Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 1 | //===-- EDDisassembler.h - LLVM Enhanced Disassembler -----------*- C++ -*-===// |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 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 | |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 16 | #ifndef LLVM_EDDISASSEMBLER_H |
| 17 | #define LLVM_EDDISASSEMBLER_H |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 18 | |
Chris Lattner | a4f15d6 | 2010-07-20 18:59:58 +0000 | [diff] [blame] | 19 | #include "EDInfo.h" |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 20 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 21 | #include "llvm/ADT/OwningPtr.h" |
| 22 | #include "llvm/ADT/Triple.h" |
| 23 | #include "llvm/Support/raw_ostream.h" |
Michael J. Spencer | 1f6efa3 | 2010-11-29 18:16:10 +0000 | [diff] [blame] | 24 | #include "llvm/Support/Mutex.h" |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 25 | |
| 26 | #include <map> |
| 27 | #include <set> |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 28 | #include <string> |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 29 | #include <vector> |
| 30 | |
| 31 | namespace llvm { |
| 32 | class AsmLexer; |
Evan Cheng | 0e6a052 | 2011-07-18 20:57:22 +0000 | [diff] [blame] | 33 | class AsmParser; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 34 | class AsmToken; |
| 35 | class MCContext; |
| 36 | class MCAsmInfo; |
| 37 | class MCAsmLexer; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 38 | class MCDisassembler; |
| 39 | class MCInstPrinter; |
| 40 | class MCInst; |
| 41 | class MCParsedAsmOperand; |
Evan Cheng | 0e6a052 | 2011-07-18 20:57:22 +0000 | [diff] [blame] | 42 | class MCRegisterInfo; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 43 | class MCStreamer; |
Evan Cheng | ebdeeab | 2011-07-08 01:53:10 +0000 | [diff] [blame] | 44 | class MCSubtargetInfo; |
Evan Cheng | 94b9550 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 45 | class MCTargetAsmLexer; |
| 46 | class MCTargetAsmParser; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 47 | template <typename T> class SmallVectorImpl; |
| 48 | class SourceMgr; |
| 49 | class Target; |
Sean Callanan | 9899f70 | 2010-04-13 21:21:57 +0000 | [diff] [blame] | 50 | |
| 51 | struct EDInstInfo; |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 52 | struct EDInst; |
| 53 | struct EDOperand; |
| 54 | struct EDToken; |
| 55 | |
| 56 | typedef int (*EDByteReaderCallback)(uint8_t *byte, uint64_t address, void *arg); |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 57 | |
| 58 | /// EDDisassembler - Encapsulates a disassembler for a single architecture and |
| 59 | /// disassembly syntax. Also manages the static disassembler registry. |
| 60 | struct EDDisassembler { |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 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 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 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 |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 78 | std::string Triple; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 79 | |
| 80 | /// The assembly syntax |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 81 | AssemblySyntax Syntax; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 82 | |
| 83 | /// operator== - Equality operator |
| 84 | bool operator==(const CPUKey &key) const { |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 85 | return (Triple == key.Triple && |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 86 | Syntax == key.Syntax); |
| 87 | } |
| 88 | |
| 89 | /// operator< - Less-than operator |
| 90 | bool operator<(const CPUKey &key) const { |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 91 | return ((Triple < key.Triple) || |
| 92 | ((Triple == key.Triple) && Syntax < (key.Syntax))); |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 93 | } |
| 94 | }; |
| 95 | |
| 96 | typedef std::map<CPUKey, EDDisassembler*> DisassemblerMap_t; |
| 97 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 98 | /// A map from disassembler specifications to disassemblers. Populated |
| 99 | /// lazily. |
| 100 | static DisassemblerMap_t sDisassemblers; |
| 101 | |
| 102 | /// getDisassembler - Returns the specified disassemble, or NULL on failure |
| 103 | /// |
| 104 | /// @arg arch - The desired architecture |
| 105 | /// @arg syntax - The desired disassembly syntax |
| 106 | static EDDisassembler *getDisassembler(llvm::Triple::ArchType arch, |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 107 | AssemblySyntax syntax); |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 108 | |
| 109 | /// getDisassembler - Returns the disassembler for a given combination of |
| 110 | /// CPU type, CPU subtype, and assembly syntax, or NULL on failure |
| 111 | /// |
| 112 | /// @arg str - The string representation of the architecture triple, e.g., |
| 113 | /// "x86_64-apple-darwin" |
| 114 | /// @arg syntax - The disassembly syntax for the required disassembler |
| 115 | static EDDisassembler *getDisassembler(llvm::StringRef str, |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 116 | AssemblySyntax syntax); |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 117 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 118 | //////////////////////// |
| 119 | // Per-object members // |
| 120 | //////////////////////// |
| 121 | |
Sean Callanan | 8f993b8 | 2010-04-08 00:48:21 +0000 | [diff] [blame] | 122 | /// True only if the object has been successfully initialized |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 123 | bool Valid; |
Sean Callanan | 8f993b8 | 2010-04-08 00:48:21 +0000 | [diff] [blame] | 124 | /// True if the disassembler can provide semantic information |
| 125 | bool HasSemantics; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 126 | |
Sean Callanan | 8f993b8 | 2010-04-08 00:48:21 +0000 | [diff] [blame] | 127 | /// The stream to write errors to |
| 128 | llvm::raw_ostream &ErrorStream; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 129 | |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 130 | /// The triple/syntax pair for the current architecture |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 131 | CPUKey Key; |
Sean Callanan | 3f4f34c | 2012-02-04 07:45:35 +0000 | [diff] [blame] | 132 | /// The Triple fur the current architecture |
| 133 | Triple TgtTriple; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 134 | /// The LLVM target corresponding to the disassembler |
| 135 | const llvm::Target *Tgt; |
| 136 | /// The assembly information for the target architecture |
| 137 | llvm::OwningPtr<const llvm::MCAsmInfo> AsmInfo; |
James Molloy | b950585 | 2011-09-07 17:24:38 +0000 | [diff] [blame] | 138 | /// The subtarget information for the target architecture |
| 139 | llvm::OwningPtr<const llvm::MCSubtargetInfo> STI; |
Evan Cheng | 0e6a052 | 2011-07-18 20:57:22 +0000 | [diff] [blame] | 140 | // The register information for the target architecture. |
| 141 | llvm::OwningPtr<const llvm::MCRegisterInfo> MRI; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 142 | /// The disassembler for the target architecture |
| 143 | llvm::OwningPtr<const llvm::MCDisassembler> Disassembler; |
| 144 | /// The output string for the instruction printer; must be guarded with |
| 145 | /// PrinterMutex |
| 146 | llvm::OwningPtr<std::string> InstString; |
| 147 | /// The output stream for the disassembler; must be guarded with |
| 148 | /// PrinterMutex |
| 149 | llvm::OwningPtr<llvm::raw_string_ostream> InstStream; |
| 150 | /// The instruction printer for the target architecture; must be guarded with |
| 151 | /// PrinterMutex when printing |
| 152 | llvm::OwningPtr<llvm::MCInstPrinter> InstPrinter; |
| 153 | /// The mutex that guards the instruction printer's printing functions, which |
| 154 | /// use a shared stream |
| 155 | llvm::sys::Mutex PrinterMutex; |
| 156 | /// The array of instruction information provided by the TableGen backend for |
| 157 | /// the target architecture |
Sean Callanan | 9899f70 | 2010-04-13 21:21:57 +0000 | [diff] [blame] | 158 | const llvm::EDInstInfo *InstInfos; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 159 | /// The target-specific lexer for use in tokenizing strings, in |
| 160 | /// target-independent and target-specific portions |
| 161 | llvm::OwningPtr<llvm::AsmLexer> GenericAsmLexer; |
Evan Cheng | 94b9550 | 2011-07-26 00:24:13 +0000 | [diff] [blame] | 162 | llvm::OwningPtr<llvm::MCTargetAsmLexer> SpecificAsmLexer; |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 163 | /// The guard for the above |
| 164 | llvm::sys::Mutex ParserMutex; |
| 165 | /// The LLVM number used for the target disassembly syntax variant |
| 166 | int LLVMSyntaxVariant; |
| 167 | |
| 168 | typedef std::vector<std::string> regvec_t; |
| 169 | typedef std::map<std::string, unsigned> regrmap_t; |
| 170 | |
| 171 | /// A vector of registers for quick mapping from LLVM register IDs to names |
| 172 | regvec_t RegVec; |
| 173 | /// A map of registers for quick mapping from register names to LLVM IDs |
| 174 | regrmap_t RegRMap; |
| 175 | |
| 176 | /// A set of register IDs for aliases of the stack pointer for the current |
| 177 | /// architecture |
| 178 | std::set<unsigned> stackPointers; |
| 179 | /// A set of register IDs for aliases of the program counter for the current |
| 180 | /// architecture |
| 181 | std::set<unsigned> programCounters; |
| 182 | |
| 183 | /// Constructor - initializes a disassembler with all the necessary objects, |
| 184 | /// which come pre-allocated from the registry accessor function |
| 185 | /// |
| 186 | /// @arg key - the architecture and disassembly syntax for the |
| 187 | /// disassembler |
| 188 | EDDisassembler(CPUKey& key); |
| 189 | |
| 190 | /// valid - reports whether there was a failure in the constructor. |
| 191 | bool valid() { |
| 192 | return Valid; |
| 193 | } |
| 194 | |
Sean Callanan | 8f993b8 | 2010-04-08 00:48:21 +0000 | [diff] [blame] | 195 | /// hasSemantics - reports whether the disassembler can provide operands and |
| 196 | /// tokens. |
| 197 | bool hasSemantics() { |
| 198 | return HasSemantics; |
| 199 | } |
| 200 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 201 | ~EDDisassembler(); |
| 202 | |
| 203 | /// createInst - creates and returns an instruction given a callback and |
| 204 | /// memory address, or NULL on failure |
| 205 | /// |
| 206 | /// @arg byteReader - A callback function that provides machine code bytes |
| 207 | /// @arg address - The address of the first byte of the instruction, |
| 208 | /// suitable for passing to byteReader |
| 209 | /// @arg arg - An opaque argument for byteReader |
| 210 | EDInst *createInst(EDByteReaderCallback byteReader, |
| 211 | uint64_t address, |
| 212 | void *arg); |
| 213 | |
| 214 | /// initMaps - initializes regVec and regRMap using the provided register |
| 215 | /// info |
| 216 | /// |
| 217 | /// @arg registerInfo - the register information to use as a source |
Evan Cheng | 1b0fc9b | 2011-07-25 20:53:02 +0000 | [diff] [blame] | 218 | void initMaps(const llvm::MCRegisterInfo ®isterInfo); |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 219 | /// nameWithRegisterID - Returns the name (owned by the EDDisassembler) of a |
| 220 | /// register for a given register ID, or NULL on failure |
| 221 | /// |
| 222 | /// @arg registerID - the ID of the register to be queried |
| 223 | const char *nameWithRegisterID(unsigned registerID) const; |
| 224 | /// registerIDWithName - Returns the ID of a register for a given register |
| 225 | /// name, or (unsigned)-1 on failure |
| 226 | /// |
| 227 | /// @arg name - The name of the register |
| 228 | unsigned registerIDWithName(const char *name) const; |
| 229 | |
| 230 | /// registerIsStackPointer - reports whether a register ID is an alias for the |
| 231 | /// stack pointer register |
| 232 | /// |
| 233 | /// @arg registerID - The LLVM register ID |
| 234 | bool registerIsStackPointer(unsigned registerID); |
| 235 | /// registerIsStackPointer - reports whether a register ID is an alias for the |
| 236 | /// stack pointer register |
| 237 | /// |
| 238 | /// @arg registerID - The LLVM register ID |
| 239 | bool registerIsProgramCounter(unsigned registerID); |
| 240 | |
| 241 | /// printInst - prints an MCInst to a string, returning 0 on success, or -1 |
| 242 | /// otherwise |
| 243 | /// |
| 244 | /// @arg str - A reference to a string which is filled in with the string |
| 245 | /// representation of the instruction |
| 246 | /// @arg inst - A reference to the MCInst to be printed |
| 247 | int printInst(std::string& str, |
| 248 | llvm::MCInst& inst); |
| 249 | |
| 250 | /// parseInst - extracts operands and tokens from a string for use in |
| 251 | /// tokenizing the string. Returns 0 on success, or -1 otherwise. |
| 252 | /// |
| 253 | /// @arg operands - A reference to a vector that will be filled in with the |
| 254 | /// parsed operands |
| 255 | /// @arg tokens - A reference to a vector that will be filled in with the |
| 256 | /// tokens |
| 257 | /// @arg str - The string representation of the instruction |
| 258 | int parseInst(llvm::SmallVectorImpl<llvm::MCParsedAsmOperand*> &operands, |
| 259 | llvm::SmallVectorImpl<llvm::AsmToken> &tokens, |
| 260 | const std::string &str); |
| 261 | |
| 262 | /// llvmSyntaxVariant - returns the LLVM syntax variant for this disassembler |
| 263 | int llvmSyntaxVariant() const; |
| 264 | }; |
| 265 | |
Chris Lattner | 847da55 | 2010-07-20 18:25:19 +0000 | [diff] [blame] | 266 | } // end namespace llvm |
| 267 | |
Sean Callanan | ee5dfd4 | 2010-02-01 08:49:35 +0000 | [diff] [blame] | 268 | #endif |