1 //===-- EDDisassembler.h - LLVM Enhanced Disassembler -----------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
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.
14 //===----------------------------------------------------------------------===//
16 #ifndef LLVM_EDDISASSEMBLER_H
17 #define LLVM_EDDISASSEMBLER_H
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"
41 class MCParsedAsmOperand;
44 class MCSubtargetInfo;
45 class MCTargetAsmLexer;
46 class MCTargetAsmParser;
47 template <typename T> class SmallVectorImpl;
56 typedef int (*EDByteReaderCallback)(uint8_t *byte, uint64_t address, void *arg);
58 /// EDDisassembler - Encapsulates a disassembler for a single architecture and
59 /// disassembly syntax. Also manages the static disassembler registry.
60 struct EDDisassembler {
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
74 /// CPUKey - Encapsulates the descriptor of an architecture/disassembly-syntax
77 /// The architecture type
80 /// The assembly syntax
81 AssemblySyntax Syntax;
83 /// operator== - Equality operator
84 bool operator==(const CPUKey &key) const {
85 return (Triple == key.Triple &&
86 Syntax == key.Syntax);
89 /// operator< - Less-than operator
90 bool operator<(const CPUKey &key) const {
91 return ((Triple < key.Triple) ||
92 ((Triple == key.Triple) && Syntax < (key.Syntax)));
96 typedef std::map<CPUKey, EDDisassembler*> DisassemblerMap_t;
98 /// A map from disassembler specifications to disassemblers. Populated
100 static DisassemblerMap_t sDisassemblers;
102 /// getDisassembler - Returns the specified disassemble, or NULL on failure
104 /// @arg arch - The desired architecture
105 /// @arg syntax - The desired disassembly syntax
106 static EDDisassembler *getDisassembler(llvm::Triple::ArchType arch,
107 AssemblySyntax syntax);
109 /// getDisassembler - Returns the disassembler for a given combination of
110 /// CPU type, CPU subtype, and assembly syntax, or NULL on failure
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,
116 AssemblySyntax syntax);
118 ////////////////////////
119 // Per-object members //
120 ////////////////////////
122 /// True only if the object has been successfully initialized
124 /// True if the disassembler can provide semantic information
127 /// The stream to write errors to
128 llvm::raw_ostream &ErrorStream;
130 /// The triple/syntax pair for the current architecture
132 /// The Triple fur the current architecture
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;
138 /// The subtarget information for the target architecture
139 llvm::OwningPtr<const llvm::MCSubtargetInfo> STI;
140 // The register information for the target architecture.
141 llvm::OwningPtr<const llvm::MCRegisterInfo> MRI;
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
146 llvm::OwningPtr<std::string> InstString;
147 /// The output stream for the disassembler; must be guarded with
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
158 const llvm::EDInstInfo *InstInfos;
159 /// The target-specific lexer for use in tokenizing strings, in
160 /// target-independent and target-specific portions
161 llvm::OwningPtr<llvm::AsmLexer> GenericAsmLexer;
162 llvm::OwningPtr<llvm::MCTargetAsmLexer> SpecificAsmLexer;
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;
168 typedef std::vector<std::string> regvec_t;
169 typedef std::map<std::string, unsigned> regrmap_t;
171 /// A vector of registers for quick mapping from LLVM register IDs to names
173 /// A map of registers for quick mapping from register names to LLVM IDs
176 /// A set of register IDs for aliases of the stack pointer for the current
178 std::set<unsigned> stackPointers;
179 /// A set of register IDs for aliases of the program counter for the current
181 std::set<unsigned> programCounters;
183 /// Constructor - initializes a disassembler with all the necessary objects,
184 /// which come pre-allocated from the registry accessor function
186 /// @arg key - the architecture and disassembly syntax for the
188 EDDisassembler(CPUKey& key);
190 /// valid - reports whether there was a failure in the constructor.
195 /// hasSemantics - reports whether the disassembler can provide operands and
197 bool hasSemantics() {
203 /// createInst - creates and returns an instruction given a callback and
204 /// memory address, or NULL on failure
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,
214 /// initMaps - initializes regVec and regRMap using the provided register
217 /// @arg registerInfo - the register information to use as a source
218 void initMaps(const llvm::MCRegisterInfo ®isterInfo);
219 /// nameWithRegisterID - Returns the name (owned by the EDDisassembler) of a
220 /// register for a given register ID, or NULL on failure
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
227 /// @arg name - The name of the register
228 unsigned registerIDWithName(const char *name) const;
230 /// registerIsStackPointer - reports whether a register ID is an alias for the
231 /// stack pointer register
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
238 /// @arg registerID - The LLVM register ID
239 bool registerIsProgramCounter(unsigned registerID);
241 /// printInst - prints an MCInst to a string, returning 0 on success, or -1
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,
250 /// parseInst - extracts operands and tokens from a string for use in
251 /// tokenizing the string. Returns 0 on success, or -1 otherwise.
253 /// @arg operands - A reference to a vector that will be filled in with the
255 /// @arg tokens - A reference to a vector that will be filled in with the
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);
262 /// llvmSyntaxVariant - returns the LLVM syntax variant for this disassembler
263 int llvmSyntaxVariant() const;
266 } // end namespace llvm