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