llvm-mc.cpp revision ab8be96fd30ca9396e6b84fdddf1ac6208984cad 
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/SubtargetFeature.h"
23#include "llvm/Target/TargetAsmBackend.h"
24#include "llvm/Target/TargetAsmParser.h"
25#include "llvm/Target/TargetData.h"
26#include "llvm/Target/TargetRegistry.h"
27#include "llvm/Target/TargetAsmInfo.h"  // FIXME.
28#include "llvm/Target/TargetLowering.h"  // FIXME.
29#include "llvm/Target/TargetLoweringObjectFile.h"  // FIXME.
30#include "llvm/Target/TargetMachine.h"  // FIXME.
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<bool>
113NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
114                                   "in the text section"));
115
116static cl::opt<bool>
117SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
118
119enum ActionType {
120  AC_AsLex,
121  AC_Assemble,
122  AC_Disassemble,
123  AC_EDisassemble
124};
125
126static cl::opt<ActionType>
127Action(cl::desc("Action to perform:"),
128       cl::init(AC_Assemble),
129       cl::values(clEnumValN(AC_AsLex, "as-lex",
130                             "Lex tokens from a .s file"),
131                  clEnumValN(AC_Assemble, "assemble",
132                             "Assemble a .s file (default)"),
133                  clEnumValN(AC_Disassemble, "disassemble",
134                             "Disassemble strings of hex bytes"),
135                  clEnumValN(AC_EDisassemble, "edis",
136                             "Enhanced disassembly of strings of hex bytes"),
137                  clEnumValEnd));
138
139static const Target *GetTarget(const char *ProgName) {
140  // Figure out the target triple.
141  if (TripleName.empty())
142    TripleName = sys::getHostTriple();
143  if (!ArchName.empty()) {
144    llvm::Triple TT(TripleName);
145    TT.setArchName(ArchName);
146    TripleName = TT.str();
147  }
148
149  // Get the target specific parser.
150  std::string Error;
151  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
152  if (TheTarget)
153    return TheTarget;
154
155  errs() << ProgName << ": error: unable to get target for '" << TripleName
156         << "', see --version and --triple.\n";
157  return 0;
158}
159
160static tool_output_file *GetOutputStream() {
161  if (OutputFilename == "")
162    OutputFilename = "-";
163
164  std::string Err;
165  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
166                                               raw_fd_ostream::F_Binary);
167  if (!Err.empty()) {
168    errs() << Err << '\n';
169    delete Out;
170    return 0;
171  }
172
173  return Out;
174}
175
176static int AsLexInput(const char *ProgName) {
177  OwningPtr<MemoryBuffer> BufferPtr;
178  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
179    errs() << ProgName << ": " << ec.message() << '\n';
180    return 1;
181  }
182  MemoryBuffer *Buffer = BufferPtr.take();
183
184  SourceMgr SrcMgr;
185
186  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
187  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
188
189  // Record the location of the include directories so that the lexer can find
190  // it later.
191  SrcMgr.setIncludeDirs(IncludeDirs);
192
193  const Target *TheTarget = GetTarget(ProgName);
194  if (!TheTarget)
195    return 1;
196
197  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
198  assert(MAI && "Unable to create target asm info!");
199
200  AsmLexer Lexer(*MAI);
201  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
202
203  OwningPtr<tool_output_file> Out(GetOutputStream());
204  if (!Out)
205    return 1;
206
207  bool Error = false;
208  while (Lexer.Lex().isNot(AsmToken::Eof)) {
209    AsmToken Tok = Lexer.getTok();
210
211    switch (Tok.getKind()) {
212    default:
213      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
214      Error = true;
215      break;
216    case AsmToken::Error:
217      Error = true; // error already printed.
218      break;
219    case AsmToken::Identifier:
220      Out->os() << "identifier: " << Lexer.getTok().getString();
221      break;
222    case AsmToken::Integer:
223      Out->os() << "int: " << Lexer.getTok().getString();
224      break;
225    case AsmToken::Real:
226      Out->os() << "real: " << Lexer.getTok().getString();
227      break;
228    case AsmToken::Register:
229      Out->os() << "register: " << Lexer.getTok().getRegVal();
230      break;
231    case AsmToken::String:
232      Out->os() << "string: " << Lexer.getTok().getString();
233      break;
234
235    case AsmToken::Amp:            Out->os() << "Amp"; break;
236    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
237    case AsmToken::At:             Out->os() << "At"; break;
238    case AsmToken::Caret:          Out->os() << "Caret"; break;
239    case AsmToken::Colon:          Out->os() << "Colon"; break;
240    case AsmToken::Comma:          Out->os() << "Comma"; break;
241    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
242    case AsmToken::Dot:            Out->os() << "Dot"; break;
243    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
244    case AsmToken::Eof:            Out->os() << "Eof"; break;
245    case AsmToken::Equal:          Out->os() << "Equal"; break;
246    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
247    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
248    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
249    case AsmToken::Greater:        Out->os() << "Greater"; break;
250    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
251    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
252    case AsmToken::Hash:           Out->os() << "Hash"; break;
253    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
254    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
255    case AsmToken::LParen:         Out->os() << "LParen"; break;
256    case AsmToken::Less:           Out->os() << "Less"; break;
257    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
258    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
259    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
260    case AsmToken::Minus:          Out->os() << "Minus"; break;
261    case AsmToken::Percent:        Out->os() << "Percent"; break;
262    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
263    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
264    case AsmToken::Plus:           Out->os() << "Plus"; break;
265    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
266    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
267    case AsmToken::RParen:         Out->os() << "RParen"; break;
268    case AsmToken::Slash:          Out->os() << "Slash"; break;
269    case AsmToken::Star:           Out->os() << "Star"; break;
270    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
271    }
272
273    // Print the token string.
274    Out->os() << " (\"";
275    Out->os().write_escaped(Tok.getString());
276    Out->os() << "\")\n";
277  }
278
279  // Keep output if no errors.
280  if (Error == 0) Out->keep();
281
282  return Error;
283}
284
285static int AssembleInput(const char *ProgName) {
286  const Target *TheTarget = GetTarget(ProgName);
287  if (!TheTarget)
288    return 1;
289
290  OwningPtr<MemoryBuffer> BufferPtr;
291  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
292    errs() << ProgName << ": " << ec.message() << '\n';
293    return 1;
294  }
295  MemoryBuffer *Buffer = BufferPtr.take();
296
297  SourceMgr SrcMgr;
298
299  // Tell SrcMgr about this buffer, which is what the parser will pick up.
300  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
301
302  // Record the location of the include directories so that the lexer can find
303  // it later.
304  SrcMgr.setIncludeDirs(IncludeDirs);
305
306
307  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
308  assert(MAI && "Unable to create target asm info!");
309
310  // Package up features to be passed to target/subtarget
311  std::string FeaturesStr;
312  if (MCPU.size()) {
313    SubtargetFeatures Features;
314    Features.setCPU(MCPU);
315    FeaturesStr = Features.getString();
316  }
317
318  // FIXME: We shouldn't need to do this (and link in codegen).
319  //        When we split this out, we should do it in a way that makes
320  //        it straightforward to switch subtargets on the fly (.e.g,
321  //        the .cpu and .code16 directives).
322  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
323                                                             FeaturesStr));
324
325  if (!TM) {
326    errs() << ProgName << ": error: could not create target for triple '"
327           << TripleName << "'.\n";
328    return 1;
329  }
330
331  const TargetAsmInfo *tai = new TargetAsmInfo(*TM);
332  MCContext Ctx(*MAI, tai);
333  if (SaveTempLabels)
334    Ctx.setAllowTemporaryLabels(false);
335
336  OwningPtr<tool_output_file> Out(GetOutputStream());
337  if (!Out)
338    return 1;
339
340  formatted_raw_ostream FOS(Out->os());
341  OwningPtr<MCStreamer> Str;
342
343  const TargetLoweringObjectFile &TLOF =
344    TM->getTargetLowering()->getObjFileLowering();
345  const_cast<TargetLoweringObjectFile&>(TLOF).Initialize(Ctx, *TM);
346
347  // FIXME: There is a bit of code duplication with addPassesToEmitFile.
348  if (FileType == OFT_AssemblyFile) {
349    MCInstPrinter *IP =
350      TheTarget->createMCInstPrinter(*TM, 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(*Parser, *TM));
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
396static 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    if (MCPU.size()) {
419      SubtargetFeatures Features;
420      Features.setCPU(MCPU);
421      FeaturesStr = Features.getString();
422    }
423
424    // FIXME: We shouldn't need to do this (and link in codegen).
425    //        When we split this out, we should do it in a way that makes
426    //        it straightforward to switch subtargets on the fly (.e.g,
427    //        the .cpu and .code16 directives).
428    OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
429                                                               FeaturesStr));
430
431    if (!TM) {
432      errs() << ProgName << ": error: could not create target for triple '"
433             << TripleName << "'.\n";
434      return 1;
435    }
436
437    Res = Disassembler::disassemble(*TheTarget, *TM, 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::InitializeAllAsmPrinters();
459  llvm::InitializeAllAsmParsers();
460  llvm::InitializeAllDisassemblers();
461
462  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
463  TripleName = Triple::normalize(TripleName);
464
465  switch (Action) {
466  default:
467  case AC_AsLex:
468    return AsLexInput(argv[0]);
469  case AC_Assemble:
470    return AssembleInput(argv[0]);
471  case AC_Disassemble:
472    return DisassembleInput(argv[0], false);
473  case AC_EDisassemble:
474    return DisassembleInput(argv[0], true);
475  }
476
477  return 0;
478}
479
480