llvm-mc.cpp revision 4c42a6de9f5456cc1b28f2d37db589f580f2adc7
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/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/Host.h"
30#include "llvm/System/Signals.h"
31#include "llvm/Target/TargetAsmBackend.h"
32#include "llvm/Target/TargetAsmParser.h"
33#include "llvm/Target/TargetData.h"
34#include "llvm/Target/TargetRegistry.h"
35#include "llvm/Target/TargetMachine.h"  // FIXME.
36#include "llvm/Target/TargetSelect.h"
37#include "llvm/MC/MCParser/AsmParser.h"
38#include "Disassembler.h"
39using namespace llvm;
40
41static cl::opt<std::string>
42InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
43
44static cl::opt<std::string>
45OutputFilename("o", cl::desc("Output filename"),
46               cl::value_desc("filename"));
47
48static cl::opt<bool>
49ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
50
51static cl::opt<bool>
52ShowInst("show-inst", cl::desc("Show internal instruction representation"));
53
54static cl::opt<unsigned>
55OutputAsmVariant("output-asm-variant",
56                 cl::desc("Syntax variant to use for output printing"));
57
58enum OutputFileType {
59  OFT_AssemblyFile,
60  OFT_ObjectFile
61};
62static cl::opt<OutputFileType>
63FileType("filetype", cl::init(OFT_AssemblyFile),
64  cl::desc("Choose an output file type:"),
65  cl::values(
66       clEnumValN(OFT_AssemblyFile, "asm",
67                  "Emit an assembly ('.s') file"),
68       clEnumValN(OFT_ObjectFile, "obj",
69                  "Emit a native object ('.o') file"),
70       clEnumValEnd));
71
72static cl::opt<bool>
73Force("f", cl::desc("Enable binary output on terminals"));
74
75static cl::list<std::string>
76IncludeDirs("I", cl::desc("Directory of include files"),
77            cl::value_desc("directory"), cl::Prefix);
78
79static cl::opt<std::string>
80ArchName("arch", cl::desc("Target arch to assemble for, "
81                            "see -version for available targets"));
82
83static cl::opt<std::string>
84TripleName("triple", cl::desc("Target triple to assemble for, "
85                              "see -version for available targets"));
86
87static cl::opt<bool>
88NoInitialTextSection("n", cl::desc(
89                   "Don't assume assembly file starts in the text section"));
90
91enum ActionType {
92  AC_AsLex,
93  AC_Assemble,
94  AC_Disassemble
95};
96
97static cl::opt<ActionType>
98Action(cl::desc("Action to perform:"),
99       cl::init(AC_Assemble),
100       cl::values(clEnumValN(AC_AsLex, "as-lex",
101                             "Lex tokens from a .s file"),
102                  clEnumValN(AC_Assemble, "assemble",
103                             "Assemble a .s file (default)"),
104                  clEnumValN(AC_Disassemble, "disassemble",
105                             "Disassemble strings of hex bytes"),
106                  clEnumValEnd));
107
108static const Target *GetTarget(const char *ProgName) {
109  // Figure out the target triple.
110  if (TripleName.empty())
111    TripleName = sys::getHostTriple();
112  if (!ArchName.empty()) {
113    llvm::Triple TT(TripleName);
114    TT.setArchName(ArchName);
115    TripleName = TT.str();
116  }
117
118  // Get the target specific parser.
119  std::string Error;
120  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
121  if (TheTarget)
122    return TheTarget;
123
124  errs() << ProgName << ": error: unable to get target for '" << TripleName
125         << "', see --version and --triple.\n";
126  return 0;
127}
128
129static int AsLexInput(const char *ProgName) {
130  std::string ErrorMessage;
131  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
132                                                      &ErrorMessage);
133  if (Buffer == 0) {
134    errs() << ProgName << ": ";
135    if (ErrorMessage.size())
136      errs() << ErrorMessage << "\n";
137    else
138      errs() << "input file didn't read correctly.\n";
139    return 1;
140  }
141
142  SourceMgr SrcMgr;
143
144  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
145  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
146
147  // Record the location of the include directories so that the lexer can find
148  // it later.
149  SrcMgr.setIncludeDirs(IncludeDirs);
150
151  const Target *TheTarget = GetTarget(ProgName);
152  if (!TheTarget)
153    return 1;
154
155  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
156  assert(MAI && "Unable to create target asm info!");
157
158  AsmLexer Lexer(*MAI);
159
160  bool Error = false;
161
162  while (Lexer.Lex().isNot(AsmToken::Eof)) {
163    switch (Lexer.getKind()) {
164    default:
165      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
166      Error = true;
167      break;
168    case AsmToken::Error:
169      Error = true; // error already printed.
170      break;
171    case AsmToken::Identifier:
172      outs() << "identifier: " << Lexer.getTok().getString() << '\n';
173      break;
174    case AsmToken::String:
175      outs() << "string: " << Lexer.getTok().getString() << '\n';
176      break;
177    case AsmToken::Integer:
178      outs() << "int: " << Lexer.getTok().getString() << '\n';
179      break;
180
181    case AsmToken::Amp:            outs() << "Amp\n"; break;
182    case AsmToken::AmpAmp:         outs() << "AmpAmp\n"; break;
183    case AsmToken::Caret:          outs() << "Caret\n"; break;
184    case AsmToken::Colon:          outs() << "Colon\n"; break;
185    case AsmToken::Comma:          outs() << "Comma\n"; break;
186    case AsmToken::Dollar:         outs() << "Dollar\n"; break;
187    case AsmToken::EndOfStatement: outs() << "EndOfStatement\n"; break;
188    case AsmToken::Eof:            outs() << "Eof\n"; break;
189    case AsmToken::Equal:          outs() << "Equal\n"; break;
190    case AsmToken::EqualEqual:     outs() << "EqualEqual\n"; break;
191    case AsmToken::Exclaim:        outs() << "Exclaim\n"; break;
192    case AsmToken::ExclaimEqual:   outs() << "ExclaimEqual\n"; break;
193    case AsmToken::Greater:        outs() << "Greater\n"; break;
194    case AsmToken::GreaterEqual:   outs() << "GreaterEqual\n"; break;
195    case AsmToken::GreaterGreater: outs() << "GreaterGreater\n"; break;
196    case AsmToken::LParen:         outs() << "LParen\n"; break;
197    case AsmToken::Less:           outs() << "Less\n"; break;
198    case AsmToken::LessEqual:      outs() << "LessEqual\n"; break;
199    case AsmToken::LessGreater:    outs() << "LessGreater\n"; break;
200    case AsmToken::LessLess:       outs() << "LessLess\n"; break;
201    case AsmToken::Minus:          outs() << "Minus\n"; break;
202    case AsmToken::Percent:        outs() << "Percent\n"; break;
203    case AsmToken::Pipe:           outs() << "Pipe\n"; break;
204    case AsmToken::PipePipe:       outs() << "PipePipe\n"; break;
205    case AsmToken::Plus:           outs() << "Plus\n"; break;
206    case AsmToken::RParen:         outs() << "RParen\n"; break;
207    case AsmToken::Slash:          outs() << "Slash\n"; break;
208    case AsmToken::Star:           outs() << "Star\n"; break;
209    case AsmToken::Tilde:          outs() << "Tilde\n"; break;
210    }
211  }
212
213  return Error;
214}
215
216static formatted_raw_ostream *GetOutputStream() {
217  if (OutputFilename == "")
218    OutputFilename = "-";
219
220  // Make sure that the Out file gets unlinked from the disk if we get a
221  // SIGINT.
222  if (OutputFilename != "-")
223    sys::RemoveFileOnSignal(sys::Path(OutputFilename));
224
225  std::string Err;
226  raw_fd_ostream *Out = new raw_fd_ostream(OutputFilename.c_str(), Err,
227                                           raw_fd_ostream::F_Binary);
228  if (!Err.empty()) {
229    errs() << Err << '\n';
230    delete Out;
231    return 0;
232  }
233
234  return new formatted_raw_ostream(*Out, formatted_raw_ostream::DELETE_STREAM);
235}
236
237static int AssembleInput(const char *ProgName) {
238  const Target *TheTarget = GetTarget(ProgName);
239  if (!TheTarget)
240    return 1;
241
242  std::string Error;
243  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, &Error);
244  if (Buffer == 0) {
245    errs() << ProgName << ": ";
246    if (Error.size())
247      errs() << Error << "\n";
248    else
249      errs() << "input file didn't read correctly.\n";
250    return 1;
251  }
252
253  SourceMgr SrcMgr;
254
255  // Tell SrcMgr about this buffer, which is what the parser will pick up.
256  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
257
258  // Record the location of the include directories so that the lexer can find
259  // it later.
260  SrcMgr.setIncludeDirs(IncludeDirs);
261
262
263  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
264  assert(MAI && "Unable to create target asm info!");
265
266  MCContext Ctx(*MAI);
267  formatted_raw_ostream *Out = GetOutputStream();
268  if (!Out)
269    return 1;
270
271
272  // FIXME: We shouldn't need to do this (and link in codegen).
273  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName, ""));
274
275  if (!TM) {
276    errs() << ProgName << ": error: could not create target for triple '"
277           << TripleName << "'.\n";
278    return 1;
279  }
280
281  OwningPtr<MCCodeEmitter> CE;
282  OwningPtr<MCStreamer> Str;
283  OwningPtr<TargetAsmBackend> TAB;
284
285  if (FileType == OFT_AssemblyFile) {
286    MCInstPrinter *IP =
287      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI, *Out);
288    if (ShowEncoding)
289      CE.reset(TheTarget->createCodeEmitter(*TM, Ctx));
290    Str.reset(createAsmStreamer(Ctx, *Out,TM->getTargetData()->isLittleEndian(),
291                                /*asmverbose*/true, IP, CE.get(), ShowInst));
292  } else {
293    assert(FileType == OFT_ObjectFile && "Invalid file type!");
294    CE.reset(TheTarget->createCodeEmitter(*TM, Ctx));
295    TAB.reset(TheTarget->createAsmBackend(TripleName));
296    Str.reset(createMachOStreamer(Ctx, *TAB, *Out, CE.get()));
297  }
298
299  AsmParser Parser(SrcMgr, Ctx, *Str.get(), *MAI);
300  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(Parser));
301  if (!TAP) {
302    errs() << ProgName
303           << ": error: this target does not support assembly parsing.\n";
304    return 1;
305  }
306
307  Parser.setTargetParser(*TAP.get());
308
309  int Res = Parser.Run(NoInitialTextSection);
310  if (Out != &fouts())
311    delete Out;
312
313  // Delete output on errors.
314  if (Res && OutputFilename != "-")
315    sys::Path(OutputFilename).eraseFromDisk();
316
317  return Res;
318}
319
320static int DisassembleInput(const char *ProgName) {
321  std::string Error;
322  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
323  if (TheTarget == 0) {
324    errs() << ProgName << ": error: unable to get target for '" << TripleName
325    << "', see --version and --triple.\n";
326    return 0;
327  }
328
329  std::string ErrorMessage;
330
331  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
332                                                      &ErrorMessage);
333
334  if (Buffer == 0) {
335    errs() << ProgName << ": ";
336    if (ErrorMessage.size())
337      errs() << ErrorMessage << "\n";
338    else
339      errs() << "input file didn't read correctly.\n";
340    return 1;
341  }
342
343  return Disassembler::disassemble(*TheTarget, TripleName, *Buffer);
344}
345
346
347int main(int argc, char **argv) {
348  // Print a stack trace if we signal out.
349  sys::PrintStackTraceOnErrorSignal();
350  PrettyStackTraceProgram X(argc, argv);
351  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
352
353  // Initialize targets and assembly printers/parsers.
354  llvm::InitializeAllTargetInfos();
355  // FIXME: We shouldn't need to initialize the Target(Machine)s.
356  llvm::InitializeAllTargets();
357  llvm::InitializeAllAsmPrinters();
358  llvm::InitializeAllAsmParsers();
359  llvm::InitializeAllDisassemblers();
360
361  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
362
363  switch (Action) {
364  default:
365  case AC_AsLex:
366    return AsLexInput(argv[0]);
367  case AC_Assemble:
368    return AssembleInput(argv[0]);
369  case AC_Disassemble:
370    return DisassembleInput(argv[0]);
371  }
372
373  return 0;
374}
375
376