llvm-mc.cpp revision 333fb04506233255f10d8095c9e2de5e5f0fdc6f 
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/SubtargetFeature.h" // FIXME.
27#include "llvm/Target/TargetMachine.h"  // FIXME.
28#include "llvm/Target/TargetSelect.h"
29#include "llvm/ADT/OwningPtr.h"
30#include "llvm/Support/CommandLine.h"
31#include "llvm/Support/FileUtilities.h"
32#include "llvm/Support/FormattedStream.h"
33#include "llvm/Support/ManagedStatic.h"
34#include "llvm/Support/MemoryBuffer.h"
35#include "llvm/Support/PrettyStackTrace.h"
36#include "llvm/Support/SourceMgr.h"
37#include "llvm/Support/ToolOutputFile.h"
38#include "llvm/Support/Host.h"
39#include "llvm/Support/Signals.h"
40#include "llvm/Support/system_error.h"
41#include "Disassembler.h"
42using namespace llvm;
43
44static cl::opt<std::string>
45InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
46
47static cl::opt<std::string>
48OutputFilename("o", cl::desc("Output filename"),
49               cl::value_desc("filename"));
50
51static cl::opt<bool>
52ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
53
54static cl::opt<bool>
55ShowInst("show-inst", cl::desc("Show internal instruction representation"));
56
57static cl::opt<bool>
58ShowInstOperands("show-inst-operands",
59                 cl::desc("Show instructions operands as parsed"));
60
61static cl::opt<unsigned>
62OutputAsmVariant("output-asm-variant",
63                 cl::desc("Syntax variant to use for output printing"));
64
65static cl::opt<bool>
66RelaxAll("mc-relax-all", cl::desc("Relax all fixups"));
67
68static cl::opt<bool>
69EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
70
71enum OutputFileType {
72  OFT_Null,
73  OFT_AssemblyFile,
74  OFT_ObjectFile
75};
76static cl::opt<OutputFileType>
77FileType("filetype", cl::init(OFT_AssemblyFile),
78  cl::desc("Choose an output file type:"),
79  cl::values(
80       clEnumValN(OFT_AssemblyFile, "asm",
81                  "Emit an assembly ('.s') file"),
82       clEnumValN(OFT_Null, "null",
83                  "Don't emit anything (for timing purposes)"),
84       clEnumValN(OFT_ObjectFile, "obj",
85                  "Emit a native object ('.o') file"),
86       clEnumValEnd));
87
88static cl::list<std::string>
89IncludeDirs("I", cl::desc("Directory of include files"),
90            cl::value_desc("directory"), cl::Prefix);
91
92static cl::opt<std::string>
93ArchName("arch", cl::desc("Target arch to assemble for, "
94                            "see -version for available targets"));
95
96static cl::opt<std::string>
97TripleName("triple", cl::desc("Target triple to assemble for, "
98                              "see -version for available targets"));
99
100static cl::opt<std::string>
101MCPU("mcpu",
102     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
103     cl::value_desc("cpu-name"),
104     cl::init(""));
105
106static cl::opt<bool>
107NoInitialTextSection("n", cl::desc(
108                   "Don't assume assembly file starts in the text section"));
109
110enum ActionType {
111  AC_AsLex,
112  AC_Assemble,
113  AC_Disassemble,
114  AC_EDisassemble
115};
116
117static cl::opt<ActionType>
118Action(cl::desc("Action to perform:"),
119       cl::init(AC_Assemble),
120       cl::values(clEnumValN(AC_AsLex, "as-lex",
121                             "Lex tokens from a .s file"),
122                  clEnumValN(AC_Assemble, "assemble",
123                             "Assemble a .s file (default)"),
124                  clEnumValN(AC_Disassemble, "disassemble",
125                             "Disassemble strings of hex bytes"),
126                  clEnumValN(AC_EDisassemble, "edis",
127                             "Enhanced disassembly of strings of hex bytes"),
128                  clEnumValEnd));
129
130static const Target *GetTarget(const char *ProgName) {
131  // Figure out the target triple.
132  if (TripleName.empty())
133    TripleName = sys::getHostTriple();
134  if (!ArchName.empty()) {
135    llvm::Triple TT(TripleName);
136    TT.setArchName(ArchName);
137    TripleName = TT.str();
138  }
139
140  // Get the target specific parser.
141  std::string Error;
142  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
143  if (TheTarget)
144    return TheTarget;
145
146  errs() << ProgName << ": error: unable to get target for '" << TripleName
147         << "', see --version and --triple.\n";
148  return 0;
149}
150
151static tool_output_file *GetOutputStream() {
152  if (OutputFilename == "")
153    OutputFilename = "-";
154
155  std::string Err;
156  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
157                                               raw_fd_ostream::F_Binary);
158  if (!Err.empty()) {
159    errs() << Err << '\n';
160    delete Out;
161    return 0;
162  }
163
164  return Out;
165}
166
167static int AsLexInput(const char *ProgName) {
168  error_code ec;
169  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
170  if (Buffer == 0) {
171    errs() << ProgName << ": " << ec.message() << '\n';
172    return 1;
173  }
174
175  SourceMgr SrcMgr;
176
177  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
178  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
179
180  // Record the location of the include directories so that the lexer can find
181  // it later.
182  SrcMgr.setIncludeDirs(IncludeDirs);
183
184  const Target *TheTarget = GetTarget(ProgName);
185  if (!TheTarget)
186    return 1;
187
188  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
189  assert(MAI && "Unable to create target asm info!");
190
191  AsmLexer Lexer(*MAI);
192  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
193
194  OwningPtr<tool_output_file> Out(GetOutputStream());
195  if (!Out)
196    return 1;
197
198  bool Error = false;
199  while (Lexer.Lex().isNot(AsmToken::Eof)) {
200    AsmToken Tok = Lexer.getTok();
201
202    switch (Tok.getKind()) {
203    default:
204      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
205      Error = true;
206      break;
207    case AsmToken::Error:
208      Error = true; // error already printed.
209      break;
210    case AsmToken::Identifier:
211      Out->os() << "identifier: " << Lexer.getTok().getString();
212      break;
213    case AsmToken::Integer:
214      Out->os() << "int: " << Lexer.getTok().getString();
215      break;
216    case AsmToken::Real:
217      Out->os() << "real: " << Lexer.getTok().getString();
218      break;
219    case AsmToken::Register:
220      Out->os() << "register: " << Lexer.getTok().getRegVal();
221      break;
222    case AsmToken::String:
223      Out->os() << "string: " << Lexer.getTok().getString();
224      break;
225
226    case AsmToken::Amp:            Out->os() << "Amp"; break;
227    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
228    case AsmToken::At:             Out->os() << "At"; break;
229    case AsmToken::Caret:          Out->os() << "Caret"; break;
230    case AsmToken::Colon:          Out->os() << "Colon"; break;
231    case AsmToken::Comma:          Out->os() << "Comma"; break;
232    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
233    case AsmToken::Dot:            Out->os() << "Dot"; break;
234    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
235    case AsmToken::Eof:            Out->os() << "Eof"; break;
236    case AsmToken::Equal:          Out->os() << "Equal"; break;
237    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
238    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
239    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
240    case AsmToken::Greater:        Out->os() << "Greater"; break;
241    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
242    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
243    case AsmToken::Hash:           Out->os() << "Hash"; break;
244    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
245    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
246    case AsmToken::LParen:         Out->os() << "LParen"; break;
247    case AsmToken::Less:           Out->os() << "Less"; break;
248    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
249    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
250    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
251    case AsmToken::Minus:          Out->os() << "Minus"; break;
252    case AsmToken::Percent:        Out->os() << "Percent"; break;
253    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
254    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
255    case AsmToken::Plus:           Out->os() << "Plus"; break;
256    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
257    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
258    case AsmToken::RParen:         Out->os() << "RParen"; break;
259    case AsmToken::Slash:          Out->os() << "Slash"; break;
260    case AsmToken::Star:           Out->os() << "Star"; break;
261    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
262    }
263
264    // Print the token string.
265    Out->os() << " (\"";
266    Out->os().write_escaped(Tok.getString());
267    Out->os() << "\")\n";
268  }
269
270  // Keep output if no errors.
271  if (Error == 0) Out->keep();
272
273  return Error;
274}
275
276static int AssembleInput(const char *ProgName) {
277  const Target *TheTarget = GetTarget(ProgName);
278  if (!TheTarget)
279    return 1;
280
281  error_code ec;
282  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
283  if (Buffer == 0) {
284    errs() << ProgName << ": " << ec.message() << '\n';
285    return 1;
286  }
287
288  SourceMgr SrcMgr;
289
290  // Tell SrcMgr about this buffer, which is what the parser will pick up.
291  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
292
293  // Record the location of the include directories so that the lexer can find
294  // it later.
295  SrcMgr.setIncludeDirs(IncludeDirs);
296
297
298  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
299  assert(MAI && "Unable to create target asm info!");
300
301  MCContext Ctx(*MAI);
302
303  // Package up features to be passed to target/subtarget
304  std::string FeaturesStr;
305  if (MCPU.size()) {
306    SubtargetFeatures Features;
307    Features.setCPU(MCPU);
308    FeaturesStr = Features.getString();
309  }
310
311  // FIXME: We shouldn't need to do this (and link in codegen).
312  //        When we split this out, we should do it in a way that makes
313  //        it straightforward to switch subtargets on the fly (.e.g,
314  //        the .cpu and .code16 directives).
315  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
316                                                             FeaturesStr));
317
318  if (!TM) {
319    errs() << ProgName << ": error: could not create target for triple '"
320           << TripleName << "'.\n";
321    return 1;
322  }
323
324  OwningPtr<tool_output_file> Out(GetOutputStream());
325  if (!Out)
326    return 1;
327
328  formatted_raw_ostream FOS(Out->os());
329  OwningPtr<MCStreamer> Str;
330
331  if (FileType == OFT_AssemblyFile) {
332    MCInstPrinter *IP =
333      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
334    MCCodeEmitter *CE = 0;
335    if (ShowEncoding)
336      CE = TheTarget->createCodeEmitter(*TM, Ctx);
337    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS,
338                                           TM->getTargetData()->isLittleEndian(),
339                                           /*asmverbose*/true, IP, CE, ShowInst));
340  } else if (FileType == OFT_Null) {
341    Str.reset(createNullStreamer(Ctx));
342  } else {
343    assert(FileType == OFT_ObjectFile && "Invalid file type!");
344    MCCodeEmitter *CE = TheTarget->createCodeEmitter(*TM, Ctx);
345    TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName);
346    Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB,
347                                              FOS, CE, RelaxAll));
348  }
349
350  if (EnableLogging) {
351    Str.reset(createLoggingStreamer(Str.take(), errs()));
352  }
353
354  OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
355                                                   *Str.get(), *MAI));
356  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*Parser, *TM));
357  if (!TAP) {
358    errs() << ProgName
359           << ": error: this target does not support assembly parsing.\n";
360    return 1;
361  }
362
363  Parser->setShowParsedOperands(ShowInstOperands);
364  Parser->setTargetParser(*TAP.get());
365
366  int Res = Parser->Run(NoInitialTextSection);
367
368  // Keep output if no errors.
369  if (Res == 0) Out->keep();
370
371  return Res;
372}
373
374static int DisassembleInput(const char *ProgName, bool Enhanced) {
375  const Target *TheTarget = GetTarget(ProgName);
376  if (!TheTarget)
377    return 0;
378
379  error_code ec;
380  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
381  if (Buffer == 0) {
382    errs() << ProgName << ": " << ec.message() << '\n';
383    return 1;
384  }
385
386  OwningPtr<tool_output_file> Out(GetOutputStream());
387  if (!Out)
388    return 1;
389
390  int Res;
391  if (Enhanced)
392    Res = Disassembler::disassembleEnhanced(TripleName, *Buffer, Out->os());
393  else
394    Res = Disassembler::disassemble(*TheTarget, TripleName, *Buffer, Out->os());
395
396  // Keep output if no errors.
397  if (Res == 0) Out->keep();
398
399  return Res;
400}
401
402
403int main(int argc, char **argv) {
404  // Print a stack trace if we signal out.
405  sys::PrintStackTraceOnErrorSignal();
406  PrettyStackTraceProgram X(argc, argv);
407  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
408
409  // Initialize targets and assembly printers/parsers.
410  llvm::InitializeAllTargetInfos();
411  // FIXME: We shouldn't need to initialize the Target(Machine)s.
412  llvm::InitializeAllTargets();
413  llvm::InitializeAllAsmPrinters();
414  llvm::InitializeAllAsmParsers();
415  llvm::InitializeAllDisassemblers();
416
417  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
418  TripleName = Triple::normalize(TripleName);
419
420  switch (Action) {
421  default:
422  case AC_AsLex:
423    return AsLexInput(argv[0]);
424  case AC_Assemble:
425    return AssembleInput(argv[0]);
426  case AC_Disassemble:
427    return DisassembleInput(argv[0], false);
428  case AC_EDisassemble:
429    return DisassembleInput(argv[0], true);
430  }
431
432  return 0;
433}
434
435