llvm-mc.cpp revision e78085a3c03de648a481e9751c3094c517bd7123
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/MCInstrInfo.h"
21#include "llvm/MC/MCObjectFileInfo.h"
22#include "llvm/MC/MCRegisterInfo.h"
23#include "llvm/MC/MCSectionMachO.h"
24#include "llvm/MC/MCStreamer.h"
25#include "llvm/MC/MCSubtargetInfo.h"
26#include "llvm/MC/SubtargetFeature.h"
27#include "llvm/Target/TargetAsmBackend.h"
28#include "llvm/Target/TargetAsmParser.h"
29#include "llvm/Target/TargetData.h"
30#include "llvm/Target/TargetRegistry.h"
31#include "llvm/Target/TargetSelect.h"
32#include "llvm/ADT/OwningPtr.h"
33#include "llvm/Support/CommandLine.h"
34#include "llvm/Support/FileUtilities.h"
35#include "llvm/Support/FormattedStream.h"
36#include "llvm/Support/ManagedStatic.h"
37#include "llvm/Support/MemoryBuffer.h"
38#include "llvm/Support/PrettyStackTrace.h"
39#include "llvm/Support/SourceMgr.h"
40#include "llvm/Support/ToolOutputFile.h"
41#include "llvm/Support/Host.h"
42#include "llvm/Support/Signals.h"
43#include "llvm/Support/system_error.h"
44#include "Disassembler.h"
45using namespace llvm;
46
47static cl::opt<std::string>
48InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
49
50static cl::opt<std::string>
51OutputFilename("o", cl::desc("Output filename"),
52               cl::value_desc("filename"));
53
54static cl::opt<bool>
55ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
56
57static cl::opt<bool>
58ShowInst("show-inst", cl::desc("Show internal instruction representation"));
59
60static cl::opt<bool>
61ShowInstOperands("show-inst-operands",
62                 cl::desc("Show instructions operands as parsed"));
63
64static cl::opt<unsigned>
65OutputAsmVariant("output-asm-variant",
66                 cl::desc("Syntax variant to use for output printing"));
67
68static cl::opt<bool>
69RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
70
71static cl::opt<bool>
72NoExecStack("mc-no-exec-stack", cl::desc("File doesn't need an exec stack"));
73
74static cl::opt<bool>
75EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
76
77enum OutputFileType {
78  OFT_Null,
79  OFT_AssemblyFile,
80  OFT_ObjectFile
81};
82static cl::opt<OutputFileType>
83FileType("filetype", cl::init(OFT_AssemblyFile),
84  cl::desc("Choose an output file type:"),
85  cl::values(
86       clEnumValN(OFT_AssemblyFile, "asm",
87                  "Emit an assembly ('.s') file"),
88       clEnumValN(OFT_Null, "null",
89                  "Don't emit anything (for timing purposes)"),
90       clEnumValN(OFT_ObjectFile, "obj",
91                  "Emit a native object ('.o') file"),
92       clEnumValEnd));
93
94static cl::list<std::string>
95IncludeDirs("I", cl::desc("Directory of include files"),
96            cl::value_desc("directory"), cl::Prefix);
97
98static cl::opt<std::string>
99ArchName("arch", cl::desc("Target arch to assemble for, "
100                          "see -version for available targets"));
101
102static cl::opt<std::string>
103TripleName("triple", cl::desc("Target triple to assemble for, "
104                              "see -version for available targets"));
105
106static cl::opt<std::string>
107MCPU("mcpu",
108     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
109     cl::value_desc("cpu-name"),
110     cl::init(""));
111
112static cl::opt<Reloc::Model>
113RelocModel("relocation-model",
114             cl::desc("Choose relocation model"),
115             cl::init(Reloc::Default),
116             cl::values(
117            clEnumValN(Reloc::Default, "default",
118                       "Target default relocation model"),
119            clEnumValN(Reloc::Static, "static",
120                       "Non-relocatable code"),
121            clEnumValN(Reloc::PIC_, "pic",
122                       "Fully relocatable, position independent code"),
123            clEnumValN(Reloc::DynamicNoPIC, "dynamic-no-pic",
124                       "Relocatable external references, non-relocatable code"),
125            clEnumValEnd));
126
127static cl::opt<llvm::CodeModel::Model>
128CMModel("code-model",
129        cl::desc("Choose code model"),
130        cl::init(CodeModel::Default),
131        cl::values(clEnumValN(CodeModel::Default, "default",
132                              "Target default code model"),
133                   clEnumValN(CodeModel::Small, "small",
134                              "Small code model"),
135                   clEnumValN(CodeModel::Kernel, "kernel",
136                              "Kernel code model"),
137                   clEnumValN(CodeModel::Medium, "medium",
138                              "Medium code model"),
139                   clEnumValN(CodeModel::Large, "large",
140                              "Large code model"),
141                   clEnumValEnd));
142
143static cl::opt<bool>
144NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
145                                   "in the text section"));
146
147static cl::opt<bool>
148SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
149
150enum ActionType {
151  AC_AsLex,
152  AC_Assemble,
153  AC_Disassemble,
154  AC_EDisassemble
155};
156
157static cl::opt<ActionType>
158Action(cl::desc("Action to perform:"),
159       cl::init(AC_Assemble),
160       cl::values(clEnumValN(AC_AsLex, "as-lex",
161                             "Lex tokens from a .s file"),
162                  clEnumValN(AC_Assemble, "assemble",
163                             "Assemble a .s file (default)"),
164                  clEnumValN(AC_Disassemble, "disassemble",
165                             "Disassemble strings of hex bytes"),
166                  clEnumValN(AC_EDisassemble, "edis",
167                             "Enhanced disassembly of strings of hex bytes"),
168                  clEnumValEnd));
169
170static const Target *GetTarget(const char *ProgName) {
171  // Figure out the target triple.
172  if (TripleName.empty())
173    TripleName = sys::getHostTriple();
174  if (!ArchName.empty()) {
175    llvm::Triple TT(TripleName);
176    TT.setArchName(ArchName);
177    TripleName = TT.str();
178  }
179
180  // Get the target specific parser.
181  std::string Error;
182  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
183  if (TheTarget)
184    return TheTarget;
185
186  errs() << ProgName << ": error: unable to get target for '" << TripleName
187         << "', see --version and --triple.\n";
188  return 0;
189}
190
191static tool_output_file *GetOutputStream() {
192  if (OutputFilename == "")
193    OutputFilename = "-";
194
195  std::string Err;
196  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
197                                               raw_fd_ostream::F_Binary);
198  if (!Err.empty()) {
199    errs() << Err << '\n';
200    delete Out;
201    return 0;
202  }
203
204  return Out;
205}
206
207static int AsLexInput(const char *ProgName) {
208  OwningPtr<MemoryBuffer> BufferPtr;
209  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
210    errs() << ProgName << ": " << ec.message() << '\n';
211    return 1;
212  }
213  MemoryBuffer *Buffer = BufferPtr.take();
214
215  SourceMgr SrcMgr;
216
217  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
218  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
219
220  // Record the location of the include directories so that the lexer can find
221  // it later.
222  SrcMgr.setIncludeDirs(IncludeDirs);
223
224  const Target *TheTarget = GetTarget(ProgName);
225  if (!TheTarget)
226    return 1;
227
228  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
229  assert(MAI && "Unable to create target asm info!");
230
231  AsmLexer Lexer(*MAI);
232  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
233
234  OwningPtr<tool_output_file> Out(GetOutputStream());
235  if (!Out)
236    return 1;
237
238  bool Error = false;
239  while (Lexer.Lex().isNot(AsmToken::Eof)) {
240    AsmToken Tok = Lexer.getTok();
241
242    switch (Tok.getKind()) {
243    default:
244      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
245      Error = true;
246      break;
247    case AsmToken::Error:
248      Error = true; // error already printed.
249      break;
250    case AsmToken::Identifier:
251      Out->os() << "identifier: " << Lexer.getTok().getString();
252      break;
253    case AsmToken::Integer:
254      Out->os() << "int: " << Lexer.getTok().getString();
255      break;
256    case AsmToken::Real:
257      Out->os() << "real: " << Lexer.getTok().getString();
258      break;
259    case AsmToken::Register:
260      Out->os() << "register: " << Lexer.getTok().getRegVal();
261      break;
262    case AsmToken::String:
263      Out->os() << "string: " << Lexer.getTok().getString();
264      break;
265
266    case AsmToken::Amp:            Out->os() << "Amp"; break;
267    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
268    case AsmToken::At:             Out->os() << "At"; break;
269    case AsmToken::Caret:          Out->os() << "Caret"; break;
270    case AsmToken::Colon:          Out->os() << "Colon"; break;
271    case AsmToken::Comma:          Out->os() << "Comma"; break;
272    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
273    case AsmToken::Dot:            Out->os() << "Dot"; break;
274    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
275    case AsmToken::Eof:            Out->os() << "Eof"; break;
276    case AsmToken::Equal:          Out->os() << "Equal"; break;
277    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
278    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
279    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
280    case AsmToken::Greater:        Out->os() << "Greater"; break;
281    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
282    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
283    case AsmToken::Hash:           Out->os() << "Hash"; break;
284    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
285    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
286    case AsmToken::LParen:         Out->os() << "LParen"; break;
287    case AsmToken::Less:           Out->os() << "Less"; break;
288    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
289    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
290    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
291    case AsmToken::Minus:          Out->os() << "Minus"; break;
292    case AsmToken::Percent:        Out->os() << "Percent"; break;
293    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
294    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
295    case AsmToken::Plus:           Out->os() << "Plus"; break;
296    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
297    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
298    case AsmToken::RParen:         Out->os() << "RParen"; break;
299    case AsmToken::Slash:          Out->os() << "Slash"; break;
300    case AsmToken::Star:           Out->os() << "Star"; break;
301    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
302    }
303
304    // Print the token string.
305    Out->os() << " (\"";
306    Out->os().write_escaped(Tok.getString());
307    Out->os() << "\")\n";
308  }
309
310  // Keep output if no errors.
311  if (Error == 0) Out->keep();
312
313  return Error;
314}
315
316static int AssembleInput(const char *ProgName) {
317  const Target *TheTarget = GetTarget(ProgName);
318  if (!TheTarget)
319    return 1;
320
321  OwningPtr<MemoryBuffer> BufferPtr;
322  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
323    errs() << ProgName << ": " << ec.message() << '\n';
324    return 1;
325  }
326  MemoryBuffer *Buffer = BufferPtr.take();
327
328  SourceMgr SrcMgr;
329
330  // Tell SrcMgr about this buffer, which is what the parser will pick up.
331  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
332
333  // Record the location of the include directories so that the lexer can find
334  // it later.
335  SrcMgr.setIncludeDirs(IncludeDirs);
336
337
338  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
339  assert(MAI && "Unable to create target asm info!");
340
341  llvm::OwningPtr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
342  assert(MRI && "Unable to create target register info!");
343
344  // Package up features to be passed to target/subtarget
345  std::string FeaturesStr;
346
347  // FIXME: This is not pretty. MCContext has a ptr to MCObjectFileInfo and
348  // MCObjectFileInfo needs a MCContext reference in order to initialize itself.
349  OwningPtr<MCObjectFileInfo> MOFI(new MCObjectFileInfo());
350  MCContext Ctx(*MAI, *MRI, MOFI.get());
351  MOFI->InitMCObjectFileInfo(TripleName, RelocModel, CMModel, Ctx);
352
353  if (SaveTempLabels)
354    Ctx.setAllowTemporaryLabels(false);
355
356  OwningPtr<tool_output_file> Out(GetOutputStream());
357  if (!Out)
358    return 1;
359
360  formatted_raw_ostream FOS(Out->os());
361  OwningPtr<MCStreamer> Str;
362
363  OwningPtr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo());
364  OwningPtr<MCSubtargetInfo>
365    STI(TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
366
367  // FIXME: There is a bit of code duplication with addPassesToEmitFile.
368  if (FileType == OFT_AssemblyFile) {
369    MCInstPrinter *IP =
370      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
371    MCCodeEmitter *CE = 0;
372    TargetAsmBackend *TAB = 0;
373    if (ShowEncoding) {
374      CE = TheTarget->createCodeEmitter(*MCII, *STI, Ctx);
375      TAB = TheTarget->createAsmBackend(TripleName);
376    }
377    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
378                                           /*useLoc*/ true,
379                                           /*useCFI*/ true, IP, CE, TAB,
380                                           ShowInst));
381  } else if (FileType == OFT_Null) {
382    Str.reset(createNullStreamer(Ctx));
383  } else {
384    assert(FileType == OFT_ObjectFile && "Invalid file type!");
385    MCCodeEmitter *CE = TheTarget->createCodeEmitter(*MCII, *STI, Ctx);
386    TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName);
387    Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB,
388                                              FOS, CE, RelaxAll,
389                                              NoExecStack));
390  }
391
392  if (EnableLogging) {
393    Str.reset(createLoggingStreamer(Str.take(), errs()));
394  }
395
396  OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
397                                                   *Str.get(), *MAI));
398  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*STI, *Parser));
399  if (!TAP) {
400    errs() << ProgName
401           << ": error: this target does not support assembly parsing.\n";
402    return 1;
403  }
404
405  Parser->setShowParsedOperands(ShowInstOperands);
406  Parser->setTargetParser(*TAP.get());
407
408  int Res = Parser->Run(NoInitialTextSection);
409
410  // Keep output if no errors.
411  if (Res == 0) Out->keep();
412
413  return Res;
414}
415
416static int DisassembleInput(const char *ProgName, bool Enhanced) {
417  const Target *TheTarget = GetTarget(ProgName);
418  if (!TheTarget)
419    return 0;
420
421  OwningPtr<MemoryBuffer> Buffer;
422  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, Buffer)) {
423    errs() << ProgName << ": " << ec.message() << '\n';
424    return 1;
425  }
426
427  OwningPtr<tool_output_file> Out(GetOutputStream());
428  if (!Out)
429    return 1;
430
431  int Res;
432  if (Enhanced) {
433    Res =
434      Disassembler::disassembleEnhanced(TripleName, *Buffer.take(), Out->os());
435  } else {
436    Res = Disassembler::disassemble(*TheTarget, TripleName,
437                                    *Buffer.take(), Out->os());
438  }
439
440  // Keep output if no errors.
441  if (Res == 0) Out->keep();
442
443  return Res;
444}
445
446
447int main(int argc, char **argv) {
448  // Print a stack trace if we signal out.
449  sys::PrintStackTraceOnErrorSignal();
450  PrettyStackTraceProgram X(argc, argv);
451  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
452
453  // Initialize targets and assembly printers/parsers.
454  llvm::InitializeAllTargetInfos();
455  // FIXME: We shouldn't need to initialize the Target(Machine)s.
456  llvm::InitializeAllTargets();
457  llvm::InitializeAllTargetMCs();
458  llvm::InitializeAllAsmPrinters();
459  llvm::InitializeAllAsmParsers();
460  llvm::InitializeAllDisassemblers();
461
462  // Register the target printer for --version.
463  // FIXME: Remove when we stop initializing the Target(Machine)s above.
464  cl::AddExtraVersionPrinter(TargetRegistry::printRegisteredTargetsForVersion);
465
466  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
467  TripleName = Triple::normalize(TripleName);
468
469  switch (Action) {
470  default:
471  case AC_AsLex:
472    return AsLexInput(argv[0]);
473  case AC_Assemble:
474    return AssembleInput(argv[0]);
475  case AC_Disassemble:
476    return DisassembleInput(argv[0], false);
477  case AC_EDisassemble:
478    return DisassembleInput(argv[0], true);
479  }
480
481  return 0;
482}
483
484