encoder should do the wacky 3DNow! thing.
-    Has3DNow0F0FOpcode = 1U << 5
-  };
-
-  // getBaseOpcodeFor - This function returns the "base" X86 opcode for the
-  // specified machine instruction.
-  //
-  static inline unsigned char getBaseOpcodeFor(uint64_t TSFlags) {
-    return TSFlags >> X86II::OpcodeShift;
-  }
-
-  static inline bool hasImm(uint64_t TSFlags) {
-    return (TSFlags & X86II::ImmMask) != 0;
-  }
-
-  /// getSizeOfImm - Decode the "size of immediate" field from the TSFlags field
-  /// of the specified instruction.
-  static inline unsigned getSizeOfImm(uint64_t TSFlags) {
-    switch (TSFlags & X86II::ImmMask) {
-    default: assert(0 && "Unknown immediate size");
-    case X86II::Imm8:
-    case X86II::Imm8PCRel:  return 1;
-    case X86II::Imm16:
-    case X86II::Imm16PCRel: return 2;
-    case X86II::Imm32:
-    case X86II::Imm32PCRel: return 4;
-    case X86II::Imm64:      return 8;
-    }
-  }
-
-  /// isImmPCRel - Return true if the immediate of the specified instruction's
-  /// TSFlags indicates that it is pc relative.
-  static inline unsigned isImmPCRel(uint64_t TSFlags) {
-    switch (TSFlags & X86II::ImmMask) {
-    default: assert(0 && "Unknown immediate size");
-    case X86II::Imm8PCRel:
-    case X86II::Imm16PCRel:
-    case X86II::Imm32PCRel:
-      return true;
-    case X86II::Imm8:
-    case X86II::Imm16:
-    case X86II::Imm32:
-    case X86II::Imm64:
-      return false;
-    }
-  }
-
-  /// getMemoryOperandNo - The function returns the MCInst operand # for the
-  /// first field of the memory operand.  If the instruction doesn't have a
-  /// memory operand, this returns -1.
-  ///
-  /// Note that this ignores tied operands.  If there is a tied register which
-  /// is duplicated in the MCInst (e.g. "EAX = addl EAX, [mem]") it is only
-  /// counted as one operand.
-  ///
-  static inline int getMemoryOperandNo(uint64_t TSFlags) {
-    switch (TSFlags & X86II::FormMask) {
-    case X86II::MRMInitReg:  assert(0 && "FIXME: Remove this form");
-    default: assert(0 && "Unknown FormMask value in getMemoryOperandNo!");
-    case X86II::Pseudo:
-    case X86II::RawFrm:
-    case X86II::AddRegFrm:
-    case X86II::MRMDestReg:
-    case X86II::MRMSrcReg:
-    case X86II::RawFrmImm8:
-    case X86II::RawFrmImm16:
-       return -1;
-    case X86II::MRMDestMem:
-      return 0;
-    case X86II::MRMSrcMem: {
-      bool HasVEX_4V = (TSFlags >> X86II::VEXShift) & X86II::VEX_4V;
-      unsigned FirstMemOp = 1;
-      if (HasVEX_4V)
-        ++FirstMemOp;// Skip the register source (which is encoded in VEX_VVVV).
-
-      // FIXME: Maybe lea should have its own form?  This is a horrible hack.
-      //if (Opcode == X86::LEA64r || Opcode == X86::LEA64_32r ||
-      //    Opcode == X86::LEA16r || Opcode == X86::LEA32r)
-      return FirstMemOp;
-    }
-    case X86II::MRM0r: case X86II::MRM1r:
-    case X86II::MRM2r: case X86II::MRM3r:
-    case X86II::MRM4r: case X86II::MRM5r:
-    case X86II::MRM6r: case X86II::MRM7r:
-      return -1;
-    case X86II::MRM0m: case X86II::MRM1m:
-    case X86II::MRM2m: case X86II::MRM3m:
-    case X86II::MRM4m: case X86II::MRM5m:
-    case X86II::MRM6m: case X86II::MRM7m:
-      return 0;
-    case X86II::MRM_C1:
-    case X86II::MRM_C2:
-    case X86II::MRM_C3:
-    case X86II::MRM_C4:
-    case X86II::MRM_C8:
-    case X86II::MRM_C9:
-    case X86II::MRM_E8:
-    case X86II::MRM_F0:
-    case X86II::MRM_F8:
-    case X86II::MRM_F9:
-    case X86II::MRM_D0:
-    case X86II::MRM_D1:
-      return -1;
-    }
-  }
-}
-
 inline static bool isScale(const MachineOperand &MO) {
   return MO.isImm() &&
     (MO.getImm() == 1 || MO.getImm() == 2 ||
@@ -829,20 +339,11 @@ public:
   /// instruction that defines the specified register class.
   bool isSafeToMoveRegClassDefs(const TargetRegisterClass *RC) const;
 
-  static bool isX86_64NonExtLowByteReg(unsigned reg) {
-    return (reg == X86::SPL || reg == X86::BPL ||
-          reg == X86::SIL || reg == X86::DIL);
-  }
-
   static bool isX86_64ExtendedReg(const MachineOperand &MO) {
     if (!MO.isReg()) return false;
-    return isX86_64ExtendedReg(MO.getReg());
+    return X86II::isX86_64ExtendedReg(MO.getReg());
   }
 
-  /// isX86_64ExtendedReg - Is the MachineOperand a x86-64 extended (r8 or
-  /// higher) register?  e.g. r8, xmm8, xmm13, etc.
-  static bool isX86_64ExtendedReg(unsigned RegNo);
-
   /// getGlobalBaseReg - Return a virtual register initialized with the
   /// the global base register value. Output instructions required to
   /// initialize the register in the function entry block, if necessary.

lib/Target/X86/X86MCCodeEmitter.cpp

@@ -12,12 +12,14 @@
 //===----------------------------------------------------------------------===//
 
 #define DEBUG_TYPE "mccodeemitter"
-#include "X86.h"
+#include "MCTargetDesc/X86MCTargetDesc.h"
-#include "X86InstrInfo.h"
+#include "MCTargetDesc/X86BaseInfo.h"
-#include "X86FixupKinds.h"
+#include "MCTargetDesc/X86FixupKinds.h"
 #include "llvm/MC/MCCodeEmitter.h"
 #include "llvm/MC/MCExpr.h"
 #include "llvm/MC/MCInst.h"
+#include "llvm/MC/MCInstrInfo.h"
+#include "llvm/MC/MCRegisterInfo.h"
 #include "llvm/MC/MCSubtargetInfo.h"
 #include "llvm/MC/MCSymbol.h"
 #include "llvm/Support/raw_ostream.h"
@@ -153,6 +155,11 @@ static MCFixupKind getImmFixupKind(uint64_t TSFlags) {
   return MCFixup::getKindForSize(Size, isPCRel);
 }
 
+namespace llvm {
+  // FIXME: TableGen this?
+  extern MCRegisterClass X86MCRegisterClasses[]; // In X86GenRegisterInfo.inc.
+}
+
 /// Is32BitMemOperand - Return true if the specified instruction with a memory
 /// operand should emit the 0x67 prefix byte in 64-bit mode due to a 32-bit
 /// memory operand.  Op specifies the operand # of the memoperand.
@@ -160,8 +167,10 @@ static bool Is32BitMemOperand(const MCInst &MI, unsigned Op) {
   const MCOperand &BaseReg  = MI.getOperand(Op+X86::AddrBaseReg);
   const MCOperand &IndexReg = MI.getOperand(Op+X86::AddrIndexReg);
   
-  if ((BaseReg.getReg() != 0 && X86::GR32RegClass.contains(BaseReg.getReg())) ||
+  if ((BaseReg.getReg() != 0 &&
-      (IndexReg.getReg() != 0 && X86::GR32RegClass.contains(IndexReg.getReg())))
+       X86MCRegisterClasses[X86::GR32RegClassID].contains(BaseReg.getReg())) ||
+      (IndexReg.getReg() != 0 &&
+       X86MCRegisterClasses[X86::GR32RegClassID].contains(IndexReg.getReg())))
     return true;
   return false;
 }
@@ -506,7 +515,7 @@ void X86MCCodeEmitter::EmitVEXOpcodePrefix(uint64_t TSFlags, unsigned &CurByte,
   case X86II::MRMSrcMem:
   case X86II::MRMSrcReg:
     if (MI.getNumOperands() > CurOp && MI.getOperand(CurOp).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(CurOp).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(CurOp).getReg()))
       VEX_R = 0x0;
     CurOp++;
 
@@ -527,11 +536,11 @@ void X86MCCodeEmitter::EmitVEXOpcodePrefix(uint64_t TSFlags, unsigned &CurByte,
 
     for (; CurOp != NumOps; ++CurOp) {
       const MCOperand &MO = MI.getOperand(CurOp);
-      if (MO.isReg() && X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+      if (MO.isReg() && X86II::isX86_64ExtendedReg(MO.getReg()))
         VEX_B = 0x0;
       if (!VEX_B && MO.isReg() &&
           ((TSFlags & X86II::FormMask) == X86II::MRMSrcMem) &&
-          X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+          X86II::isX86_64ExtendedReg(MO.getReg()))
         VEX_X = 0x0;
     }
     break;
@@ -540,7 +549,7 @@ void X86MCCodeEmitter::EmitVEXOpcodePrefix(uint64_t TSFlags, unsigned &CurByte,
       break;
 
     if (MI.getOperand(CurOp).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(CurOp).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(CurOp).getReg()))
       VEX_B = 0;
 
     if (HasVEX_4V)
@@ -550,7 +559,7 @@ void X86MCCodeEmitter::EmitVEXOpcodePrefix(uint64_t TSFlags, unsigned &CurByte,
     for (; CurOp != NumOps; ++CurOp) {
       const MCOperand &MO = MI.getOperand(CurOp);
       if (MO.isReg() && !HasVEX_4V &&
-          X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+          X86II::isX86_64ExtendedReg(MO.getReg()))
         VEX_R = 0x0;
     }
     break;
@@ -606,7 +615,7 @@ static unsigned DetermineREXPrefix(const MCInst &MI, uint64_t TSFlags,
     const MCOperand &MO = MI.getOperand(i);
     if (!MO.isReg()) continue;
     unsigned Reg = MO.getReg();
-    if (!X86InstrInfo::isX86_64NonExtLowByteReg(Reg)) continue;
+    if (!X86II::isX86_64NonExtLowByteReg(Reg)) continue;
     // FIXME: The caller of DetermineREXPrefix slaps this prefix onto anything
     // that returns non-zero.
     REX |= 0x40; // REX fixed encoding prefix
@@ -617,25 +626,25 @@ static unsigned DetermineREXPrefix(const MCInst &MI, uint64_t TSFlags,
   case X86II::MRMInitReg: assert(0 && "FIXME: Remove this!");
   case X86II::MRMSrcReg:
     if (MI.getOperand(0).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
       REX |= 1 << 2; // set REX.R
     i = isTwoAddr ? 2 : 1;
     for (; i != NumOps; ++i) {
       const MCOperand &MO = MI.getOperand(i);
-      if (MO.isReg() && X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+      if (MO.isReg() && X86II::isX86_64ExtendedReg(MO.getReg()))
         REX |= 1 << 0; // set REX.B
     }
     break;
   case X86II::MRMSrcMem: {
     if (MI.getOperand(0).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
       REX |= 1 << 2; // set REX.R
     unsigned Bit = 0;
     i = isTwoAddr ? 2 : 1;
     for (; i != NumOps; ++i) {
       const MCOperand &MO = MI.getOperand(i);
       if (MO.isReg()) {
-        if (X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+        if (X86II::isX86_64ExtendedReg(MO.getReg()))
           REX |= 1 << Bit; // set REX.B (Bit=0) and REX.X (Bit=1)
         Bit++;
       }
@@ -650,13 +659,13 @@ static unsigned DetermineREXPrefix(const MCInst &MI, uint64_t TSFlags,
     unsigned e = (isTwoAddr ? X86::AddrNumOperands+1 : X86::AddrNumOperands);
     i = isTwoAddr ? 1 : 0;
     if (NumOps > e && MI.getOperand(e).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(e).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(e).getReg()))
       REX |= 1 << 2; // set REX.R
     unsigned Bit = 0;
     for (; i != e; ++i) {
       const MCOperand &MO = MI.getOperand(i);
       if (MO.isReg()) {
-        if (X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+        if (X86II::isX86_64ExtendedReg(MO.getReg()))
           REX |= 1 << Bit; // REX.B (Bit=0) and REX.X (Bit=1)
         Bit++;
       }
@@ -665,12 +674,12 @@ static unsigned DetermineREXPrefix(const MCInst &MI, uint64_t TSFlags,
   }
   default:
     if (MI.getOperand(0).isReg() &&
-        X86InstrInfo::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
+        X86II::isX86_64ExtendedReg(MI.getOperand(0).getReg()))
       REX |= 1 << 0; // set REX.B
     i = isTwoAddr ? 2 : 1;
     for (unsigned e = NumOps; i != e; ++i) {
       const MCOperand &MO = MI.getOperand(i);
-      if (MO.isReg() && X86InstrInfo::isX86_64ExtendedReg(MO.getReg()))
+      if (MO.isReg() && X86II::isX86_64ExtendedReg(MO.getReg()))
         REX |= 1 << 2; // set REX.R
     }
     break;
@@ -1009,7 +1018,7 @@ EncodeInstruction(const MCInst &MI, raw_ostream &OS,
     if ((TSFlags >> X86II::VEXShift) & X86II::VEX_I8IMM) {
       const MCOperand &MO = MI.getOperand(CurOp++);
       bool IsExtReg =
-        X86InstrInfo::isX86_64ExtendedReg(MO.getReg());
+        X86II::isX86_64ExtendedReg(MO.getReg());
       unsigned RegNum = (IsExtReg ? (1 << 7) : 0);
       RegNum |= GetX86RegNum(MO) << 4;
       EmitImmediate(MCOperand::CreateImm(RegNum), 1, FK_Data_1, CurByte, OS,

lib/Target/X86/X86MachObjectWriter.cpp

@@ -8,7 +8,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "X86.h"
-#include "X86FixupKinds.h"
+#include "MCTargetDesc/X86FixupKinds.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/MC/MCAssembler.h"
 #include "llvm/MC/MCAsmLayout.h"