llvm-mc.cpp revision 28c85a81a17dd719a254dc00cbeb484774893197
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/MCAsmBackend.h"
18#include "llvm/MC/MCContext.h"
19#include "llvm/MC/MCCodeEmitter.h"
20#include "llvm/MC/MCInstPrinter.h"
21#include "llvm/MC/MCInstrInfo.h"
22#include "llvm/MC/MCObjectFileInfo.h"
23#include "llvm/MC/MCRegisterInfo.h"
24#include "llvm/MC/MCSectionMachO.h"
25#include "llvm/MC/MCStreamer.h"
26#include "llvm/MC/MCSubtargetInfo.h"
27#include "llvm/MC/MCTargetAsmParser.h"
28#include "llvm/MC/SubtargetFeature.h"
29#include "llvm/Target/TargetRegistry.h"
30#include "llvm/Target/TargetSelect.h"
31#include "llvm/ADT/OwningPtr.h"
32#include "llvm/Support/CommandLine.h"
33#include "llvm/Support/FileUtilities.h"
34#include "llvm/Support/FormattedStream.h"
35#include "llvm/Support/ManagedStatic.h"
36#include "llvm/Support/MemoryBuffer.h"
37#include "llvm/Support/PrettyStackTrace.h"
38#include "llvm/Support/SourceMgr.h"
39#include "llvm/Support/ToolOutputFile.h"
40#include "llvm/Support/Host.h"
41#include "llvm/Support/Signals.h"
42#include "llvm/Support/system_error.h"
43#include "Disassembler.h"
44using namespace llvm;
45
46static cl::opt<std::string>
47InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
48
49static cl::opt<std::string>
50OutputFilename("o", cl::desc("Output filename"),
51               cl::value_desc("filename"));
52
53static cl::opt<bool>
54ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
55
56static cl::opt<bool>
57ShowInst("show-inst", cl::desc("Show internal instruction representation"));
58
59static cl::opt<bool>
60ShowInstOperands("show-inst-operands",
61                 cl::desc("Show instructions operands as parsed"));
62
63static cl::opt<unsigned>
64OutputAsmVariant("output-asm-variant",
65                 cl::desc("Syntax variant to use for output printing"));
66
67static cl::opt<bool>
68RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
69
70static cl::opt<bool>
71NoExecStack("mc-no-exec-stack", cl::desc("File doesn't need an exec stack"));
72
73static cl::opt<bool>
74EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
75
76enum OutputFileType {
77  OFT_Null,
78  OFT_AssemblyFile,
79  OFT_ObjectFile
80};
81static cl::opt<OutputFileType>
82FileType("filetype", cl::init(OFT_AssemblyFile),
83  cl::desc("Choose an output file type:"),
84  cl::values(
85       clEnumValN(OFT_AssemblyFile, "asm",
86                  "Emit an assembly ('.s') file"),
87       clEnumValN(OFT_Null, "null",
88                  "Don't emit anything (for timing purposes)"),
89       clEnumValN(OFT_ObjectFile, "obj",
90                  "Emit a native object ('.o') file"),
91       clEnumValEnd));
92
93static cl::list<std::string>
94IncludeDirs("I", cl::desc("Directory of include files"),
95            cl::value_desc("directory"), cl::Prefix);
96
97static cl::opt<std::string>
98ArchName("arch", cl::desc("Target arch to assemble for, "
99                          "see -version for available targets"));
100
101static cl::opt<std::string>
102TripleName("triple", cl::desc("Target triple to assemble for, "
103                              "see -version for available targets"));
104
105static cl::opt<std::string>
106MCPU("mcpu",
107     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
108     cl::value_desc("cpu-name"),
109     cl::init(""));
110
111static cl::opt<Reloc::Model>
112RelocModel("relocation-model",
113             cl::desc("Choose relocation model"),
114             cl::init(Reloc::Default),
115             cl::values(
116            clEnumValN(Reloc::Default, "default",
117                       "Target default relocation model"),
118            clEnumValN(Reloc::Static, "static",
119                       "Non-relocatable code"),
120            clEnumValN(Reloc::PIC_, "pic",
121                       "Fully relocatable, position independent code"),
122            clEnumValN(Reloc::DynamicNoPIC, "dynamic-no-pic",
123                       "Relocatable external references, non-relocatable code"),
124            clEnumValEnd));
125
126static cl::opt<llvm::CodeModel::Model>
127CMModel("code-model",
128        cl::desc("Choose code model"),
129        cl::init(CodeModel::Default),
130        cl::values(clEnumValN(CodeModel::Default, "default",
131                              "Target default code model"),
132                   clEnumValN(CodeModel::Small, "small",
133                              "Small code model"),
134                   clEnumValN(CodeModel::Kernel, "kernel",
135                              "Kernel code model"),
136                   clEnumValN(CodeModel::Medium, "medium",
137                              "Medium code model"),
138                   clEnumValN(CodeModel::Large, "large",
139                              "Large code model"),
140                   clEnumValEnd));
141
142static cl::opt<bool>
143NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
144                                   "in the text section"));
145
146static cl::opt<bool>
147SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
148
149enum ActionType {
150  AC_AsLex,
151  AC_Assemble,
152  AC_Disassemble,
153  AC_EDisassemble
154};
155
156static cl::opt<ActionType>
157Action(cl::desc("Action to perform:"),
158       cl::init(AC_Assemble),
159       cl::values(clEnumValN(AC_AsLex, "as-lex",
160                             "Lex tokens from a .s file"),
161                  clEnumValN(AC_Assemble, "assemble",
162                             "Assemble a .s file (default)"),
163                  clEnumValN(AC_Disassemble, "disassemble",
164                             "Disassemble strings of hex bytes"),
165                  clEnumValN(AC_EDisassemble, "edis",
166                             "Enhanced disassembly of strings of hex bytes"),
167                  clEnumValEnd));
168
169static const Target *GetTarget(const char *ProgName) {
170  // Figure out the target triple.
171  if (TripleName.empty())
172    TripleName = sys::getHostTriple();
173  if (!ArchName.empty()) {
174    llvm::Triple TT(TripleName);
175    TT.setArchName(ArchName);
176    TripleName = TT.str();
177  }
178
179  // Get the target specific parser.
180  std::string Error;
181  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
182  if (TheTarget)
183    return TheTarget;
184
185  errs() << ProgName << ": error: unable to get target for '" << TripleName
186         << "', see --version and --triple.\n";
187  return 0;
188}
189
190static tool_output_file *GetOutputStream() {
191  if (OutputFilename == "")
192    OutputFilename = "-";
193
194  std::string Err;
195  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
196                                               raw_fd_ostream::F_Binary);
197  if (!Err.empty()) {
198    errs() << Err << '\n';
199    delete Out;
200    return 0;
201  }
202
203  return Out;
204}
205
206static int AsLexInput(const char *ProgName) {
207  OwningPtr<MemoryBuffer> BufferPtr;
208  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
209    errs() << ProgName << ": " << ec.message() << '\n';
210    return 1;
211  }
212  MemoryBuffer *Buffer = BufferPtr.take();
213
214  SourceMgr SrcMgr;
215
216  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
217  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
218
219  // Record the location of the include directories so that the lexer can find
220  // it later.
221  SrcMgr.setIncludeDirs(IncludeDirs);
222
223  const Target *TheTarget = GetTarget(ProgName);
224  if (!TheTarget)
225    return 1;
226
227  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
228  assert(MAI && "Unable to create target asm info!");
229
230  AsmLexer Lexer(*MAI);
231  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
232
233  OwningPtr<tool_output_file> Out(GetOutputStream());
234  if (!Out)
235    return 1;
236
237  bool Error = false;
238  while (Lexer.Lex().isNot(AsmToken::Eof)) {
239    AsmToken Tok = Lexer.getTok();
240
241    switch (Tok.getKind()) {
242    default:
243      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
244      Error = true;
245      break;
246    case AsmToken::Error:
247      Error = true; // error already printed.
248      break;
249    case AsmToken::Identifier:
250      Out->os() << "identifier: " << Lexer.getTok().getString();
251      break;
252    case AsmToken::Integer:
253      Out->os() << "int: " << Lexer.getTok().getString();
254      break;
255    case AsmToken::Real:
256      Out->os() << "real: " << Lexer.getTok().getString();
257      break;
258    case AsmToken::Register:
259      Out->os() << "register: " << Lexer.getTok().getRegVal();
260      break;
261    case AsmToken::String:
262      Out->os() << "string: " << Lexer.getTok().getString();
263      break;
264
265    case AsmToken::Amp:            Out->os() << "Amp"; break;
266    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
267    case AsmToken::At:             Out->os() << "At"; break;
268    case AsmToken::Caret:          Out->os() << "Caret"; break;
269    case AsmToken::Colon:          Out->os() << "Colon"; break;
270    case AsmToken::Comma:          Out->os() << "Comma"; break;
271    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
272    case AsmToken::Dot:            Out->os() << "Dot"; break;
273    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
274    case AsmToken::Eof:            Out->os() << "Eof"; break;
275    case AsmToken::Equal:          Out->os() << "Equal"; break;
276    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
277    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
278    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
279    case AsmToken::Greater:        Out->os() << "Greater"; break;
280    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
281    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
282    case AsmToken::Hash:           Out->os() << "Hash"; break;
283    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
284    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
285    case AsmToken::LParen:         Out->os() << "LParen"; break;
286    case AsmToken::Less:           Out->os() << "Less"; break;
287    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
288    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
289    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
290    case AsmToken::Minus:          Out->os() << "Minus"; break;
291    case AsmToken::Percent:        Out->os() << "Percent"; break;
292    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
293    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
294    case AsmToken::Plus:           Out->os() << "Plus"; break;
295    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
296    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
297    case AsmToken::RParen:         Out->os() << "RParen"; break;
298    case AsmToken::Slash:          Out->os() << "Slash"; break;
299    case AsmToken::Star:           Out->os() << "Star"; break;
300    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
301    }
302
303    // Print the token string.
304    Out->os() << " (\"";
305    Out->os().write_escaped(Tok.getString());
306    Out->os() << "\")\n";
307  }
308
309  // Keep output if no errors.
310  if (Error == 0) Out->keep();
311
312  return Error;
313}
314
315static int AssembleInput(const char *ProgName) {
316  const Target *TheTarget = GetTarget(ProgName);
317  if (!TheTarget)
318    return 1;
319
320  OwningPtr<MemoryBuffer> BufferPtr;
321  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
322    errs() << ProgName << ": " << ec.message() << '\n';
323    return 1;
324  }
325  MemoryBuffer *Buffer = BufferPtr.take();
326
327  SourceMgr SrcMgr;
328
329  // Tell SrcMgr about this buffer, which is what the parser will pick up.
330  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
331
332  // Record the location of the include directories so that the lexer can find
333  // it later.
334  SrcMgr.setIncludeDirs(IncludeDirs);
335
336
337  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
338  assert(MAI && "Unable to create target asm info!");
339
340  llvm::OwningPtr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
341  assert(MRI && "Unable to create target register info!");
342
343  // Package up features to be passed to target/subtarget
344  std::string FeaturesStr;
345
346  // FIXME: This is not pretty. MCContext has a ptr to MCObjectFileInfo and
347  // MCObjectFileInfo needs a MCContext reference in order to initialize itself.
348  OwningPtr<MCObjectFileInfo> MOFI(new MCObjectFileInfo());
349  MCContext Ctx(*MAI, *MRI, MOFI.get());
350  MOFI->InitMCObjectFileInfo(TripleName, RelocModel, CMModel, Ctx);
351
352  if (SaveTempLabels)
353    Ctx.setAllowTemporaryLabels(false);
354
355  OwningPtr<tool_output_file> Out(GetOutputStream());
356  if (!Out)
357    return 1;
358
359  formatted_raw_ostream FOS(Out->os());
360  OwningPtr<MCStreamer> Str;
361
362  OwningPtr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo());
363  OwningPtr<MCSubtargetInfo>
364    STI(TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
365
366  // FIXME: There is a bit of code duplication with addPassesToEmitFile.
367  if (FileType == OFT_AssemblyFile) {
368    MCInstPrinter *IP =
369      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
370    MCCodeEmitter *CE = 0;
371    MCAsmBackend *MAB = 0;
372    if (ShowEncoding) {
373      CE = TheTarget->createMCCodeEmitter(*MCII, *STI, Ctx);
374      MAB = TheTarget->createMCAsmBackend(TripleName);
375    }
376    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
377                                           /*useLoc*/ true,
378                                           /*useCFI*/ true, IP, CE, MAB,
379                                           ShowInst));
380  } else if (FileType == OFT_Null) {
381    Str.reset(createNullStreamer(Ctx));
382  } else {
383    assert(FileType == OFT_ObjectFile && "Invalid file type!");
384    MCCodeEmitter *CE = TheTarget->createMCCodeEmitter(*MCII, *STI, Ctx);
385    MCAsmBackend *MAB = TheTarget->createMCAsmBackend(TripleName);
386    Str.reset(TheTarget->createMCObjectStreamer(TripleName, Ctx, *MAB,
387                                                FOS, CE, RelaxAll,
388                                                NoExecStack));
389  }
390
391  if (EnableLogging) {
392    Str.reset(createLoggingStreamer(Str.take(), errs()));
393  }
394
395  OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
396                                                   *Str.get(), *MAI));
397  OwningPtr<MCTargetAsmParser> TAP(TheTarget->createMCAsmParser(*STI, *Parser));
398  if (!TAP) {
399    errs() << ProgName
400           << ": error: this target does not support assembly parsing.\n";
401    return 1;
402  }
403
404  Parser->setShowParsedOperands(ShowInstOperands);
405  Parser->setTargetParser(*TAP.get());
406
407  int Res = Parser->Run(NoInitialTextSection);
408
409  // Keep output if no errors.
410  if (Res == 0) Out->keep();
411
412  return Res;
413}
414
415static int DisassembleInput(const char *ProgName, bool Enhanced) {
416  const Target *TheTarget = GetTarget(ProgName);
417  if (!TheTarget)
418    return 0;
419
420  OwningPtr<MemoryBuffer> Buffer;
421  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, Buffer)) {
422    errs() << ProgName << ": " << ec.message() << '\n';
423    return 1;
424  }
425
426  OwningPtr<tool_output_file> Out(GetOutputStream());
427  if (!Out)
428    return 1;
429
430  int Res;
431  if (Enhanced) {
432    Res =
433      Disassembler::disassembleEnhanced(TripleName, *Buffer.take(), Out->os());
434  } else {
435    Res = Disassembler::disassemble(*TheTarget, TripleName,
436                                    *Buffer.take(), Out->os());
437  }
438
439  // Keep output if no errors.
440  if (Res == 0) Out->keep();
441
442  return Res;
443}
444
445
446int main(int argc, char **argv) {
447  // Print a stack trace if we signal out.
448  sys::PrintStackTraceOnErrorSignal();
449  PrettyStackTraceProgram X(argc, argv);
450  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
451
452  // Initialize targets and assembly printers/parsers.
453  llvm::InitializeAllTargetInfos();
454  llvm::InitializeAllTargetMCs();
455  llvm::InitializeAllAsmParsers();
456  llvm::InitializeAllDisassemblers();
457
458  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
459  TripleName = Triple::normalize(TripleName);
460
461  switch (Action) {
462  default:
463  case AC_AsLex:
464    return AsLexInput(argv[0]);
465  case AC_Assemble:
466    return AssembleInput(argv[0]);
467  case AC_Disassemble:
468    return DisassembleInput(argv[0], false);
469  case AC_EDisassemble:
470    return DisassembleInput(argv[0], true);
471  }
472
473  return 0;
474}
475
476