1 //===- Disassembler.cpp - Disassembler for hex strings --------------------===// 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 class implements the disassembler of strings of bytes written in 11 // hexadecimal, from standard input or from a file. 12 // 13 //===----------------------------------------------------------------------===// 14 15 #include "Disassembler.h" 16 #include "../../lib/MC/MCDisassembler/EDDisassembler.h" 17 #include "../../lib/MC/MCDisassembler/EDInst.h" 18 #include "../../lib/MC/MCDisassembler/EDOperand.h" 19 #include "../../lib/MC/MCDisassembler/EDToken.h" 20 #include "llvm/MC/MCAsmInfo.h" 21 #include "llvm/MC/MCDisassembler.h" 22 #include "llvm/MC/MCInst.h" 23 #include "llvm/MC/MCInstPrinter.h" 24 #include "llvm/Target/TargetRegistry.h" 25 #include "llvm/ADT/OwningPtr.h" 26 #include "llvm/ADT/Triple.h" 27 #include "llvm/ADT/Twine.h" 28 #include "llvm/Support/MemoryBuffer.h" 29 #include "llvm/Support/MemoryObject.h" 30 #include "llvm/Support/raw_ostream.h" 31 #include "llvm/Support/SourceMgr.h" 32 using namespace llvm; 33 34 typedef std::vector<std::pair<unsigned char, const char*> > ByteArrayTy; 35 36 namespace { 37 class VectorMemoryObject : public MemoryObject { 38 private: 39 const ByteArrayTy &Bytes; 40 public: 41 VectorMemoryObject(const ByteArrayTy &bytes) : Bytes(bytes) {} 42 43 uint64_t getBase() const { return 0; } 44 uint64_t getExtent() const { return Bytes.size(); } 45 46 int readByte(uint64_t Addr, uint8_t *Byte) const { 47 if (Addr > getExtent()) 48 return -1; 49 *Byte = Bytes[Addr].first; 50 return 0; 51 } 52 }; 53 } 54 55 static bool PrintInsts(const MCDisassembler &DisAsm, 56 MCInstPrinter &Printer, const ByteArrayTy &Bytes, 57 SourceMgr &SM, raw_ostream &Out) { 58 // Wrap the vector in a MemoryObject. 59 VectorMemoryObject memoryObject(Bytes); 60 61 // Disassemble it to strings. 62 uint64_t Size; 63 uint64_t Index; 64 65 for (Index = 0; Index < Bytes.size(); Index += Size) { 66 MCInst Inst; 67 68 if (DisAsm.getInstruction(Inst, Size, memoryObject, Index, 69 /*REMOVE*/ nulls())) { 70 Printer.printInst(&Inst, Out); 71 Out << "\n"; 72 } else { 73 SM.PrintMessage(SMLoc::getFromPointer(Bytes[Index].second), 74 "invalid instruction encoding", "warning"); 75 if (Size == 0) 76 Size = 1; // skip illegible bytes 77 } 78 } 79 80 return false; 81 } 82 83 static bool ByteArrayFromString(ByteArrayTy &ByteArray, 84 StringRef &Str, 85 SourceMgr &SM) { 86 while (!Str.empty()) { 87 // Strip horizontal whitespace. 88 if (size_t Pos = Str.find_first_not_of(" \t\r")) { 89 Str = Str.substr(Pos); 90 continue; 91 } 92 93 // If this is the end of a line or start of a comment, remove the rest of 94 // the line. 95 if (Str[0] == '\n' || Str[0] == '#') { 96 // Strip to the end of line if we already processed any bytes on this 97 // line. This strips the comment and/or the \n. 98 if (Str[0] == '\n') { 99 Str = Str.substr(1); 100 } else { 101 Str = Str.substr(Str.find_first_of('\n')); 102 if (!Str.empty()) 103 Str = Str.substr(1); 104 } 105 continue; 106 } 107 108 // Get the current token. 109 size_t Next = Str.find_first_of(" \t\n\r#"); 110 StringRef Value = Str.substr(0, Next); 111 112 // Convert to a byte and add to the byte vector. 113 unsigned ByteVal; 114 if (Value.getAsInteger(0, ByteVal) || ByteVal > 255) { 115 // If we have an error, print it and skip to the end of line. 116 SM.PrintMessage(SMLoc::getFromPointer(Value.data()), 117 "invalid input token", "error"); 118 Str = Str.substr(Str.find('\n')); 119 ByteArray.clear(); 120 continue; 121 } 122 123 ByteArray.push_back(std::make_pair((unsigned char)ByteVal, Value.data())); 124 Str = Str.substr(Next); 125 } 126 127 return false; 128 } 129 130 int Disassembler::disassemble(const Target &T, const std::string &Triple, 131 MemoryBuffer &Buffer, 132 raw_ostream &Out) { 133 // Set up disassembler. 134 OwningPtr<const MCAsmInfo> AsmInfo(T.createAsmInfo(Triple)); 135 136 if (!AsmInfo) { 137 errs() << "error: no assembly info for target " << Triple << "\n"; 138 return -1; 139 } 140 141 OwningPtr<const MCDisassembler> DisAsm(T.createMCDisassembler()); 142 if (!DisAsm) { 143 errs() << "error: no disassembler for target " << Triple << "\n"; 144 return -1; 145 } 146 147 int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); 148 OwningPtr<MCInstPrinter> IP(T.createMCInstPrinter(AsmPrinterVariant, 149 *AsmInfo)); 150 if (!IP) { 151 errs() << "error: no instruction printer for target " << Triple << '\n'; 152 return -1; 153 } 154 155 bool ErrorOccurred = false; 156 157 SourceMgr SM; 158 SM.AddNewSourceBuffer(&Buffer, SMLoc()); 159 160 // Convert the input to a vector for disassembly. 161 ByteArrayTy ByteArray; 162 StringRef Str = Buffer.getBuffer(); 163 164 ErrorOccurred |= ByteArrayFromString(ByteArray, Str, SM); 165 166 if (!ByteArray.empty()) 167 ErrorOccurred |= PrintInsts(*DisAsm, *IP, ByteArray, SM, Out); 168 169 return ErrorOccurred; 170 } 171 172 static int byteArrayReader(uint8_t *B, uint64_t A, void *Arg) { 173 ByteArrayTy &ByteArray = *((ByteArrayTy*)Arg); 174 175 if (A >= ByteArray.size()) 176 return -1; 177 178 *B = ByteArray[A].first; 179 180 return 0; 181 } 182 183 static int verboseEvaluator(uint64_t *V, unsigned R, void *Arg) { 184 EDDisassembler &disassembler = *(EDDisassembler *)((void **)Arg)[0]; 185 raw_ostream &Out = *(raw_ostream *)((void **)Arg)[1]; 186 187 if (const char *regName = disassembler.nameWithRegisterID(R)) 188 Out << "[" << regName << "/" << R << "]"; 189 190 if (disassembler.registerIsStackPointer(R)) 191 Out << "(sp)"; 192 if (disassembler.registerIsProgramCounter(R)) 193 Out << "(pc)"; 194 195 *V = 0; 196 return 0; 197 } 198 199 int Disassembler::disassembleEnhanced(const std::string &TS, 200 MemoryBuffer &Buffer, 201 raw_ostream &Out) { 202 ByteArrayTy ByteArray; 203 StringRef Str = Buffer.getBuffer(); 204 SourceMgr SM; 205 206 SM.AddNewSourceBuffer(&Buffer, SMLoc()); 207 208 if (ByteArrayFromString(ByteArray, Str, SM)) { 209 return -1; 210 } 211 212 Triple T(TS); 213 EDDisassembler::AssemblySyntax AS; 214 215 switch (T.getArch()) { 216 default: 217 errs() << "error: no default assembly syntax for " << TS.c_str() << "\n"; 218 return -1; 219 case Triple::arm: 220 case Triple::thumb: 221 AS = EDDisassembler::kEDAssemblySyntaxARMUAL; 222 break; 223 case Triple::x86: 224 case Triple::x86_64: 225 AS = EDDisassembler::kEDAssemblySyntaxX86ATT; 226 break; 227 } 228 229 EDDisassembler::initialize(); 230 EDDisassembler *disassembler = 231 EDDisassembler::getDisassembler(TS.c_str(), AS); 232 233 if (disassembler == 0) { 234 errs() << "error: couldn't get disassembler for " << TS << '\n'; 235 return -1; 236 } 237 238 EDInst *inst = 239 disassembler->createInst(byteArrayReader, 0, &ByteArray); 240 241 if (inst == 0) { 242 errs() << "error: Didn't get an instruction\n"; 243 return -1; 244 } 245 246 unsigned numTokens = inst->numTokens(); 247 if ((int)numTokens < 0) { 248 errs() << "error: couldn't count the instruction's tokens\n"; 249 return -1; 250 } 251 252 for (unsigned tokenIndex = 0; tokenIndex != numTokens; ++tokenIndex) { 253 EDToken *token; 254 255 if (inst->getToken(token, tokenIndex)) { 256 errs() << "error: Couldn't get token\n"; 257 return -1; 258 } 259 260 const char *buf; 261 if (token->getString(buf)) { 262 errs() << "error: Couldn't get string for token\n"; 263 return -1; 264 } 265 266 Out << '['; 267 int operandIndex = token->operandID(); 268 269 if (operandIndex >= 0) 270 Out << operandIndex << "-"; 271 272 switch (token->type()) { 273 default: Out << "?"; break; 274 case EDToken::kTokenWhitespace: Out << "w"; break; 275 case EDToken::kTokenPunctuation: Out << "p"; break; 276 case EDToken::kTokenOpcode: Out << "o"; break; 277 case EDToken::kTokenLiteral: Out << "l"; break; 278 case EDToken::kTokenRegister: Out << "r"; break; 279 } 280 281 Out << ":" << buf; 282 283 if (token->type() == EDToken::kTokenLiteral) { 284 Out << "="; 285 if (token->literalSign()) 286 Out << "-"; 287 uint64_t absoluteValue; 288 if (token->literalAbsoluteValue(absoluteValue)) { 289 errs() << "error: Couldn't get the value of a literal token\n"; 290 return -1; 291 } 292 Out << absoluteValue; 293 } else if (token->type() == EDToken::kTokenRegister) { 294 Out << "="; 295 unsigned regID; 296 if (token->registerID(regID)) { 297 errs() << "error: Couldn't get the ID of a register token\n"; 298 return -1; 299 } 300 Out << "r" << regID; 301 } 302 303 Out << "]"; 304 } 305 306 Out << " "; 307 308 if (inst->isBranch()) 309 Out << "<br> "; 310 if (inst->isMove()) 311 Out << "<mov> "; 312 313 unsigned numOperands = inst->numOperands(); 314 315 if ((int)numOperands < 0) { 316 errs() << "error: Couldn't count operands\n"; 317 return -1; 318 } 319 320 for (unsigned operandIndex = 0; operandIndex != numOperands; ++operandIndex) { 321 Out << operandIndex << ":"; 322 323 EDOperand *operand; 324 if (inst->getOperand(operand, operandIndex)) { 325 errs() << "error: couldn't get operand\n"; 326 return -1; 327 } 328 329 uint64_t evaluatedResult; 330 void *Arg[] = { disassembler, &Out }; 331 evaluatedResult = operand->evaluate(evaluatedResult, verboseEvaluator, Arg); 332 Out << "=" << evaluatedResult << " "; 333 } 334 335 Out << '\n'; 336 337 return 0; 338 } 339 340