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