llvm-mc.cpp revision 439661395fd2a2a832dba01c65bc88718528313c 
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/MCSectionMachO.h"
22#include "llvm/MC/MCStreamer.h"
23#include "llvm/MC/MCSubtargetInfo.h"
24#include "llvm/MC/SubtargetFeature.h"
25#include "llvm/Target/TargetAsmBackend.h"
26#include "llvm/Target/TargetAsmParser.h"
27#include "llvm/Target/TargetData.h"
28#include "llvm/Target/TargetRegistry.h"
29#include "llvm/Target/TargetAsmInfo.h"  // FIXME.
30#include "llvm/Target/TargetLowering.h"  // FIXME.
31#include "llvm/Target/TargetLoweringObjectFile.h"  // FIXME.
32#include "llvm/Target/TargetMachine.h"  // FIXME.
33#include "llvm/Target/TargetSelect.h"
34#include "llvm/ADT/OwningPtr.h"
35#include "llvm/Support/CommandLine.h"
36#include "llvm/Support/FileUtilities.h"
37#include "llvm/Support/FormattedStream.h"
38#include "llvm/Support/ManagedStatic.h"
39#include "llvm/Support/MemoryBuffer.h"
40#include "llvm/Support/PrettyStackTrace.h"
41#include "llvm/Support/SourceMgr.h"
42#include "llvm/Support/ToolOutputFile.h"
43#include "llvm/Support/Host.h"
44#include "llvm/Support/Signals.h"
45#include "llvm/Support/system_error.h"
46#include "Disassembler.h"
47using namespace llvm;
48
49static cl::opt<std::string>
50InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
51
52static cl::opt<std::string>
53OutputFilename("o", cl::desc("Output filename"),
54               cl::value_desc("filename"));
55
56static cl::opt<bool>
57ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
58
59static cl::opt<bool>
60ShowInst("show-inst", cl::desc("Show internal instruction representation"));
61
62static cl::opt<bool>
63ShowInstOperands("show-inst-operands",
64                 cl::desc("Show instructions operands as parsed"));
65
66static cl::opt<unsigned>
67OutputAsmVariant("output-asm-variant",
68                 cl::desc("Syntax variant to use for output printing"));
69
70static cl::opt<bool>
71RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
72
73static cl::opt<bool>
74NoExecStack("mc-no-exec-stack", cl::desc("File doesn't need an exec stack"));
75
76static cl::opt<bool>
77EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
78
79enum OutputFileType {
80  OFT_Null,
81  OFT_AssemblyFile,
82  OFT_ObjectFile
83};
84static cl::opt<OutputFileType>
85FileType("filetype", cl::init(OFT_AssemblyFile),
86  cl::desc("Choose an output file type:"),
87  cl::values(
88       clEnumValN(OFT_AssemblyFile, "asm",
89                  "Emit an assembly ('.s') file"),
90       clEnumValN(OFT_Null, "null",
91                  "Don't emit anything (for timing purposes)"),
92       clEnumValN(OFT_ObjectFile, "obj",
93                  "Emit a native object ('.o') file"),
94       clEnumValEnd));
95
96static cl::list<std::string>
97IncludeDirs("I", cl::desc("Directory of include files"),
98            cl::value_desc("directory"), cl::Prefix);
99
100static cl::opt<std::string>
101ArchName("arch", cl::desc("Target arch to assemble for, "
102                          "see -version for available targets"));
103
104static cl::opt<std::string>
105TripleName("triple", cl::desc("Target triple to assemble for, "
106                              "see -version for available targets"));
107
108static cl::opt<std::string>
109MCPU("mcpu",
110     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
111     cl::value_desc("cpu-name"),
112     cl::init(""));
113
114static cl::opt<Reloc::Model>
115RelocModel("relocation-model",
116             cl::desc("Choose relocation model"),
117             cl::init(Reloc::Default),
118             cl::values(
119            clEnumValN(Reloc::Default, "default",
120                       "Target default relocation model"),
121            clEnumValN(Reloc::Static, "static",
122                       "Non-relocatable code"),
123            clEnumValN(Reloc::PIC_, "pic",
124                       "Fully relocatable, position independent code"),
125            clEnumValN(Reloc::DynamicNoPIC, "dynamic-no-pic",
126                       "Relocatable external references, non-relocatable code"),
127            clEnumValEnd));
128
129static cl::opt<bool>
130NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
131                                   "in the text section"));
132
133static cl::opt<bool>
134SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
135
136enum ActionType {
137  AC_AsLex,
138  AC_Assemble,
139  AC_Disassemble,
140  AC_EDisassemble
141};
142
143static cl::opt<ActionType>
144Action(cl::desc("Action to perform:"),
145       cl::init(AC_Assemble),
146       cl::values(clEnumValN(AC_AsLex, "as-lex",
147                             "Lex tokens from a .s file"),
148                  clEnumValN(AC_Assemble, "assemble",
149                             "Assemble a .s file (default)"),
150                  clEnumValN(AC_Disassemble, "disassemble",
151                             "Disassemble strings of hex bytes"),
152                  clEnumValN(AC_EDisassemble, "edis",
153                             "Enhanced disassembly of strings of hex bytes"),
154                  clEnumValEnd));
155
156static const Target *GetTarget(const char *ProgName) {
157  // Figure out the target triple.
158  if (TripleName.empty())
159    TripleName = sys::getHostTriple();
160  if (!ArchName.empty()) {
161    llvm::Triple TT(TripleName);
162    TT.setArchName(ArchName);
163    TripleName = TT.str();
164  }
165
166  // Get the target specific parser.
167  std::string Error;
168  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
169  if (TheTarget)
170    return TheTarget;
171
172  errs() << ProgName << ": error: unable to get target for '" << TripleName
173         << "', see --version and --triple.\n";
174  return 0;
175}
176
177static tool_output_file *GetOutputStream() {
178  if (OutputFilename == "")
179    OutputFilename = "-";
180
181  std::string Err;
182  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
183                                               raw_fd_ostream::F_Binary);
184  if (!Err.empty()) {
185    errs() << Err << '\n';
186    delete Out;
187    return 0;
188  }
189
190  return Out;
191}
192
193static int AsLexInput(const char *ProgName) {
194  OwningPtr<MemoryBuffer> BufferPtr;
195  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
196    errs() << ProgName << ": " << ec.message() << '\n';
197    return 1;
198  }
199  MemoryBuffer *Buffer = BufferPtr.take();
200
201  SourceMgr SrcMgr;
202
203  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
204  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
205
206  // Record the location of the include directories so that the lexer can find
207  // it later.
208  SrcMgr.setIncludeDirs(IncludeDirs);
209
210  const Target *TheTarget = GetTarget(ProgName);
211  if (!TheTarget)
212    return 1;
213
214  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
215  assert(MAI && "Unable to create target asm info!");
216
217  AsmLexer Lexer(*MAI);
218  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
219
220  OwningPtr<tool_output_file> Out(GetOutputStream());
221  if (!Out)
222    return 1;
223
224  bool Error = false;
225  while (Lexer.Lex().isNot(AsmToken::Eof)) {
226    AsmToken Tok = Lexer.getTok();
227
228    switch (Tok.getKind()) {
229    default:
230      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
231      Error = true;
232      break;
233    case AsmToken::Error:
234      Error = true; // error already printed.
235      break;
236    case AsmToken::Identifier:
237      Out->os() << "identifier: " << Lexer.getTok().getString();
238      break;
239    case AsmToken::Integer:
240      Out->os() << "int: " << Lexer.getTok().getString();
241      break;
242    case AsmToken::Real:
243      Out->os() << "real: " << Lexer.getTok().getString();
244      break;
245    case AsmToken::Register:
246      Out->os() << "register: " << Lexer.getTok().getRegVal();
247      break;
248    case AsmToken::String:
249      Out->os() << "string: " << Lexer.getTok().getString();
250      break;
251
252    case AsmToken::Amp:            Out->os() << "Amp"; break;
253    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
254    case AsmToken::At:             Out->os() << "At"; break;
255    case AsmToken::Caret:          Out->os() << "Caret"; break;
256    case AsmToken::Colon:          Out->os() << "Colon"; break;
257    case AsmToken::Comma:          Out->os() << "Comma"; break;
258    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
259    case AsmToken::Dot:            Out->os() << "Dot"; break;
260    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
261    case AsmToken::Eof:            Out->os() << "Eof"; break;
262    case AsmToken::Equal:          Out->os() << "Equal"; break;
263    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
264    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
265    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
266    case AsmToken::Greater:        Out->os() << "Greater"; break;
267    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
268    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
269    case AsmToken::Hash:           Out->os() << "Hash"; break;
270    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
271    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
272    case AsmToken::LParen:         Out->os() << "LParen"; break;
273    case AsmToken::Less:           Out->os() << "Less"; break;
274    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
275    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
276    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
277    case AsmToken::Minus:          Out->os() << "Minus"; break;
278    case AsmToken::Percent:        Out->os() << "Percent"; break;
279    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
280    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
281    case AsmToken::Plus:           Out->os() << "Plus"; break;
282    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
283    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
284    case AsmToken::RParen:         Out->os() << "RParen"; break;
285    case AsmToken::Slash:          Out->os() << "Slash"; break;
286    case AsmToken::Star:           Out->os() << "Star"; break;
287    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
288    }
289
290    // Print the token string.
291    Out->os() << " (\"";
292    Out->os().write_escaped(Tok.getString());
293    Out->os() << "\")\n";
294  }
295
296  // Keep output if no errors.
297  if (Error == 0) Out->keep();
298
299  return Error;
300}
301
302static int AssembleInput(const char *ProgName) {
303  const Target *TheTarget = GetTarget(ProgName);
304  if (!TheTarget)
305    return 1;
306
307  OwningPtr<MemoryBuffer> BufferPtr;
308  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
309    errs() << ProgName << ": " << ec.message() << '\n';
310    return 1;
311  }
312  MemoryBuffer *Buffer = BufferPtr.take();
313
314  SourceMgr SrcMgr;
315
316  // Tell SrcMgr about this buffer, which is what the parser will pick up.
317  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
318
319  // Record the location of the include directories so that the lexer can find
320  // it later.
321  SrcMgr.setIncludeDirs(IncludeDirs);
322
323
324  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
325  assert(MAI && "Unable to create target asm info!");
326
327  llvm::OwningPtr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
328  assert(MRI && "Unable to create target register info!");
329
330  // Package up features to be passed to target/subtarget
331  std::string FeaturesStr;
332
333  // FIXME: We shouldn't need to do this (and link in codegen).
334  //        When we split this out, we should do it in a way that makes
335  //        it straightforward to switch subtargets on the fly (.e.g,
336  //        the .cpu and .code16 directives).
337  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
338                                                             MCPU,
339                                                             FeaturesStr,
340                                                             RelocModel));
341
342  if (!TM) {
343    errs() << ProgName << ": error: could not create target for triple '"
344           << TripleName << "'.\n";
345    return 1;
346  }
347
348  const TargetAsmInfo *tai = new TargetAsmInfo(*TM);
349  MCContext Ctx(*MAI, *MRI, tai);
350  if (SaveTempLabels)
351    Ctx.setAllowTemporaryLabels(false);
352
353  OwningPtr<tool_output_file> Out(GetOutputStream());
354  if (!Out)
355    return 1;
356
357  formatted_raw_ostream FOS(Out->os());
358  OwningPtr<MCStreamer> Str;
359
360  const TargetLoweringObjectFile &TLOF =
361    TM->getTargetLowering()->getObjFileLowering();
362  const_cast<TargetLoweringObjectFile&>(TLOF).Initialize(Ctx, *TM);
363
364  OwningPtr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo());
365  OwningPtr<MCSubtargetInfo>
366    STI(TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
367
368  // FIXME: There is a bit of code duplication with addPassesToEmitFile.
369  if (FileType == OFT_AssemblyFile) {
370    MCInstPrinter *IP =
371      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
372    MCCodeEmitter *CE = 0;
373    TargetAsmBackend *TAB = 0;
374    if (ShowEncoding) {
375      CE = TheTarget->createCodeEmitter(*MCII, *STI, Ctx);
376      TAB = TheTarget->createAsmBackend(TripleName);
377    }
378    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
379                                           /*useLoc*/ true,
380                                           /*useCFI*/ true, IP, CE, TAB,
381                                           ShowInst));
382  } else if (FileType == OFT_Null) {
383    Str.reset(createNullStreamer(Ctx));
384  } else {
385    assert(FileType == OFT_ObjectFile && "Invalid file type!");
386    MCCodeEmitter *CE = TheTarget->createCodeEmitter(*MCII, *STI, Ctx);
387    TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName);
388    Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB,
389                                              FOS, CE, RelaxAll,
390                                              NoExecStack));
391  }
392
393  if (EnableLogging) {
394    Str.reset(createLoggingStreamer(Str.take(), errs()));
395  }
396
397  OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
398                                                   *Str.get(), *MAI));
399  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*STI, *Parser));
400  if (!TAP) {
401    errs() << ProgName
402           << ": error: this target does not support assembly parsing.\n";
403    return 1;
404  }
405
406  Parser->setShowParsedOperands(ShowInstOperands);
407  Parser->setTargetParser(*TAP.get());
408
409  int Res = Parser->Run(NoInitialTextSection);
410
411  // Keep output if no errors.
412  if (Res == 0) Out->keep();
413
414  return Res;
415}
416
417static int DisassembleInput(const char *ProgName, bool Enhanced) {
418  const Target *TheTarget = GetTarget(ProgName);
419  if (!TheTarget)
420    return 0;
421
422  OwningPtr<MemoryBuffer> Buffer;
423  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, Buffer)) {
424    errs() << ProgName << ": " << ec.message() << '\n';
425    return 1;
426  }
427
428  OwningPtr<tool_output_file> Out(GetOutputStream());
429  if (!Out)
430    return 1;
431
432  int Res;
433  if (Enhanced) {
434    Res =
435      Disassembler::disassembleEnhanced(TripleName, *Buffer.take(), Out->os());
436  } else {
437    Res = Disassembler::disassemble(*TheTarget, TripleName,
438                                    *Buffer.take(), Out->os());
439  }
440
441  // Keep output if no errors.
442  if (Res == 0) Out->keep();
443
444  return Res;
445}
446
447
448int main(int argc, char **argv) {
449  // Print a stack trace if we signal out.
450  sys::PrintStackTraceOnErrorSignal();
451  PrettyStackTraceProgram X(argc, argv);
452  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
453
454  // Initialize targets and assembly printers/parsers.
455  llvm::InitializeAllTargetInfos();
456  // FIXME: We shouldn't need to initialize the Target(Machine)s.
457  llvm::InitializeAllTargets();
458  llvm::InitializeAllMCAsmInfos();
459  llvm::InitializeAllMCCodeGenInfos();
460  llvm::InitializeAllMCInstrInfos();
461  llvm::InitializeAllMCRegisterInfos();
462  llvm::InitializeAllMCSubtargetInfos();
463  llvm::InitializeAllAsmPrinters();
464  llvm::InitializeAllAsmParsers();
465  llvm::InitializeAllDisassemblers();
466
467  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
468  TripleName = Triple::normalize(TripleName);
469
470  switch (Action) {
471  default:
472  case AC_AsLex:
473    return AsLexInput(argv[0]);
474  case AC_Assemble:
475    return AssembleInput(argv[0]);
476  case AC_Disassemble:
477    return DisassembleInput(argv[0], false);
478  case AC_EDisassemble:
479    return DisassembleInput(argv[0], true);
480  }
481
482  return 0;
483}
484
485