llvm-mc.cpp revision 745dacc91d7ee9531bfba76b21beb5d4eef93a7d
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/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>
72EnableLogging("enable-api-logging", cl::desc("Enable MC API logging"));
73
74enum OutputFileType {
75  OFT_Null,
76  OFT_AssemblyFile,
77  OFT_ObjectFile
78};
79static cl::opt<OutputFileType>
80FileType("filetype", cl::init(OFT_AssemblyFile),
81  cl::desc("Choose an output file type:"),
82  cl::values(
83       clEnumValN(OFT_AssemblyFile, "asm",
84                  "Emit an assembly ('.s') file"),
85       clEnumValN(OFT_Null, "null",
86                  "Don't emit anything (for timing purposes)"),
87       clEnumValN(OFT_ObjectFile, "obj",
88                  "Emit a native object ('.o') file"),
89       clEnumValEnd));
90
91static cl::list<std::string>
92IncludeDirs("I", cl::desc("Directory of include files"),
93            cl::value_desc("directory"), cl::Prefix);
94
95static cl::opt<std::string>
96ArchName("arch", cl::desc("Target arch to assemble for, "
97                            "see -version for available targets"));
98
99static cl::opt<std::string>
100TripleName("triple", cl::desc("Target triple to assemble for, "
101                              "see -version for available targets"));
102
103static cl::opt<std::string>
104MCPU("mcpu",
105     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
106     cl::value_desc("cpu-name"),
107     cl::init(""));
108
109static cl::opt<bool>
110NoInitialTextSection("n", cl::desc(
111                   "Don't assume assembly file starts in the text section"));
112
113enum ActionType {
114  AC_AsLex,
115  AC_Assemble,
116  AC_Disassemble,
117  AC_EDisassemble
118};
119
120static cl::opt<ActionType>
121Action(cl::desc("Action to perform:"),
122       cl::init(AC_Assemble),
123       cl::values(clEnumValN(AC_AsLex, "as-lex",
124                             "Lex tokens from a .s file"),
125                  clEnumValN(AC_Assemble, "assemble",
126                             "Assemble a .s file (default)"),
127                  clEnumValN(AC_Disassemble, "disassemble",
128                             "Disassemble strings of hex bytes"),
129                  clEnumValN(AC_EDisassemble, "edis",
130                             "Enhanced disassembly of strings of hex bytes"),
131                  clEnumValEnd));
132
133static const Target *GetTarget(const char *ProgName) {
134  // Figure out the target triple.
135  if (TripleName.empty())
136    TripleName = sys::getHostTriple();
137  if (!ArchName.empty()) {
138    llvm::Triple TT(TripleName);
139    TT.setArchName(ArchName);
140    TripleName = TT.str();
141  }
142
143  // Get the target specific parser.
144  std::string Error;
145  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
146  if (TheTarget)
147    return TheTarget;
148
149  errs() << ProgName << ": error: unable to get target for '" << TripleName
150         << "', see --version and --triple.\n";
151  return 0;
152}
153
154static tool_output_file *GetOutputStream() {
155  if (OutputFilename == "")
156    OutputFilename = "-";
157
158  std::string Err;
159  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
160                                               raw_fd_ostream::F_Binary);
161  if (!Err.empty()) {
162    errs() << Err << '\n';
163    delete Out;
164    return 0;
165  }
166
167  return Out;
168}
169
170static int AsLexInput(const char *ProgName) {
171  error_code ec;
172  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
173  if (Buffer == 0) {
174    errs() << ProgName << ": " << ec.message() << '\n';
175    return 1;
176  }
177
178  SourceMgr SrcMgr;
179
180  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
181  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
182
183  // Record the location of the include directories so that the lexer can find
184  // it later.
185  SrcMgr.setIncludeDirs(IncludeDirs);
186
187  const Target *TheTarget = GetTarget(ProgName);
188  if (!TheTarget)
189    return 1;
190
191  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
192  assert(MAI && "Unable to create target asm info!");
193
194  AsmLexer Lexer(*MAI);
195  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
196
197  OwningPtr<tool_output_file> Out(GetOutputStream());
198  if (!Out)
199    return 1;
200
201  bool Error = false;
202  while (Lexer.Lex().isNot(AsmToken::Eof)) {
203    AsmToken Tok = Lexer.getTok();
204
205    switch (Tok.getKind()) {
206    default:
207      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
208      Error = true;
209      break;
210    case AsmToken::Error:
211      Error = true; // error already printed.
212      break;
213    case AsmToken::Identifier:
214      Out->os() << "identifier: " << Lexer.getTok().getString();
215      break;
216    case AsmToken::Integer:
217      Out->os() << "int: " << Lexer.getTok().getString();
218      break;
219    case AsmToken::Real:
220      Out->os() << "real: " << Lexer.getTok().getString();
221      break;
222    case AsmToken::Register:
223      Out->os() << "register: " << Lexer.getTok().getRegVal();
224      break;
225    case AsmToken::String:
226      Out->os() << "string: " << Lexer.getTok().getString();
227      break;
228
229    case AsmToken::Amp:            Out->os() << "Amp"; break;
230    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
231    case AsmToken::At:             Out->os() << "At"; break;
232    case AsmToken::Caret:          Out->os() << "Caret"; break;
233    case AsmToken::Colon:          Out->os() << "Colon"; break;
234    case AsmToken::Comma:          Out->os() << "Comma"; break;
235    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
236    case AsmToken::Dot:            Out->os() << "Dot"; break;
237    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
238    case AsmToken::Eof:            Out->os() << "Eof"; break;
239    case AsmToken::Equal:          Out->os() << "Equal"; break;
240    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
241    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
242    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
243    case AsmToken::Greater:        Out->os() << "Greater"; break;
244    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
245    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
246    case AsmToken::Hash:           Out->os() << "Hash"; break;
247    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
248    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
249    case AsmToken::LParen:         Out->os() << "LParen"; break;
250    case AsmToken::Less:           Out->os() << "Less"; break;
251    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
252    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
253    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
254    case AsmToken::Minus:          Out->os() << "Minus"; break;
255    case AsmToken::Percent:        Out->os() << "Percent"; break;
256    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
257    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
258    case AsmToken::Plus:           Out->os() << "Plus"; break;
259    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
260    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
261    case AsmToken::RParen:         Out->os() << "RParen"; break;
262    case AsmToken::Slash:          Out->os() << "Slash"; break;
263    case AsmToken::Star:           Out->os() << "Star"; break;
264    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
265    }
266
267    // Print the token string.
268    Out->os() << " (\"";
269    Out->os().write_escaped(Tok.getString());
270    Out->os() << "\")\n";
271  }
272
273  // Keep output if no errors.
274  if (Error == 0) Out->keep();
275
276  return Error;
277}
278
279static int AssembleInput(const char *ProgName) {
280  const Target *TheTarget = GetTarget(ProgName);
281  if (!TheTarget)
282    return 1;
283
284  error_code ec;
285  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
286  if (Buffer == 0) {
287    errs() << ProgName << ": " << ec.message() << '\n';
288    return 1;
289  }
290
291  SourceMgr SrcMgr;
292
293  // Tell SrcMgr about this buffer, which is what the parser will pick up.
294  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
295
296  // Record the location of the include directories so that the lexer can find
297  // it later.
298  SrcMgr.setIncludeDirs(IncludeDirs);
299
300
301  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createAsmInfo(TripleName));
302  assert(MAI && "Unable to create target asm info!");
303
304  // Package up features to be passed to target/subtarget
305  std::string FeaturesStr;
306  if (MCPU.size()) {
307    SubtargetFeatures Features;
308    Features.setCPU(MCPU);
309    FeaturesStr = Features.getString();
310  }
311
312  // FIXME: We shouldn't need to do this (and link in codegen).
313  //        When we split this out, we should do it in a way that makes
314  //        it straightforward to switch subtargets on the fly (.e.g,
315  //        the .cpu and .code16 directives).
316  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName,
317                                                             FeaturesStr));
318
319  if (!TM) {
320    errs() << ProgName << ": error: could not create target for triple '"
321           << TripleName << "'.\n";
322    return 1;
323  }
324
325  const TargetAsmInfo *tai = new TargetAsmInfo(*TM);
326  MCContext Ctx(*MAI, tai);
327
328  OwningPtr<tool_output_file> Out(GetOutputStream());
329  if (!Out)
330    return 1;
331
332  formatted_raw_ostream FOS(Out->os());
333  OwningPtr<MCStreamer> Str;
334
335  const TargetLoweringObjectFile &TLOF =
336    TM->getTargetLowering()->getObjFileLowering();
337  const_cast<TargetLoweringObjectFile&>(TLOF).Initialize(Ctx, *TM);
338
339  if (FileType == OFT_AssemblyFile) {
340    MCInstPrinter *IP =
341      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI);
342    MCCodeEmitter *CE = 0;
343    TargetAsmBackend *TAB = 0;
344    if (ShowEncoding) {
345      CE = TheTarget->createCodeEmitter(*TM, Ctx);
346      TAB = TheTarget->createAsmBackend(TripleName);
347    }
348    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
349                                           /*useLoc*/ true, IP, CE, TAB,
350                                           ShowInst));
351  } else if (FileType == OFT_Null) {
352    Str.reset(createNullStreamer(Ctx));
353  } else {
354    assert(FileType == OFT_ObjectFile && "Invalid file type!");
355    MCCodeEmitter *CE = TheTarget->createCodeEmitter(*TM, Ctx);
356    TargetAsmBackend *TAB = TheTarget->createAsmBackend(TripleName);
357    Str.reset(TheTarget->createObjectStreamer(TripleName, Ctx, *TAB,
358                                              FOS, CE, RelaxAll));
359  }
360
361  if (EnableLogging) {
362    Str.reset(createLoggingStreamer(Str.take(), errs()));
363  }
364
365  OwningPtr<MCAsmParser> Parser(createMCAsmParser(*TheTarget, SrcMgr, Ctx,
366                                                   *Str.get(), *MAI));
367  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(*Parser, *TM));
368  if (!TAP) {
369    errs() << ProgName
370           << ": error: this target does not support assembly parsing.\n";
371    return 1;
372  }
373
374  Parser->setShowParsedOperands(ShowInstOperands);
375  Parser->setTargetParser(*TAP.get());
376
377  int Res = Parser->Run(NoInitialTextSection);
378
379  // Keep output if no errors.
380  if (Res == 0) Out->keep();
381
382  return Res;
383}
384
385static int DisassembleInput(const char *ProgName, bool Enhanced) {
386  const Target *TheTarget = GetTarget(ProgName);
387  if (!TheTarget)
388    return 0;
389
390  error_code ec;
391  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, ec);
392  if (Buffer == 0) {
393    errs() << ProgName << ": " << ec.message() << '\n';
394    return 1;
395  }
396
397  OwningPtr<tool_output_file> Out(GetOutputStream());
398  if (!Out)
399    return 1;
400
401  int Res;
402  if (Enhanced)
403    Res = Disassembler::disassembleEnhanced(TripleName, *Buffer, Out->os());
404  else
405    Res = Disassembler::disassemble(*TheTarget, TripleName, *Buffer, Out->os());
406
407  // Keep output if no errors.
408  if (Res == 0) Out->keep();
409
410  return Res;
411}
412
413
414int main(int argc, char **argv) {
415  // Print a stack trace if we signal out.
416  sys::PrintStackTraceOnErrorSignal();
417  PrettyStackTraceProgram X(argc, argv);
418  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
419
420  // Initialize targets and assembly printers/parsers.
421  llvm::InitializeAllTargetInfos();
422  // FIXME: We shouldn't need to initialize the Target(Machine)s.
423  llvm::InitializeAllTargets();
424  llvm::InitializeAllAsmPrinters();
425  llvm::InitializeAllAsmParsers();
426  llvm::InitializeAllDisassemblers();
427
428  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
429  TripleName = Triple::normalize(TripleName);
430
431  switch (Action) {
432  default:
433  case AC_AsLex:
434    return AsLexInput(argv[0]);
435  case AC_Assemble:
436    return AssembleInput(argv[0]);
437  case AC_Disassemble:
438    return DisassembleInput(argv[0], false);
439  case AC_EDisassemble:
440    return DisassembleInput(argv[0], true);
441  }
442
443  return 0;
444}
445
446