tools/llvm-mc/llvm-mc.cpp

   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/MC/MCSubtargetInfo.h"
  23 #include "llvm/MC/SubtargetFeature.h"
  24 #include "llvm/Target/TargetAsmBackend.h"
  25 #include "llvm/Target/TargetAsmParser.h"
  26 #include "llvm/Target/TargetData.h"
  27 #include "llvm/Target/TargetRegistry.h"
  28 #include "llvm/Target/TargetAsmInfo.h"  // FIXME.
  29 #include "llvm/Target/TargetLowering.h"  // FIXME.
  30 #include "llvm/Target/TargetLoweringObjectFile.h"  // FIXME.
  31 #include "llvm/Target/TargetMachine.h"  // FIXME.
  32 #include "llvm/Target/TargetSelect.h"
  33 #include "llvm/ADT/OwningPtr.h"
  34 #include "llvm/Support/CommandLine.h"
  35 #include "llvm/Support/FileUtilities.h"
  36 #include "llvm/Support/FormattedStream.h"
  37 #include "llvm/Support/ManagedStatic.h"
  38 #include "llvm/Support/MemoryBuffer.h"
  39 #include "llvm/Support/PrettyStackTrace.h"
  40 #include "llvm/Support/SourceMgr.h"
  41 #include "llvm/Support/ToolOutputFile.h"
  42 #include "llvm/Support/Host.h"
  43 #include "llvm/Support/Signals.h"
  44 #include "llvm/Support/system_error.h"
  45 #include "Disassembler.h"
  46 using namespace llvm;
  47
  48 static cl::opt<std::string>
  49 InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
  50
  51 static cl::opt<std::string>
  52 OutputFilename("o", cl::desc("Output filename"),
  53                cl::value_desc("filename"));
  54
  55 static cl::opt<bool>
  56 ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
  57
  58 static cl::opt<bool>
  59 ShowInst("show-inst", cl::desc("Show internal instruction representation"));
  60
  61 static cl::opt<bool>
  62 ShowInstOperands("show-inst-operands",
  63                  cl::desc("Show instructions operands as parsed"));
  64
  65 static cl::opt<unsigned>
  66 OutputAsmVariant("output-asm-variant",
  67                  cl::desc("Syntax variant to use for output printing"));
  68
  69 static cl::opt<bool>
  70 RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
  71
  72 static cl::opt<bool>
  73 NoExecStack("mc-no-exec-stack", cl::desc("File doesn't need an exec stack"));
  74
  75 static cl::opt<bool>
  76 EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
  77
  78 enum OutputFileType {
  79   OFT_Null,
  80   OFT_AssemblyFile,
  81   OFT_ObjectFile
  82 };
  83 static cl::opt<OutputFileType>
  84 FileType("filetype", cl::init(OFT_AssemblyFile),
  85   cl::desc("Choose an output file type:"),
  86   cl::values(
  87        clEnumValN(OFT_AssemblyFile, "asm",
  88                   "Emit an assembly ('.s') file"),
  89        clEnumValN(OFT_Null, "null",
  90                   "Don't emit anything (for timing purposes)"),
  91        clEnumValN(OFT_ObjectFile, "obj",
  92                   "Emit a native object ('.o') file"),
  93        clEnumValEnd));
  94
  95 static cl::list<std::string>
  96 IncludeDirs("I", cl::desc("Directory of include files"),
  97             cl::value_desc("directory"), cl::Prefix);
  98
  99 static cl::opt<std::string>
 100 ArchName("arch", cl::desc("Target arch to assemble for, "
 101                           "see -version for available targets"));
 102
 103 static cl::opt<std::string>
 104 TripleName("triple", cl::desc("Target triple to assemble for, "
 105                               "see -version for available targets"));
 106
 107 static cl::opt<std::string>
 108 MCPU("mcpu",
 109      cl::desc("Target a specific cpu type (-mcpu=help for details)"),
 110      cl::value_desc("cpu-name"),
 111      cl::init(""));
 112
 113 static cl::opt<bool>
 114 NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
 115                                    "in the text section"));
 116
 117 static cl::opt<bool>
 118 SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
 119
 120 enum ActionType {
 121   AC_AsLex,
 122   AC_Assemble,
 123   AC_Disassemble,
 124   AC_EDisassemble
 125 };
 126
 127 static cl::opt<ActionType>
 128 Action(cl::desc("Action to perform:"),
 129        cl::init(AC_Assemble),
 130        cl::values(clEnumValN(AC_AsLex, "as-lex",
 131                              "Lex tokens from a .s file"),
 132                   clEnumValN(AC_Assemble, "assemble",
 133                              "Assemble a .s file (default)"),
 134                   clEnumValN(AC_Disassemble, "disassemble",
 135                              "Disassemble strings of hex bytes"),
 136                   clEnumValN(AC_EDisassemble, "edis",
 137                              "Enhanced disassembly of strings of hex bytes"),
 138                   clEnumValEnd));
 139
 140 static const Target *GetTarget(const char *ProgName) {
 141   // Figure out the target triple.
 142   if (TripleName.empty())
 143     TripleName = sys::getHostTriple();
 144   if (!ArchName.empty()) {
 145     llvm::Triple TT(TripleName);
 146     TT.setArchName(ArchName);
 147     TripleName = TT.str();
 148   }
 149
 150   // Get the target specific parser.
 151   std::string Error;
 152   const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
 153   if (TheTarget)
 154     return TheTarget;
 155
 156   errs() << ProgName << ": error: unable to get target for '" << TripleName
 157          << "', see --version and --triple.\n";
 158   return 0;
 159 }
 160
 161 static tool_output_file *GetOutputStream() {
 162   if (OutputFilename == "")
 163     OutputFilename = "-";
 164
 165   std::string Err;
 166   tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
 167                                                raw_fd_ostream::F_Binary);
 168   if (!Err.empty()) {
 169     errs() << Err << '\n';
 170     delete Out;
 171     return 0;
 172   }
 173
 174   return Out;
 175 }
 176
 177 static int AsLexInput(const char *ProgName) {
 178   OwningPtr<MemoryBuffer> BufferPtr;
 179   if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
 180     errs() << ProgName << ": " << ec.message() << '\n';
 181     return 1;
 182   }
 183   MemoryBuffer *Buffer = BufferPtr.take();
 184
 185   SourceMgr SrcMgr;
 186
 187   // Tell SrcMgr about this buffer, which is what TGParser will pick up.
 188   SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
 189
 190   // Record the location of the include directories so that the lexer can find
 191   // it later.
 192   SrcMgr.setIncludeDirs(IncludeDirs);
 193
 194   const Target *TheTarget = GetTarget(ProgName);
 195   if (!TheTarget)
 196     return 1;
 197
 198   llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
 199   assert(MAI && "Unable to create target asm info!");
 200
 201   AsmLexer Lexer(*MAI);
 202   Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
 203
 204   OwningPtr<tool_output_file> Out(GetOutputStream());
 205   if (!Out)
 206     return 1;
 207
 208   bool Error = false;
 209   while (Lexer.Lex().isNot(AsmToken::Eof)) {
 210     AsmToken Tok = Lexer.getTok();
 211
 212     switch (Tok.getKind()) {
 213     default:
 214       SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
 215       Error = true;
 216       break;
 217     case AsmToken::Error:
 218       Error = true; // error already printed.
 219       break;
 220     case AsmToken::Identifier:
 221       Out->os() << "identifier: " << Lexer.getTok().getString();
 222       break;
 223     case AsmToken::Integer:
 224       Out->os() << "int: " << Lexer.getTok().getString();
 225       break;
 226     case AsmToken::Real:
 227       Out->os() << "real: " << Lexer.getTok().getString();
 228       break;
 229     case AsmToken::Register:
 230       Out->os() << "register: " << Lexer.getTok().getRegVal();
 231       break;
 232     case AsmToken::String:
 233       Out->os() << "string: " << Lexer.getTok().getString();
 234       break;
 235
 236     case AsmToken::Amp:            Out->os() << "Amp"; break;
 237     case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
 238     case AsmToken::At:             Out->os() << "At"; break;
 239     case AsmToken::Caret:          Out->os() << "Caret"; break;
 240     case AsmToken::Colon:          Out->os() << "Colon"; break;
 241     case AsmToken::Comma:          Out->os() << "Comma"; break;
 242     case AsmToken::Dollar:         Out->os() << "Dollar"; break;
 243     case AsmToken::Dot:            Out->os() << "Dot"; break;
 244     case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
 245     case AsmToken::Eof:            Out->os() << "Eof"; break;
 246     case AsmToken::Equal:          Out->os() << "Equal"; break;
 247     case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
 248     case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
 249     case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
 250     case AsmToken::Greater:        Out->os() << "Greater"; break;
 251     case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
 252     case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
 253     case AsmToken::Hash:           Out->os() << "Hash"; break;
 254     case AsmToken::LBrac:          Out->os() << "LBrac"; break;
 255     case AsmToken::LCurly:         Out->os() << "LCurly"; break;
 256     case AsmToken::LParen:         Out->os() << "LParen"; break;
 257     case AsmToken::Less:           Out->os() << "Less"; break;
 258     case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
 259     case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
 260     case AsmToken::LessLess:       Out->os() << "LessLess"; break;
 261     case AsmToken::Minus:          Out->os() << "Minus"; break;
 262     case AsmToken::Percent:        Out->os() << "Percent"; break;
 263     case AsmToken::Pipe:           Out->os() << "Pipe"; break;
 264     case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
 265     case AsmToken::Plus:           Out->os() << "Plus"; break;
 266     case AsmToken::RBrac:          Out->os() << "RBrac"; break;
 267     case AsmToken::RCurly:         Out->os() << "RCurly"; break;
 268     case AsmToken::RParen:         Out->os() << "RParen"; break;
 269     case AsmToken::Slash:          Out->os() << "Slash"; break;
 270     case AsmToken::Star:           Out->os() << "Star"; break;
 271     case AsmToken::Tilde:          Out->os() << "Tilde"; break;
 272     }
 273
 274     // Print the token string.
 275     Out->os() << " (\"";
 276     Out->os().write_escaped(Tok.getString());
 277     Out->os() << "\")\n";
 278   }
 279
 280   // Keep output if no errors.
 281   if (Error == 0) Out->keep();
 282
 283   return Error;
 284 }
 285
 286 static int AssembleInput(const char *ProgName) {
 287   const Target *TheTarget = GetTarget(ProgName);
 288   if (!TheTarget)
 289     return 1;
 290
 291   OwningPtr<MemoryBuffer> BufferPtr;
 292   if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
 293     errs() << ProgName << ": " << ec.message() << '\n';
 294     return 1;
 295   }
 296   MemoryBuffer *Buffer = BufferPtr.take();
 297
 298   SourceMgr SrcMgr;
 299
 300   // Tell SrcMgr about this buffer, which is what the parser will pick up.
 301   SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
 302
 303   // Record the location of the include directories so that the lexer can find
 304   // it later.
 305   SrcMgr.setIncludeDirs(IncludeDirs);
 306
 307
 308   llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
 309   assert(MAI && "Unable to create target asm info!");
 310
 311   // Package up features to be passed to target/subtarget
 312   std::string FeaturesStr;
 313
 314   // FIXME: We shouldn't need to do this (and link in codegen).
 315   //        When we split this out, we should do it in a way that makes
 316   //        it straightforward to switch subtargets on the fly (.e.g,
 317   //        the .cpu and .code16 directives).
 318   OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
 319                                                              MCPU,
 320                                                              FeaturesStr));
 321
 322   if (!TM) {
 323     errs() << ProgName << ": error: could not create target for triple '"
 324            << TripleName << "'.\n";
 325     return 1;
 326   }
 327
 328   const TargetAsmInfo *tai = new TargetAsmInfo(*TM);
 329   MCContext Ctx(*MAI, tai);
 330   if (SaveTempLabels)
 331     Ctx.setAllowTemporaryLabels(false);
 332
 333   OwningPtr<tool_output_file> Out(GetOutputStream());
 334   if (!Out)
 335     return 1;
 336
 337   formatted_raw_ostream FOS(Out->os());
 338   OwningPtr<MCStreamer> Str;
 339
 340   const TargetLoweringObjectFile &TLOF =
 341     TM->getTargetLowering()->getObjFileLowering();
 342   const_cast<TargetLoweringObjectFile&>(TLOF).Initialize(Ctx, *TM);
 343
 344   OwningPtr<MCSubtargetInfo>
 345     STI(TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
 346
 347   // FIXME: There is a bit of code duplication with addPassesToEmitFile.
 348   if (FileType == OFT_AssemblyFile) {
 349     MCInstPrinter *IP =
 350       TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
 351     MCCodeEmitter *CE = 0;
 352     TargetAsmBackend *TAB = 0;
 353     if (ShowEncoding) {
 354       CE = TheTarget->createCodeEmitter(*TM, Ctx);
 355       TAB = TheTarget->createAsmBackend(TripleName);
 356     }
 357     Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
 358                                            /*useLoc*/ true,
 359                                            /*useCFI*/ true, IP, CE, TAB,
 360                                            ShowInst));
 361   } else if (FileType == OFT_Null) {
 362     Str.reset(createNullStreamer(Ctx));
 363   } else {
 364     assert(FileType == OFT_ObjectFile && "Invalid file type!");
 365     MCCodeEmitter *CE = TheTarget->createCodeEmitter(*TM, Ctx);
 366     TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName);
 367     Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB,
 368                                               FOS, CE, RelaxAll,
 369                                               NoExecStack));
 370   }
 371
 372   if (EnableLogging) {
 373     Str.reset(createLoggingStreamer(Str.take(), errs()));
 374   }
 375
 376   OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
 377                                                    *Str.get(), *MAI));
 378   OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*STI, *Parser));
 379   if (!TAP) {
 380     errs() << ProgName
 381            << ": error: this target does not support assembly parsing.\n";
 382     return 1;
 383   }
 384
 385   Parser->setShowParsedOperands(ShowInstOperands);
 386   Parser->setTargetParser(*TAP.get());
 387
 388   int Res = Parser->Run(NoInitialTextSection);
 389
 390   // Keep output if no errors.
 391   if (Res == 0) Out->keep();
 392
 393   return Res;
 394 }
 395
 396 static int DisassembleInput(const char *ProgName, bool Enhanced) {
 397   const Target *TheTarget = GetTarget(ProgName);
 398   if (!TheTarget)
 399     return 0;
 400
 401   OwningPtr<MemoryBuffer> Buffer;
 402   if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, Buffer)) {
 403     errs() << ProgName << ": " << ec.message() << '\n';
 404     return 1;
 405   }
 406
 407   OwningPtr<tool_output_file> Out(GetOutputStream());
 408   if (!Out)
 409     return 1;
 410
 411   int Res;
 412   if (Enhanced) {
 413     Res =
 414       Disassembler::disassembleEnhanced(TripleName, *Buffer.take(), Out->os());
 415   } else {
 416     // Package up features to be passed to target/subtarget
 417     std::string FeaturesStr;
 418
 419     // FIXME: We shouldn't need to do this (and link in codegen).
 420     //        When we split this out, we should do it in a way that makes
 421     //        it straightforward to switch subtargets on the fly (.e.g,
 422     //        the .cpu and .code16 directives).
 423     OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
 424                                                                MCPU,
 425                                                                FeaturesStr));
 426
 427     if (!TM) {
 428       errs() << ProgName << ": error: could not create target for triple '"
 429              << TripleName << "'.\n";
 430       return 1;
 431     }
 432
 433     Res = Disassembler::disassemble(*TheTarget, TripleName,
 434                                     *Buffer.take(), Out->os());
 435   }
 436
 437   // Keep output if no errors.
 438   if (Res == 0) Out->keep();
 439
 440   return Res;
 441 }
 442
 443
 444 int main(int argc, char **argv) {
 445   // Print a stack trace if we signal out.
 446   sys::PrintStackTraceOnErrorSignal();
 447   PrettyStackTraceProgram X(argc, argv);
 448   llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
 449
 450   // Initialize targets and assembly printers/parsers.
 451   llvm::InitializeAllTargetInfos();
 452   // FIXME: We shouldn't need to initialize the Target(Machine)s.
 453   llvm::InitializeAllTargets();
 454   llvm::InitializeAllMCSubtargetInfos();
 455   llvm::InitializeAllAsmPrinters();
 456   llvm::InitializeAllAsmParsers();
 457   llvm::InitializeAllDisassemblers();
 458
 459   cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
 460   TripleName = Triple::normalize(TripleName);
 461
 462   switch (Action) {
 463   default:
 464   case AC_AsLex:
 465     return AsLexInput(argv[0]);
 466   case AC_Assemble:
 467     return AssembleInput(argv[0]);
 468   case AC_Disassemble:
 469     return DisassembleInput(argv[0], false);
 470   case AC_EDisassemble:
 471     return DisassembleInput(argv[0], true);
 472   }
 473
 474   return 0;
 475 }
 476