1 //===-- llvm-mc.cpp - Machine Code Hacking Driver -------------------------===// 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 utility is a simple driver that allows command line hacking on machine 11 // code. 12 // 13 //===----------------------------------------------------------------------===// 14 15 #include "llvm/MC/MCParser/AsmLexer.h" 16 #include "llvm/MC/MCParser/MCAsmLexer.h" 17 #include "llvm/MC/MCContext.h" 18 #include "llvm/MC/MCCodeEmitter.h" 19 #include "llvm/MC/MCInstPrinter.h" 20 #include "llvm/MC/MCSectionMachO.h" 21 #include "llvm/MC/MCStreamer.h" 22 #include "llvm/Target/TargetAsmBackend.h" 23 #include "llvm/Target/TargetAsmParser.h" 24 #include "llvm/Target/TargetData.h" 25 #include "llvm/Target/TargetRegistry.h" 26 #include "llvm/Target/SubtargetFeature.h" // FIXME. 27 #include "llvm/Target/TargetMachine.h" // FIXME. 28 #include "llvm/Target/TargetSelect.h" 29 #include "llvm/ADT/OwningPtr.h" 30 #include "llvm/Support/CommandLine.h" 31 #include "llvm/Support/FileUtilities.h" 32 #include "llvm/Support/FormattedStream.h" 33 #include "llvm/Support/ManagedStatic.h" 34 #include "llvm/Support/MemoryBuffer.h" 35 #include "llvm/Support/PrettyStackTrace.h" 36 #include "llvm/Support/SourceMgr.h" 37 #include "llvm/Support/ToolOutputFile.h" 38 #include "llvm/System/Host.h" 39 #include "llvm/System/Signals.h" 40 #include "Disassembler.h" 41 using namespace llvm; 42 43 static cl::opt<std::string> 44 InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-")); 45 46 static cl::opt<std::string> 47 OutputFilename("o", cl::desc("Output filename"), 48 cl::value_desc("filename")); 49 50 static cl::opt<bool> 51 ShowEncoding("show-encoding", cl::desc("Show instruction encodings")); 52 53 static cl::opt<bool> 54 ShowInst("show-inst", cl::desc("Show internal instruction representation")); 55 56 static cl::opt<bool> 57 ShowInstOperands("show-inst-operands", 58 cl::desc("Show instructions operands as parsed")); 59 60 static cl::opt<unsigned> 61 OutputAsmVariant("output-asm-variant", 62 cl::desc("Syntax variant to use for output printing")); 63 64 static cl::opt<bool> 65 RelaxAll("mc-relax-all", cl::desc("Relax all fixups")); 66 67 static cl::opt<bool> 68 EnableLogging("enable-api-logging", cl::desc("Enable MC API logging")); 69 70 enum OutputFileType { 71 OFT_Null, 72 OFT_AssemblyFile, 73 OFT_ObjectFile 74 }; 75 static cl::opt<OutputFileType> 76 FileType("filetype", cl::init(OFT_AssemblyFile), 77 cl::desc("Choose an output file type:"), 78 cl::values( 79 clEnumValN(OFT_AssemblyFile, "asm", 80 "Emit an assembly ('.s') file"), 81 clEnumValN(OFT_Null, "null", 82 "Don't emit anything (for timing purposes)"), 83 clEnumValN(OFT_ObjectFile, "obj", 84 "Emit a native object ('.o') file"), 85 clEnumValEnd)); 86 87 static cl::list<std::string> 88 IncludeDirs("I", cl::desc("Directory of include files"), 89 cl::value_desc("directory"), cl::Prefix); 90 91 static cl::opt<std::string> 92 ArchName("arch", cl::desc("Target arch to assemble for, " 93 "see -version for available targets")); 94 95 static cl::opt<std::string> 96 TripleName("triple", cl::desc("Target triple to assemble for, " 97 "see -version for available targets")); 98 99 static cl::opt<std::string> 100 MCPU("mcpu", 101 cl::desc("Target a specific cpu type (-mcpu=help for details)"), 102 cl::value_desc("cpu-name"), 103 cl::init("")); 104 105 static cl::opt<bool> 106 NoInitialTextSection("n", cl::desc( 107 "Don't assume assembly file starts in the text section")); 108 109 enum ActionType { 110 AC_AsLex, 111 AC_Assemble, 112 AC_Disassemble, 113 AC_EDisassemble 114 }; 115 116 static cl::opt<ActionType> 117 Action(cl::desc("Action to perform:"), 118 cl::init(AC_Assemble), 119 cl::values(clEnumValN(AC_AsLex, "as-lex", 120 "Lex tokens from a .s file"), 121 clEnumValN(AC_Assemble, "assemble", 122 "Assemble a .s file (default)"), 123 clEnumValN(AC_Disassemble, "disassemble", 124 "Disassemble strings of hex bytes"), 125 clEnumValN(AC_EDisassemble, "edis", 126 "Enhanced disassembly of strings of hex bytes"), 127 clEnumValEnd)); 128 129 static const Target *GetTarget(const char *ProgName) { 130 // Figure out the target triple. 131 if (TripleName.empty()) 132 TripleName = sys::getHostTriple(); 133 if (!ArchName.empty()) { 134 llvm::Triple TT(TripleName); 135 TT.setArchName(ArchName); 136 TripleName = TT.str(); 137 } 138 139 // Get the target specific parser. 140 std::string Error; 141 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error); 142 if (TheTarget) 143 return TheTarget; 144 145 errs() << ProgName << ": error: unable to get target for '" << TripleName 146 << "', see --version and --triple.\n"; 147 return 0; 148 } 149 150 static tool_output_file *GetOutputStream() { 151 if (OutputFilename == "") 152 OutputFilename = "-"; 153 154 std::string Err; 155 tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err, 156 raw_fd_ostream::F_Binary); 157 if (!Err.empty()) { 158 errs() << Err << '\n'; 159 delete Out; 160 return 0; 161 } 162 163 return Out; 164 } 165 166 static int AsLexInput(const char *ProgName) { 167 std::string ErrorMessage; 168 MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, 169 &ErrorMessage); 170 if (Buffer == 0) { 171 errs() << ProgName << ": "; 172 if (ErrorMessage.size()) 173 errs() << ErrorMessage << "\n"; 174 else 175 errs() << "input file didn't read correctly.\n"; 176 return 1; 177 } 178 179 SourceMgr SrcMgr; 180 181 // Tell SrcMgr about this buffer, which is what TGParser will pick up. 182 SrcMgr.AddNewSourceBuffer(Buffer, SMLoc()); 183 184 // Record the location of the include directories so that the lexer can find 185 // it later. 186 SrcMgr.setIncludeDirs(IncludeDirs); 187 188 const Target *TheTarget = GetTarget(ProgName); 189 if (!TheTarget) 190 return 1; 191 192 llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName)); 193 assert(MAI && "Unable to create target asm info!"); 194 195 AsmLexer Lexer(*MAI); 196 Lexer.setBuffer(SrcMgr.getMemoryBuffer(0)); 197 198 OwningPtr<tool_output_file> Out(GetOutputStream()); 199 if (!Out) 200 return 1; 201 202 bool Error = false; 203 while (Lexer.Lex().isNot(AsmToken::Eof)) { 204 AsmToken Tok = Lexer.getTok(); 205 206 switch (Tok.getKind()) { 207 default: 208 SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning"); 209 Error = true; 210 break; 211 case AsmToken::Error: 212 Error = true; // error already printed. 213 break; 214 case AsmToken::Identifier: 215 Out->os() << "identifier: " << Lexer.getTok().getString(); 216 break; 217 case AsmToken::Integer: 218 Out->os() << "int: " << Lexer.getTok().getString(); 219 break; 220 case AsmToken::Real: 221 Out->os() << "real: " << Lexer.getTok().getString(); 222 break; 223 case AsmToken::Register: 224 Out->os() << "register: " << Lexer.getTok().getRegVal(); 225 break; 226 case AsmToken::String: 227 Out->os() << "string: " << Lexer.getTok().getString(); 228 break; 229 230 case AsmToken::Amp: Out->os() << "Amp"; break; 231 case AsmToken::AmpAmp: Out->os() << "AmpAmp"; break; 232 case AsmToken::At: Out->os() << "At"; break; 233 case AsmToken::Caret: Out->os() << "Caret"; break; 234 case AsmToken::Colon: Out->os() << "Colon"; break; 235 case AsmToken::Comma: Out->os() << "Comma"; break; 236 case AsmToken::Dollar: Out->os() << "Dollar"; break; 237 case AsmToken::Dot: Out->os() << "Dot"; break; 238 case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break; 239 case AsmToken::Eof: Out->os() << "Eof"; break; 240 case AsmToken::Equal: Out->os() << "Equal"; break; 241 case AsmToken::EqualEqual: Out->os() << "EqualEqual"; break; 242 case AsmToken::Exclaim: Out->os() << "Exclaim"; break; 243 case AsmToken::ExclaimEqual: Out->os() << "ExclaimEqual"; break; 244 case AsmToken::Greater: Out->os() << "Greater"; break; 245 case AsmToken::GreaterEqual: Out->os() << "GreaterEqual"; break; 246 case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break; 247 case AsmToken::Hash: Out->os() << "Hash"; break; 248 case AsmToken::LBrac: Out->os() << "LBrac"; break; 249 case AsmToken::LCurly: Out->os() << "LCurly"; break; 250 case AsmToken::LParen: Out->os() << "LParen"; break; 251 case AsmToken::Less: Out->os() << "Less"; break; 252 case AsmToken::LessEqual: Out->os() << "LessEqual"; break; 253 case AsmToken::LessGreater: Out->os() << "LessGreater"; break; 254 case AsmToken::LessLess: Out->os() << "LessLess"; break; 255 case AsmToken::Minus: Out->os() << "Minus"; break; 256 case AsmToken::Percent: Out->os() << "Percent"; break; 257 case AsmToken::Pipe: Out->os() << "Pipe"; break; 258 case AsmToken::PipePipe: Out->os() << "PipePipe"; break; 259 case AsmToken::Plus: Out->os() << "Plus"; break; 260 case AsmToken::RBrac: Out->os() << "RBrac"; break; 261 case AsmToken::RCurly: Out->os() << "RCurly"; break; 262 case AsmToken::RParen: Out->os() << "RParen"; break; 263 case AsmToken::Slash: Out->os() << "Slash"; break; 264 case AsmToken::Star: Out->os() << "Star"; break; 265 case AsmToken::Tilde: Out->os() << "Tilde"; break; 266 } 267 268 // Print the token string. 269 Out->os() << " (\""; 270 Out->os().write_escaped(Tok.getString()); 271 Out->os() << "\")\n"; 272 } 273 274 // Keep output if no errors. 275 if (Error == 0) Out->keep(); 276 277 return Error; 278 } 279 280 static int AssembleInput(const char *ProgName) { 281 const Target *TheTarget = GetTarget(ProgName); 282 if (!TheTarget) 283 return 1; 284 285 std::string Error; 286 MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, &Error); 287 if (Buffer == 0) { 288 errs() << ProgName << ": "; 289 if (Error.size()) 290 errs() << Error << "\n"; 291 else 292 errs() << "input file didn't read correctly.\n"; 293 return 1; 294 } 295 296 SourceMgr SrcMgr; 297 298 // Tell SrcMgr about this buffer, which is what the parser will pick up. 299 SrcMgr.AddNewSourceBuffer(Buffer, SMLoc()); 300 301 // Record the location of the include directories so that the lexer can find 302 // it later. 303 SrcMgr.setIncludeDirs(IncludeDirs); 304 305 306 llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName)); 307 assert(MAI && "Unable to create target asm info!"); 308 309 MCContext Ctx(*MAI); 310 311 // Package up features to be passed to target/subtarget 312 std::string FeaturesStr; 313 if (MCPU.size()) { 314 SubtargetFeatures Features; 315 Features.setCPU(MCPU); 316 FeaturesStr = Features.getString(); 317 } 318 319 // FIXME: We shouldn't need to do this (and link in codegen). 320 // When we split this out, we should do it in a way that makes 321 // it straightforward to switch subtargets on the fly (.e.g, 322 // the .cpu and .code16 directives). 323 OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName, 324 FeaturesStr)); 325 326 if (!TM) { 327 errs() << ProgName << ": error: could not create target for triple '" 328 << TripleName << "'.\n"; 329 return 1; 330 } 331 332 OwningPtr<tool_output_file> Out(GetOutputStream()); 333 if (!Out) 334 return 1; 335 336 formatted_raw_ostream FOS(Out->os()); 337 OwningPtr<MCStreamer> Str; 338 339 if (FileType == OFT_AssemblyFile) { 340 MCInstPrinter *IP = 341 TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI); 342 MCCodeEmitter *CE = 0; 343 if (ShowEncoding) 344 CE = TheTarget->createCodeEmitter(*TM, Ctx); 345 Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, 346 TM->getTargetData()->isLittleEndian(), 347 /*asmverbose*/true, IP, CE, ShowInst)); 348 } else if (FileType == OFT_Null) { 349 Str.reset(createNullStreamer(Ctx)); 350 } else { 351 assert(FileType == OFT_ObjectFile && "Invalid file type!"); 352 MCCodeEmitter *CE = TheTarget->createCodeEmitter(*TM, Ctx); 353 TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName); 354 Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB, 355 FOS, CE, RelaxAll)); 356 } 357 358 if (EnableLogging) { 359 Str.reset(createLoggingStreamer(Str.take(), errs())); 360 } 361 362 OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx, 363 *Str.get(), *MAI)); 364 OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*Parser, *TM)); 365 if (!TAP) { 366 errs() << ProgName 367 << ": error: this target does not support assembly parsing.\n"; 368 return 1; 369 } 370 371 Parser->setShowParsedOperands(ShowInstOperands); 372 Parser->setTargetParser(*TAP.get()); 373 374 int Res = Parser->Run(NoInitialTextSection); 375 376 // Keep output if no errors. 377 if (Res == 0) Out->keep(); 378 379 return Res; 380 } 381 382 static int DisassembleInput(const char *ProgName, bool Enhanced) { 383 const Target *TheTarget = GetTarget(ProgName); 384 if (!TheTarget) 385 return 0; 386 387 std::string ErrorMessage; 388 389 MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, 390 &ErrorMessage); 391 392 if (Buffer == 0) { 393 errs() << ProgName << ": "; 394 if (ErrorMessage.size()) 395 errs() << ErrorMessage << "\n"; 396 else 397 errs() << "input file didn't read correctly.\n"; 398 return 1; 399 } 400 401 OwningPtr<tool_output_file> Out(GetOutputStream()); 402 if (!Out) 403 return 1; 404 405 int Res; 406 if (Enhanced) 407 Res = Disassembler::disassembleEnhanced(TripleName, *Buffer, Out->os()); 408 else 409 Res = Disassembler::disassemble(*TheTarget, TripleName, *Buffer, Out->os()); 410 411 // Keep output if no errors. 412 if (Res == 0) Out->keep(); 413 414 return Res; 415 } 416 417 418 int main(int argc, char **argv) { 419 // Print a stack trace if we signal out. 420 sys::PrintStackTraceOnErrorSignal(); 421 PrettyStackTraceProgram X(argc, argv); 422 llvm_shutdown_obj Y; // Call llvm_shutdown() on exit. 423 424 // Initialize targets and assembly printers/parsers. 425 llvm::InitializeAllTargetInfos(); 426 // FIXME: We shouldn't need to initialize the Target(Machine)s. 427 llvm::InitializeAllTargets(); 428 llvm::InitializeAllAsmPrinters(); 429 llvm::InitializeAllAsmParsers(); 430 llvm::InitializeAllDisassemblers(); 431 432 cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n"); 433 TripleName = Triple::normalize(TripleName); 434 435 switch (Action) { 436 default: 437 case AC_AsLex: 438 return AsLexInput(argv[0]); 439 case AC_Assemble: 440 return AssembleInput(argv[0]); 441 case AC_Disassemble: 442 return DisassembleInput(argv[0], false); 443 case AC_EDisassemble: 444 return DisassembleInput(argv[0], true); 445 } 446 447 return 0; 448 } 449 450