lib/MC/MCDisassembler/EDDisassembler.h

   1 //===-- EDDisassembler.h - LLVM Enhanced Disassembler -----------*- C++ -*-===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This file defines the interface for the Enhanced Disassembly library's
  11 // disassembler class.  The disassembler is responsible for vending individual
  12 // instructions according to a given architecture and disassembly syntax.
  13 //
  14 //===----------------------------------------------------------------------===//
  15
  16 #ifndef LLVM_EDDISASSEMBLER_H
  17 #define LLVM_EDDISASSEMBLER_H
  18
  19 #include "EDInfo.h"
  20
  21 #include "llvm/ADT/OwningPtr.h"
  22 #include "llvm/ADT/Triple.h"
  23 #include "llvm/Support/raw_ostream.h"
  24 #include "llvm/Support/Mutex.h"
  25
  26 #include <map>
  27 #include <set>
  28 #include <vector>
  29
  30 namespace llvm {
  31 class AsmLexer;
  32 class AsmToken;
  33 class MCContext;
  34 class MCAsmInfo;
  35 class MCAsmLexer;
  36 class AsmParser;
  37 class TargetAsmLexer;
  38 class TargetAsmParser;
  39 class MCDisassembler;
  40 class MCInstPrinter;
  41 class MCInst;
  42 class MCParsedAsmOperand;
  43 class MCStreamer;
  44 class MCSubtargetInfo;
  45 template <typename T> class SmallVectorImpl;
  46 class SourceMgr;
  47 class Target;
  48 class TargetMachine;
  49 class TargetRegisterInfo;
  50
  51 struct EDInstInfo;
  52 struct EDInst;
  53 struct EDOperand;
  54 struct EDToken;
  55
  56 typedef int (*EDByteReaderCallback)(uint8_t *byte, uint64_t address, void *arg);
  57
  58 /// EDDisassembler - Encapsulates a disassembler for a single architecture and
  59 ///   disassembly syntax.  Also manages the static disassembler registry.
  60 struct EDDisassembler {
  61   typedef enum {
  62     /*! @constant kEDAssemblySyntaxX86Intel Intel syntax for i386 and x86_64. */
  63     kEDAssemblySyntaxX86Intel  = 0,
  64     /*! @constant kEDAssemblySyntaxX86ATT AT&T syntax for i386 and x86_64. */
  65     kEDAssemblySyntaxX86ATT    = 1,
  66     kEDAssemblySyntaxARMUAL    = 2
  67   } AssemblySyntax;
  68
  69
  70   ////////////////////
  71   // Static members //
  72   ////////////////////
  73
  74   /// CPUKey - Encapsulates the descriptor of an architecture/disassembly-syntax
  75   ///   pair
  76   struct CPUKey {
  77     /// The architecture type
  78     llvm::Triple::ArchType Arch;
  79
  80     /// The assembly syntax
  81     AssemblySyntax Syntax;
  82
  83     /// operator== - Equality operator
  84     bool operator==(const CPUKey &key) const {
  85       return (Arch == key.Arch &&
  86               Syntax == key.Syntax);
  87     }
  88
  89     /// operator< - Less-than operator
  90     bool operator<(const CPUKey &key) const {
  91       return ((Arch < key.Arch) ||
  92               ((Arch == key.Arch) && Syntax < (key.Syntax)));
  93     }
  94   };
  95
  96   typedef std::map<CPUKey, EDDisassembler*> DisassemblerMap_t;
  97
  98   /// True if the disassembler registry has been initialized; false if not
  99   static bool sInitialized;
 100   /// A map from disassembler specifications to disassemblers.  Populated
 101   ///   lazily.
 102   static DisassemblerMap_t sDisassemblers;
 103
 104   /// getDisassembler - Returns the specified disassemble, or NULL on failure
 105   ///
 106   /// @arg arch   - The desired architecture
 107   /// @arg syntax - The desired disassembly syntax
 108   static EDDisassembler *getDisassembler(llvm::Triple::ArchType arch,
 109                                          AssemblySyntax syntax);
 110
 111   /// getDisassembler - Returns the disassembler for a given combination of
 112   ///   CPU type, CPU subtype, and assembly syntax, or NULL on failure
 113   ///
 114   /// @arg str    - The string representation of the architecture triple, e.g.,
 115   ///               "x86_64-apple-darwin"
 116   /// @arg syntax - The disassembly syntax for the required disassembler
 117   static EDDisassembler *getDisassembler(llvm::StringRef str,
 118                                          AssemblySyntax syntax);
 119
 120   /// initialize - Initializes the disassembler registry and the LLVM backend
 121   static void initialize();
 122
 123   ////////////////////////
 124   // Per-object members //
 125   ////////////////////////
 126
 127   /// True only if the object has been successfully initialized
 128   bool Valid;
 129   /// True if the disassembler can provide semantic information
 130   bool HasSemantics;
 131
 132   /// The stream to write errors to
 133   llvm::raw_ostream &ErrorStream;
 134
 135   /// The architecture/syntax pair for the current architecture
 136   CPUKey Key;
 137   /// The LLVM target corresponding to the disassembler
 138   const llvm::Target *Tgt;
 139   /// The target machine instance.
 140   llvm::OwningPtr<llvm::TargetMachine> TargetMachine;
 141   /// The assembly information for the target architecture
 142   llvm::OwningPtr<const llvm::MCAsmInfo> AsmInfo;
 143   /// The disassembler for the target architecture
 144   llvm::OwningPtr<const llvm::MCDisassembler> Disassembler;
 145   /// The output string for the instruction printer; must be guarded with
 146   ///   PrinterMutex
 147   llvm::OwningPtr<std::string> InstString;
 148   /// The output stream for the disassembler; must be guarded with
 149   ///   PrinterMutex
 150   llvm::OwningPtr<llvm::raw_string_ostream> InstStream;
 151   /// The instruction printer for the target architecture; must be guarded with
 152   ///   PrinterMutex when printing
 153   llvm::OwningPtr<llvm::MCInstPrinter> InstPrinter;
 154   /// The mutex that guards the instruction printer's printing functions, which
 155   ///   use a shared stream
 156   llvm::sys::Mutex PrinterMutex;
 157   /// The array of instruction information provided by the TableGen backend for
 158   ///   the target architecture
 159   const llvm::EDInstInfo *InstInfos;
 160   /// The target-specific lexer for use in tokenizing strings, in
 161   ///   target-independent and target-specific portions
 162   llvm::OwningPtr<llvm::AsmLexer> GenericAsmLexer;
 163   llvm::OwningPtr<llvm::TargetAsmLexer> SpecificAsmLexer;
 164   /// The guard for the above
 165   llvm::sys::Mutex ParserMutex;
 166   /// The LLVM number used for the target disassembly syntax variant
 167   int LLVMSyntaxVariant;
 168
 169   typedef std::vector<std::string> regvec_t;
 170   typedef std::map<std::string, unsigned> regrmap_t;
 171
 172   /// A vector of registers for quick mapping from LLVM register IDs to names
 173   regvec_t RegVec;
 174   /// A map of registers for quick mapping from register names to LLVM IDs
 175   regrmap_t RegRMap;
 176
 177   /// A set of register IDs for aliases of the stack pointer for the current
 178   ///   architecture
 179   std::set<unsigned> stackPointers;
 180   /// A set of register IDs for aliases of the program counter for the current
 181   ///   architecture
 182   std::set<unsigned> programCounters;
 183
 184   /// Constructor - initializes a disassembler with all the necessary objects,
 185   ///   which come pre-allocated from the registry accessor function
 186   ///
 187   /// @arg key                - the architecture and disassembly syntax for the
 188   ///                           disassembler
 189   EDDisassembler(CPUKey& key);
 190
 191   /// valid - reports whether there was a failure in the constructor.
 192   bool valid() {
 193     return Valid;
 194   }
 195
 196   /// hasSemantics - reports whether the disassembler can provide operands and
 197   ///   tokens.
 198   bool hasSemantics() {
 199     return HasSemantics;
 200   }
 201
 202   ~EDDisassembler();
 203
 204   /// createInst - creates and returns an instruction given a callback and
 205   ///   memory address, or NULL on failure
 206   ///
 207   /// @arg byteReader - A callback function that provides machine code bytes
 208   /// @arg address    - The address of the first byte of the instruction,
 209   ///                   suitable for passing to byteReader
 210   /// @arg arg        - An opaque argument for byteReader
 211   EDInst *createInst(EDByteReaderCallback byteReader,
 212                      uint64_t address,
 213                      void *arg);
 214
 215   /// initMaps - initializes regVec and regRMap using the provided register
 216   ///   info
 217   ///
 218   /// @arg registerInfo - the register information to use as a source
 219   void initMaps(const llvm::TargetRegisterInfo &registerInfo);
 220   /// nameWithRegisterID - Returns the name (owned by the EDDisassembler) of a
 221   ///   register for a given register ID, or NULL on failure
 222   ///
 223   /// @arg registerID - the ID of the register to be queried
 224   const char *nameWithRegisterID(unsigned registerID) const;
 225   /// registerIDWithName - Returns the ID of a register for a given register
 226   ///   name, or (unsigned)-1 on failure
 227   ///
 228   /// @arg name - The name of the register
 229   unsigned registerIDWithName(const char *name) const;
 230
 231   /// registerIsStackPointer - reports whether a register ID is an alias for the
 232   ///   stack pointer register
 233   ///
 234   /// @arg registerID - The LLVM register ID
 235   bool registerIsStackPointer(unsigned registerID);
 236   /// registerIsStackPointer - reports whether a register ID is an alias for the
 237   ///   stack pointer register
 238   ///
 239   /// @arg registerID - The LLVM register ID
 240   bool registerIsProgramCounter(unsigned registerID);
 241
 242   /// printInst - prints an MCInst to a string, returning 0 on success, or -1
 243   ///   otherwise
 244   ///
 245   /// @arg str  - A reference to a string which is filled in with the string
 246   ///             representation of the instruction
 247   /// @arg inst - A reference to the MCInst to be printed
 248   int printInst(std::string& str,
 249                 llvm::MCInst& inst);
 250
 251   /// parseInst - extracts operands and tokens from a string for use in
 252   ///   tokenizing the string.  Returns 0 on success, or -1 otherwise.
 253   ///
 254   /// @arg operands - A reference to a vector that will be filled in with the
 255   ///                 parsed operands
 256   /// @arg tokens   - A reference to a vector that will be filled in with the
 257   ///                 tokens
 258   /// @arg str      - The string representation of the instruction
 259   int parseInst(llvm::SmallVectorImpl<llvm::MCParsedAsmOperand*> &operands,
 260                 llvm::SmallVectorImpl<llvm::AsmToken> &tokens,
 261                 const std::string &str);
 262
 263   /// llvmSyntaxVariant - returns the LLVM syntax variant for this disassembler
 264   int llvmSyntaxVariant() const;
 265 };
 266
 267 } // end namespace llvm
 268
 269 #endif