llvm-mc.cpp revision 5532cf44a012149ce3afce43dbd0651b4d87a505 
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/Signals.h"
30#include "llvm/Target/TargetAsmParser.h"
31#include "llvm/Target/TargetData.h"
32#include "llvm/Target/TargetRegistry.h"
33#include "llvm/Target/TargetMachine.h"  // FIXME.
34#include "llvm/Target/TargetSelect.h"
35#include "llvm/MC/MCParser/AsmParser.h"
36#include "Disassembler.h"
37using namespace llvm;
38
39static cl::opt<std::string>
40InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
41
42static cl::opt<std::string>
43OutputFilename("o", cl::desc("Output filename"),
44               cl::value_desc("filename"));
45
46static cl::opt<bool>
47ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
48
49static cl::opt<bool>
50ShowInst("show-inst", cl::desc("Show internal instruction representation"));
51
52static cl::opt<unsigned>
53OutputAsmVariant("output-asm-variant",
54                 cl::desc("Syntax variant to use for output printing"));
55
56enum OutputFileType {
57  OFT_AssemblyFile,
58  OFT_ObjectFile
59};
60static cl::opt<OutputFileType>
61FileType("filetype", cl::init(OFT_AssemblyFile),
62  cl::desc("Choose an output file type:"),
63  cl::values(
64       clEnumValN(OFT_AssemblyFile, "asm",
65                  "Emit an assembly ('.s') file"),
66       clEnumValN(OFT_ObjectFile, "obj",
67                  "Emit a native object ('.o') file"),
68       clEnumValEnd));
69
70static cl::opt<bool>
71Force("f", cl::desc("Enable binary output on terminals"));
72
73static cl::list<std::string>
74IncludeDirs("I", cl::desc("Directory of include files"),
75            cl::value_desc("directory"), cl::Prefix);
76
77static cl::opt<std::string>
78TripleName("triple", cl::desc("Target triple to assemble for, "
79                              "see -version for available targets"),
80           cl::init(LLVM_HOSTTRIPLE));
81
82enum ActionType {
83  AC_AsLex,
84  AC_Assemble,
85  AC_Disassemble
86};
87
88static cl::opt<ActionType>
89Action(cl::desc("Action to perform:"),
90       cl::init(AC_Assemble),
91       cl::values(clEnumValN(AC_AsLex, "as-lex",
92                             "Lex tokens from a .s file"),
93                  clEnumValN(AC_Assemble, "assemble",
94                             "Assemble a .s file (default)"),
95                  clEnumValN(AC_Disassemble, "disassemble",
96                             "Disassemble strings of hex bytes"),
97                  clEnumValEnd));
98
99static const Target *GetTarget(const char *ProgName) {
100  // Get the target specific parser.
101  std::string Error;
102  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
103  if (TheTarget)
104    return TheTarget;
105
106  errs() << ProgName << ": error: unable to get target for '" << TripleName
107         << "', see --version and --triple.\n";
108  return 0;
109}
110
111static int AsLexInput(const char *ProgName) {
112  std::string ErrorMessage;
113  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
114                                                      &ErrorMessage);
115  if (Buffer == 0) {
116    errs() << ProgName << ": ";
117    if (ErrorMessage.size())
118      errs() << ErrorMessage << "\n";
119    else
120      errs() << "input file didn't read correctly.\n";
121    return 1;
122  }
123
124  SourceMgr SrcMgr;
125
126  // Tell SrcMgr about this buffer, which is what TGParser will pick up.
127  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
128
129  // Record the location of the include directories so that the lexer can find
130  // it later.
131  SrcMgr.setIncludeDirs(IncludeDirs);
132
133  const Target *TheTarget = GetTarget(ProgName);
134  if (!TheTarget)
135    return 1;
136
137  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
138  assert(MAI && "Unable to create target asm info!");
139
140  AsmLexer Lexer(*MAI);
141
142  bool Error = false;
143
144  while (Lexer.Lex().isNot(AsmToken::Eof)) {
145    switch (Lexer.getKind()) {
146    default:
147      SrcMgr.PrintMessage(Lexer.getLoc(), "unknown token", "warning");
148      Error = true;
149      break;
150    case AsmToken::Error:
151      Error = true; // error already printed.
152      break;
153    case AsmToken::Identifier:
154      outs() << "identifier: " << Lexer.getTok().getString() << '\n';
155      break;
156    case AsmToken::String:
157      outs() << "string: " << Lexer.getTok().getString() << '\n';
158      break;
159    case AsmToken::Integer:
160      outs() << "int: " << Lexer.getTok().getString() << '\n';
161      break;
162
163    case AsmToken::Amp:            outs() << "Amp\n"; break;
164    case AsmToken::AmpAmp:         outs() << "AmpAmp\n"; break;
165    case AsmToken::Caret:          outs() << "Caret\n"; break;
166    case AsmToken::Colon:          outs() << "Colon\n"; break;
167    case AsmToken::Comma:          outs() << "Comma\n"; break;
168    case AsmToken::Dollar:         outs() << "Dollar\n"; break;
169    case AsmToken::EndOfStatement: outs() << "EndOfStatement\n"; break;
170    case AsmToken::Eof:            outs() << "Eof\n"; break;
171    case AsmToken::Equal:          outs() << "Equal\n"; break;
172    case AsmToken::EqualEqual:     outs() << "EqualEqual\n"; break;
173    case AsmToken::Exclaim:        outs() << "Exclaim\n"; break;
174    case AsmToken::ExclaimEqual:   outs() << "ExclaimEqual\n"; break;
175    case AsmToken::Greater:        outs() << "Greater\n"; break;
176    case AsmToken::GreaterEqual:   outs() << "GreaterEqual\n"; break;
177    case AsmToken::GreaterGreater: outs() << "GreaterGreater\n"; break;
178    case AsmToken::LParen:         outs() << "LParen\n"; break;
179    case AsmToken::Less:           outs() << "Less\n"; break;
180    case AsmToken::LessEqual:      outs() << "LessEqual\n"; break;
181    case AsmToken::LessGreater:    outs() << "LessGreater\n"; break;
182    case AsmToken::LessLess:       outs() << "LessLess\n"; break;
183    case AsmToken::Minus:          outs() << "Minus\n"; break;
184    case AsmToken::Percent:        outs() << "Percent\n"; break;
185    case AsmToken::Pipe:           outs() << "Pipe\n"; break;
186    case AsmToken::PipePipe:       outs() << "PipePipe\n"; break;
187    case AsmToken::Plus:           outs() << "Plus\n"; break;
188    case AsmToken::RParen:         outs() << "RParen\n"; break;
189    case AsmToken::Slash:          outs() << "Slash\n"; break;
190    case AsmToken::Star:           outs() << "Star\n"; break;
191    case AsmToken::Tilde:          outs() << "Tilde\n"; break;
192    }
193  }
194
195  return Error;
196}
197
198static formatted_raw_ostream *GetOutputStream() {
199  if (OutputFilename == "")
200    OutputFilename = "-";
201
202  // Make sure that the Out file gets unlinked from the disk if we get a
203  // SIGINT.
204  if (OutputFilename != "-")
205    sys::RemoveFileOnSignal(sys::Path(OutputFilename));
206
207  std::string Err;
208  raw_fd_ostream *Out = new raw_fd_ostream(OutputFilename.c_str(), Err,
209                                           raw_fd_ostream::F_Binary);
210  if (!Err.empty()) {
211    errs() << Err << '\n';
212    delete Out;
213    return 0;
214  }
215
216  return new formatted_raw_ostream(*Out, formatted_raw_ostream::DELETE_STREAM);
217}
218
219static int AssembleInput(const char *ProgName) {
220  const Target *TheTarget = GetTarget(ProgName);
221  if (!TheTarget)
222    return 1;
223
224  std::string Error;
225  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename, &Error);
226  if (Buffer == 0) {
227    errs() << ProgName << ": ";
228    if (Error.size())
229      errs() << Error << "\n";
230    else
231      errs() << "input file didn't read correctly.\n";
232    return 1;
233  }
234
235  SourceMgr SrcMgr;
236
237  // Tell SrcMgr about this buffer, which is what the parser will pick up.
238  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
239
240  // Record the location of the include directories so that the lexer can find
241  // it later.
242  SrcMgr.setIncludeDirs(IncludeDirs);
243
244  MCContext Ctx;
245  formatted_raw_ostream *Out = GetOutputStream();
246  if (!Out)
247    return 1;
248
249
250  // FIXME: We shouldn't need to do this (and link in codegen).
251  OwningPtr<TargetMachine> TM(TheTarget->createTargetMachine(TripleName, ""));
252
253  if (!TM) {
254    errs() << ProgName << ": error: could not create target for triple '"
255           << TripleName << "'.\n";
256    return 1;
257  }
258
259  OwningPtr<MCInstPrinter> IP;
260  OwningPtr<MCCodeEmitter> CE;
261  OwningPtr<MCStreamer> Str;
262
263  const MCAsmInfo *MAI = TheTarget->createAsmInfo(TripleName);
264  assert(MAI && "Unable to create target asm info!");
265
266  if (FileType == OFT_AssemblyFile) {
267    IP.reset(TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI, *Out));
268    if (ShowEncoding)
269      CE.reset(TheTarget->createCodeEmitter(*TM));
270    Str.reset(createAsmStreamer(Ctx, *Out, *MAI,
271                                TM->getTargetData()->isLittleEndian(),
272                                /*asmverbose*/true, IP.get(), CE.get(),
273                                ShowInst));
274  } else {
275    assert(FileType == OFT_ObjectFile && "Invalid file type!");
276    CE.reset(TheTarget->createCodeEmitter(*TM));
277    Str.reset(createMachOStreamer(Ctx, *Out, CE.get()));
278  }
279
280  AsmParser Parser(SrcMgr, Ctx, *Str.get(), *MAI);
281  OwningPtr<TargetAsmParser> TAP(TheTarget->createAsmParser(Parser));
282  if (!TAP) {
283    errs() << ProgName
284           << ": error: this target does not support assembly parsing.\n";
285    return 1;
286  }
287
288  Parser.setTargetParser(*TAP.get());
289
290  int Res = Parser.Run();
291  if (Out != &fouts())
292    delete Out;
293
294  return Res;
295}
296
297static int DisassembleInput(const char *ProgName) {
298  std::string Error;
299  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
300  if (TheTarget == 0) {
301    errs() << ProgName << ": error: unable to get target for '" << TripleName
302    << "', see --version and --triple.\n";
303    return 0;
304  }
305
306  std::string ErrorMessage;
307
308  MemoryBuffer *Buffer = MemoryBuffer::getFileOrSTDIN(InputFilename,
309                                                      &ErrorMessage);
310
311  if (Buffer == 0) {
312    errs() << ProgName << ": ";
313    if (ErrorMessage.size())
314      errs() << ErrorMessage << "\n";
315    else
316      errs() << "input file didn't read correctly.\n";
317    return 1;
318  }
319
320  return Disassembler::disassemble(*TheTarget, TripleName, *Buffer);
321}
322
323
324int main(int argc, char **argv) {
325  // Print a stack trace if we signal out.
326  sys::PrintStackTraceOnErrorSignal();
327  PrettyStackTraceProgram X(argc, argv);
328  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
329
330  // Initialize targets and assembly printers/parsers.
331  llvm::InitializeAllTargetInfos();
332  // FIXME: We shouldn't need to initialize the Target(Machine)s.
333  llvm::InitializeAllTargets();
334  llvm::InitializeAllAsmPrinters();
335  llvm::InitializeAllAsmParsers();
336  llvm::InitializeAllDisassemblers();
337
338  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
339
340  switch (Action) {
341  default:
342  case AC_AsLex:
343    return AsLexInput(argv[0]);
344  case AC_Assemble:
345    return AssembleInput(argv[0]);
346  case AC_Disassemble:
347    return DisassembleInput(argv[0]);
348  }
349
350  return 0;
351}
352
353