llvm-mc.cpp revision a3dcfb130044f306632a5fab43854eda4095a09c 
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/MCAsmLexer.h"
16#include "llvm/MC/MCContext.h"
17#include "llvm/MC/MCCodeEmitter.h"
18#include "llvm/MC/MCInstPrinter.h"
19#include "llvm/MC/MCSectionMachO.h"
20#include "llvm/MC/MCStreamer.h"
21#include "llvm/ADT/OwningPtr.h"
22#include "llvm/Support/CommandLine.h"
23#include "llvm/Support/FormattedStream.h"
24#include "llvm/Support/ManagedStatic.h"
25#include "llvm/Support/MemoryBuffer.h"
26#include "llvm/Support/PrettyStackTrace.h"
27#include "llvm/Support/SourceMgr.h"
28#include "llvm/Support/raw_ostream.h"
29#include "llvm/System/Signals.h"
30#include "llvm/Target/TargetAsmParser.h"
31#include "llvm/Target/TargetRegistry.h"
32#include "llvm/Target/TargetMachine.h"  // FIXME.
33#include "llvm/Target/TargetSelect.h"
34#include "AsmParser.h"
35#include "Disassembler.h"
36using namespace llvm;
37
38static cl::opt<std::string>
39InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
40
41static cl::opt<std::string>
42OutputFilename("o", cl::desc("Output filename"),
43               cl::value_desc("filename"));
44
45static cl::opt<bool>
46ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
47
48static cl::opt<unsigned>
49OutputAsmVariant("output-asm-variant",
50                 cl::desc("Syntax variant to use for output printing"));
51
52enum OutputFileType {
53  OFT_AssemblyFile,
54  OFT_ObjectFile
55};
56static cl::opt<OutputFileType>
57FileType("filetype", cl::init(OFT_AssemblyFile),
58  cl::desc("Choose an output file type:"),
59  cl::values(
60       clEnumValN(OFT_AssemblyFile, "asm",
61                  "Emit an assembly ('.s') file"),
62       clEnumValN(OFT_ObjectFile, "obj",
63                  "Emit a native object ('.o') file"),
64       clEnumValEnd));
65
66static cl::opt<bool>
67Force("f", cl::desc("Enable binary output on terminals"));
68
69static cl::list<std::string>
70IncludeDirs("I", cl::desc("Directory of include files"),
71            cl::value_desc("directory"), cl::Prefix);
72
73static cl::opt<std::string>
74TripleName("triple", cl::desc("Target triple to assemble for, "
75                              "see -version for available targets"),
76           cl::init(LLVM_HOSTTRIPLE));
77
78enum ActionType {
79  AC_AsLex,
80  AC_Assemble,
81  AC_Disassemble
82};
83
84static cl::opt<ActionType>
85Action(cl::desc("Action to perform:"),
86       cl::init(AC_Assemble),
87       cl::values(clEnumValN(AC_AsLex, "as-lex",
88                             "Lex tokens from a .s file"),
89                  clEnumValN(AC_Assemble, "assemble",
90                             "Assemble a .s file (default)"),
91                  clEnumValN(AC_Disassemble, "disassemble",
92                             "Disassemble strings of hex bytes"),
93                  clEnumValEnd));
94
95static const Target *GetTarget(const char *ProgName) {
96  // Get the target specific parser.
97  std::string Error;
98  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
99  if (TheTarget)
100    return TheTarget;
101
102  errs() << ProgName << ": error: unable to get target for '" << TripleName
103         << "', see --version and --triple.\n";
104  return 0;
105}
106
107static int AsLexInput(const char *ProgName) {
108  std::string ErrorMessage;
109  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
110                                                      &ErrorMessage);
111  if (Buffer == 0) {
112    errs() << ProgName << ": ";
113    if (ErrorMessage.size())
114      errs() << ErrorMessage << "\n";
115    else
116      errs() << "input file didn't read correctly.\n";
117    return 1;
118  }
119
120  SourceMgr SrcMgr;
121
122  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
123  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
124
125  // Record the location of the include directories so that the lexer can find
126  // it later.
127  SrcMgr.setIncludeDirs(IncludeDirs);
128
129  const Target *TheTarget = GetTarget(ProgName);
130  if (!TheTarget)
131    return 1;
132
133  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
134  assert(MAI && "Unable to create target asm info!");
135
136  AsmLexer Lexer(SrcMgr, *MAI);
137
138  bool Error = false;
139
140  while (Lexer.Lex().isNot(AsmToken::Eof)) {
141    switch (Lexer.getKind()) {
142    default:
143      Lexer.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
144      Error = true;
145      break;
146    case AsmToken::Error:
147      Error = true; // error already printed.
148      break;
149    case AsmToken::Identifier:
150      outs() << "identifier: " << Lexer.getTok().getString() << '\n';
151      break;
152    case AsmToken::String:
153      outs() << "string: " << Lexer.getTok().getString() << '\n';
154      break;
155    case AsmToken::Integer:
156      outs() << "int: " << Lexer.getTok().getString() << '\n';
157      break;
158
159    case AsmToken::Amp:            outs() << "Amp\n"; break;
160    case AsmToken::AmpAmp:         outs() << "AmpAmp\n"; break;
161    case AsmToken::Caret:          outs() << "Caret\n"; break;
162    case AsmToken::Colon:          outs() << "Colon\n"; break;
163    case AsmToken::Comma:          outs() << "Comma\n"; break;
164    case AsmToken::Dollar:         outs() << "Dollar\n"; break;
165    case AsmToken::EndOfStatement: outs() << "EndOfStatement\n"; break;
166    case AsmToken::Eof:            outs() << "Eof\n"; break;
167    case AsmToken::Equal:          outs() << "Equal\n"; break;
168    case AsmToken::EqualEqual:     outs() << "EqualEqual\n"; break;
169    case AsmToken::Exclaim:        outs() << "Exclaim\n"; break;
170    case AsmToken::ExclaimEqual:   outs() << "ExclaimEqual\n"; break;
171    case AsmToken::Greater:        outs() << "Greater\n"; break;
172    case AsmToken::GreaterEqual:   outs() << "GreaterEqual\n"; break;
173    case AsmToken::GreaterGreater: outs() << "GreaterGreater\n"; break;
174    case AsmToken::LParen:         outs() << "LParen\n"; break;
175    case AsmToken::Less:           outs() << "Less\n"; break;
176    case AsmToken::LessEqual:      outs() << "LessEqual\n"; break;
177    case AsmToken::LessGreater:    outs() << "LessGreater\n"; break;
178    case AsmToken::LessLess:       outs() << "LessLess\n"; break;
179    case AsmToken::Minus:          outs() << "Minus\n"; break;
180    case AsmToken::Percent:        outs() << "Percent\n"; break;
181    case AsmToken::Pipe:           outs() << "Pipe\n"; break;
182    case AsmToken::PipePipe:       outs() << "PipePipe\n"; break;
183    case AsmToken::Plus:           outs() << "Plus\n"; break;
184    case AsmToken::RParen:         outs() << "RParen\n"; break;
185    case AsmToken::Slash:          outs() << "Slash\n"; break;
186    case AsmToken::Star:           outs() << "Star\n"; break;
187    case AsmToken::Tilde:          outs() << "Tilde\n"; break;
188    }
189  }
190
191  return Error;
192}
193
194static formatted_raw_ostream *GetOutputStream() {
195  if (OutputFilename == "")
196    OutputFilename = "-";
197
198  // Make sure that the Out file gets unlinked from the disk if we get a
199  // SIGINT.
200  if (OutputFilename != "-")
201    sys::RemoveFileOnSignal(sys::Path(OutputFilename));
202
203  std::string Err;
204  raw_fd_ostream *Out = new raw_fd_ostream(OutputFilename.c_str(), Err,
205                                           raw_fd_ostream::F_Binary);
206  if (!Err.empty()) {
207    errs() << Err << '\n';
208    delete Out;
209    return 0;
210  }
211
212  return new formatted_raw_ostream(*Out, formatted_raw_ostream::DELETE_STREAM);
213}
214
215static int AssembleInput(const char *ProgName) {
216  const Target *TheTarget = GetTarget(ProgName);
217  if (!TheTarget)
218    return 1;
219
220  std::string Error;
221  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, &Error);
222  if (Buffer == 0) {
223    errs() << ProgName << ": ";
224    if (Error.size())
225      errs() << Error << "\n";
226    else
227      errs() << "input file didn't read correctly.\n";
228    return 1;
229  }
230
231  SourceMgr SrcMgr;
232
233  // Tell SrcMgr about this buffer, which is what the parser will pick up.
234  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
235
236  // Record the location of the include directories so that the lexer can find
237  // it later.
238  SrcMgr.setIncludeDirs(IncludeDirs);
239
240  MCContext Ctx;
241  formatted_raw_ostream *Out = GetOutputStream();
242  if (!Out)
243    return 1;
244
245
246  // FIXME: We shouldn't need to do this (and link in codegen).
247  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName, ""));
248
249  if (!TM) {
250    errs() << ProgName << ": error: could not create target for triple '"
251           << TripleName << "'.\n";
252    return 1;
253  }
254
255  OwningPtr<MCInstPrinter> IP;
256  OwningPtr<MCCodeEmitter> CE;
257  OwningPtr<MCStreamer> Str;
258
259  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
260  assert(MAI && "Unable to create target asm info!");
261
262  if (FileType == OFT_AssemblyFile) {
263    IP.reset(TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI, *Out));
264    if (ShowEncoding)
265      CE.reset(TheTarget->createCodeEmitter(*TM));
266    Str.reset(createAsmStreamer(Ctx, *Out, *MAI, IP.get(), CE.get()));
267  } else {
268    assert(FileType == OFT_ObjectFile && "Invalid file type!");
269    CE.reset(TheTarget->createCodeEmitter(*TM));
270    Str.reset(createMachOStreamer(Ctx, *Out, CE.get()));
271  }
272
273  AsmParser Parser(SrcMgr, Ctx, *Str.get(), *MAI);
274  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(Parser));
275  if (!TAP) {
276    errs() << ProgName
277           << ": error: this target does not support assembly parsing.\n";
278    return 1;
279  }
280
281  Parser.setTargetParser(*TAP.get());
282
283  int Res = Parser.Run();
284  if (Out != &fouts())
285    delete Out;
286
287  return Res;
288}
289
290static int DisassembleInput(const char *ProgName) {
291  std::string Error;
292  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
293  if (TheTarget == 0) {
294    errs() << ProgName << ": error: unable to get target for '" << TripleName
295    << "', see --version and --triple.\n";
296    return 0;
297  }
298
299  std::string ErrorMessage;
300
301  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
302                                                      &ErrorMessage);
303
304  if (Buffer == 0) {
305    errs() << ProgName << ": ";
306    if (ErrorMessage.size())
307      errs() << ErrorMessage << "\n";
308    else
309      errs() << "input file didn't read correctly.\n";
310    return 1;
311  }
312
313  return Disassembler::disassemble(*TheTarget, TripleName, *Buffer);
314}
315
316
317int main(int argc, char **argv) {
318  // Print a stack trace if we signal out.
319  sys::PrintStackTraceOnErrorSignal();
320  PrettyStackTraceProgram X(argc, argv);
321  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
322
323  // Initialize targets and assembly printers/parsers.
324  llvm::InitializeAllTargetInfos();
325  // FIXME: We shouldn't need to initialize the Target(Machine)s.
326  llvm::InitializeAllTargets();
327  llvm::InitializeAllAsmPrinters();
328  llvm::InitializeAllAsmParsers();
329  llvm::InitializeAllDisassemblers();
330
331  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
332
333  switch (Action) {
334  default:
335  case AC_AsLex:
336    return AsLexInput(argv[0]);
337  case AC_Assemble:
338    return AssembleInput(argv[0]);
339  case AC_Disassemble:
340    return DisassembleInput(argv[0]);
341  }
342
343  return 0;
344}
345
346