llvm-mc.cpp revision d0c478d95f440b4db76279fe47d6cf734a28fa9a
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/MCAsmBackend.h"
18#include "llvm/MC/MCAsmInfo.h"
19#include "llvm/MC/MCContext.h"
20#include "llvm/MC/MCCodeEmitter.h"
21#include "llvm/MC/MCInstPrinter.h"
22#include "llvm/MC/MCInstrInfo.h"
23#include "llvm/MC/MCObjectFileInfo.h"
24#include "llvm/MC/MCRegisterInfo.h"
25#include "llvm/MC/MCSectionMachO.h"
26#include "llvm/MC/MCStreamer.h"
27#include "llvm/MC/MCSubtargetInfo.h"
28#include "llvm/MC/MCTargetAsmParser.h"
29#include "llvm/MC/SubtargetFeature.h"
30#include "llvm/ADT/OwningPtr.h"
31#include "llvm/Support/CommandLine.h"
32#include "llvm/Support/FileUtilities.h"
33#include "llvm/Support/FormattedStream.h"
34#include "llvm/Support/ManagedStatic.h"
35#include "llvm/Support/MemoryBuffer.h"
36#include "llvm/Support/PrettyStackTrace.h"
37#include "llvm/Support/SourceMgr.h"
38#include "llvm/Support/ToolOutputFile.h"
39#include "llvm/Support/Host.h"
40#include "llvm/Support/Signals.h"
41#include "llvm/Support/TargetRegistry.h"
42#include "llvm/Support/TargetSelect.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
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::list<std::string>
110MAttrs("mattr",
111  cl::CommaSeparated,
112  cl::desc("Target specific attributes (-mattr=help for details)"),
113  cl::value_desc("a1,+a2,-a3,..."));
114
115static cl::opt<Reloc::Model>
116RelocModel("relocation-model",
117             cl::desc("Choose relocation model"),
118             cl::init(Reloc::Default),
119             cl::values(
120            clEnumValN(Reloc::Default, "default",
121                       "Target default relocation model"),
122            clEnumValN(Reloc::Static, "static",
123                       "Non-relocatable code"),
124            clEnumValN(Reloc::PIC_, "pic",
125                       "Fully relocatable, position independent code"),
126            clEnumValN(Reloc::DynamicNoPIC, "dynamic-no-pic",
127                       "Relocatable external references, non-relocatable code"),
128            clEnumValEnd));
129
130static cl::opt<llvm::CodeModel::Model>
131CMModel("code-model",
132        cl::desc("Choose code model"),
133        cl::init(CodeModel::Default),
134        cl::values(clEnumValN(CodeModel::Default, "default",
135                              "Target default code model"),
136                   clEnumValN(CodeModel::Small, "small",
137                              "Small code model"),
138                   clEnumValN(CodeModel::Kernel, "kernel",
139                              "Kernel code model"),
140                   clEnumValN(CodeModel::Medium, "medium",
141                              "Medium code model"),
142                   clEnumValN(CodeModel::Large, "large",
143                              "Large code model"),
144                   clEnumValEnd));
145
146static cl::opt<bool>
147NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
148                                   "in the text section"));
149
150static cl::opt<bool>
151SaveTempLabels("L", cl::desc("Don't discard temporary labels"));
152
153static cl::opt<bool>
154GenDwarfForAssembly("g", cl::desc("Generate dwarf debugging info for assembly "
155                                  "source files"));
156
157enum ActionType {
158  AC_AsLex,
159  AC_Assemble,
160  AC_Disassemble,
161  AC_EDisassemble
162};
163
164static cl::opt<ActionType>
165Action(cl::desc("Action to perform:"),
166       cl::init(AC_Assemble),
167       cl::values(clEnumValN(AC_AsLex, "as-lex",
168                             "Lex tokens from a .s file"),
169                  clEnumValN(AC_Assemble, "assemble",
170                             "Assemble a .s file (default)"),
171                  clEnumValN(AC_Disassemble, "disassemble",
172                             "Disassemble strings of hex bytes"),
173                  clEnumValN(AC_EDisassemble, "edis",
174                             "Enhanced disassembly of strings of hex bytes"),
175                  clEnumValEnd));
176
177static const Target *GetTarget(const char *ProgName) {
178  // Figure out the target triple.
179  if (TripleName.empty())
180    TripleName = sys::getDefaultTargetTriple();
181  Triple TheTriple(Triple::normalize(TripleName));
182
183  const Target *TheTarget = 0;
184  if (!ArchName.empty()) {
185    for (TargetRegistry::iterator it = TargetRegistry::begin(),
186           ie = TargetRegistry::end(); it != ie; ++it) {
187      if (ArchName == it->getName()) {
188        TheTarget = &*it;
189        break;
190      }
191    }
192
193    if (!TheTarget) {
194      errs() << ProgName << ": error: invalid target '" << ArchName << "'.\n";
195      return 0;
196    }
197
198    // Adjust the triple to match (if known), otherwise stick with the
199    // module/host triple.
200    Triple::ArchType Type = Triple::getArchTypeForLLVMName(ArchName);
201    if (Type != Triple::UnknownArch)
202      TheTriple.setArch(Type);
203  } else {
204    // Get the target specific parser.
205    std::string Error;
206    TheTarget = TargetRegistry::lookupTarget(TheTriple.getTriple(), Error);
207    if (TheTarget == 0) {
208      errs() << ProgName << ": error: unable to get target for '"
209             << TheTriple.getTriple()
210             << "', see --version and --triple.\n";
211      return 0;
212    }
213  }
214
215  TripleName = TheTriple.getTriple();
216  return TheTarget;
217}
218
219static tool_output_file *GetOutputStream() {
220  if (OutputFilename == "")
221    OutputFilename = "-";
222
223  std::string Err;
224  tool_output_file *Out = new tool_output_file(OutputFilename.c_str(), Err,
225                                               raw_fd_ostream::F_Binary);
226  if (!Err.empty()) {
227    errs() << Err << '\n';
228    delete Out;
229    return 0;
230  }
231
232  return Out;
233}
234
235static std::string DwarfDebugFlags;
236static void setDwarfDebugFlags(int argc, char **argv) {
237  if (!getenv("RC_DEBUG_OPTIONS"))
238    return;
239  for (int i = 0; i < argc; i++) {
240    DwarfDebugFlags += argv[i];
241    if (i + 1 < argc)
242      DwarfDebugFlags += " ";
243  }
244}
245
246static int AsLexInput(SourceMgr &SrcMgr, MCAsmInfo &MAI, tool_output_file *Out) {
247
248  AsmLexer Lexer(MAI);
249  Lexer.setBuffer(SrcMgr.getMemoryBuffer(0));
250
251  bool Error = false;
252  while (Lexer.Lex().isNot(AsmToken::Eof)) {
253    AsmToken Tok = Lexer.getTok();
254
255    switch (Tok.getKind()) {
256    default:
257      SrcMgr.PrintMessage(Lexer.getLoc(), SourceMgr::DK_Warning,
258                          "unknown token");
259      Error = true;
260      break;
261    case AsmToken::Error:
262      Error = true; // error already printed.
263      break;
264    case AsmToken::Identifier:
265      Out->os() << "identifier: " << Lexer.getTok().getString();
266      break;
267    case AsmToken::Integer:
268      Out->os() << "int: " << Lexer.getTok().getString();
269      break;
270    case AsmToken::Real:
271      Out->os() << "real: " << Lexer.getTok().getString();
272      break;
273    case AsmToken::Register:
274      Out->os() << "register: " << Lexer.getTok().getRegVal();
275      break;
276    case AsmToken::String:
277      Out->os() << "string: " << Lexer.getTok().getString();
278      break;
279
280    case AsmToken::Amp:            Out->os() << "Amp"; break;
281    case AsmToken::AmpAmp:         Out->os() << "AmpAmp"; break;
282    case AsmToken::At:             Out->os() << "At"; break;
283    case AsmToken::Caret:          Out->os() << "Caret"; break;
284    case AsmToken::Colon:          Out->os() << "Colon"; break;
285    case AsmToken::Comma:          Out->os() << "Comma"; break;
286    case AsmToken::Dollar:         Out->os() << "Dollar"; break;
287    case AsmToken::Dot:            Out->os() << "Dot"; break;
288    case AsmToken::EndOfStatement: Out->os() << "EndOfStatement"; break;
289    case AsmToken::Eof:            Out->os() << "Eof"; break;
290    case AsmToken::Equal:          Out->os() << "Equal"; break;
291    case AsmToken::EqualEqual:     Out->os() << "EqualEqual"; break;
292    case AsmToken::Exclaim:        Out->os() << "Exclaim"; break;
293    case AsmToken::ExclaimEqual:   Out->os() << "ExclaimEqual"; break;
294    case AsmToken::Greater:        Out->os() << "Greater"; break;
295    case AsmToken::GreaterEqual:   Out->os() << "GreaterEqual"; break;
296    case AsmToken::GreaterGreater: Out->os() << "GreaterGreater"; break;
297    case AsmToken::Hash:           Out->os() << "Hash"; break;
298    case AsmToken::LBrac:          Out->os() << "LBrac"; break;
299    case AsmToken::LCurly:         Out->os() << "LCurly"; break;
300    case AsmToken::LParen:         Out->os() << "LParen"; break;
301    case AsmToken::Less:           Out->os() << "Less"; break;
302    case AsmToken::LessEqual:      Out->os() << "LessEqual"; break;
303    case AsmToken::LessGreater:    Out->os() << "LessGreater"; break;
304    case AsmToken::LessLess:       Out->os() << "LessLess"; break;
305    case AsmToken::Minus:          Out->os() << "Minus"; break;
306    case AsmToken::Percent:        Out->os() << "Percent"; break;
307    case AsmToken::Pipe:           Out->os() << "Pipe"; break;
308    case AsmToken::PipePipe:       Out->os() << "PipePipe"; break;
309    case AsmToken::Plus:           Out->os() << "Plus"; break;
310    case AsmToken::RBrac:          Out->os() << "RBrac"; break;
311    case AsmToken::RCurly:         Out->os() << "RCurly"; break;
312    case AsmToken::RParen:         Out->os() << "RParen"; break;
313    case AsmToken::Slash:          Out->os() << "Slash"; break;
314    case AsmToken::Star:           Out->os() << "Star"; break;
315    case AsmToken::Tilde:          Out->os() << "Tilde"; break;
316    }
317
318    // Print the token string.
319    Out->os() << " (\"";
320    Out->os().write_escaped(Tok.getString());
321    Out->os() << "\")\n";
322  }
323
324  return Error;
325}
326
327static int AssembleInput(const char *ProgName, const Target *TheTarget,
328                         SourceMgr &SrcMgr, MCContext &Ctx, MCStreamer &Str,
329                         MCAsmInfo &MAI, MCSubtargetInfo &STI) {
330  OwningPtr<MCAsmParser> Parser(createMCAsmParser(SrcMgr, Ctx,
331                                                  Str, MAI));
332  OwningPtr<MCTargetAsmParser> TAP(TheTarget->createMCAsmParser(STI, *Parser));
333  if (!TAP) {
334    errs() << ProgName
335           << ": error: this target does not support assembly parsing.\n";
336    return 1;
337  }
338
339  Parser->setShowParsedOperands(ShowInstOperands);
340  Parser->setTargetParser(*TAP.get());
341
342  int Res = Parser->Run(NoInitialTextSection);
343
344  return Res;
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  llvm::InitializeAllTargetMCs();
356  llvm::InitializeAllAsmParsers();
357  llvm::InitializeAllDisassemblers();
358
359  // Register the target printer for --version.
360  cl::AddExtraVersionPrinter(TargetRegistry::printRegisteredTargetsForVersion);
361
362  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
363  TripleName = Triple::normalize(TripleName);
364  setDwarfDebugFlags(argc, argv);
365
366  const char *ProgName = argv[0];
367  const Target *TheTarget = GetTarget(ProgName);
368  if (!TheTarget)
369    return 1;
370
371  OwningPtr<MemoryBuffer> BufferPtr;
372  if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename, BufferPtr)) {
373    errs() << ProgName << ": " << ec.message() << '\n';
374    return 1;
375  }
376  MemoryBuffer *Buffer = BufferPtr.take();
377
378  SourceMgr SrcMgr;
379
380  // Tell SrcMgr about this buffer, which is what the parser will pick up.
381  SrcMgr.AddNewSourceBuffer(Buffer, SMLoc());
382
383  // Record the location of the include directories so that the lexer can find
384  // it later.
385  SrcMgr.setIncludeDirs(IncludeDirs);
386
387
388  llvm::OwningPtr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TripleName));
389  assert(MAI && "Unable to create target asm info!");
390
391  llvm::OwningPtr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
392  assert(MRI && "Unable to create target register info!");
393
394  // FIXME: This is not pretty. MCContext has a ptr to MCObjectFileInfo and
395  // MCObjectFileInfo needs a MCContext reference in order to initialize itself.
396  OwningPtr<MCObjectFileInfo> MOFI(new MCObjectFileInfo());
397  MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
398  MOFI->InitMCObjectFileInfo(TripleName, RelocModel, CMModel, Ctx);
399
400  if (SaveTempLabels)
401    Ctx.setAllowTemporaryLabels(false);
402
403  Ctx.setGenDwarfForAssembly(GenDwarfForAssembly);
404  if (!DwarfDebugFlags.empty())
405    Ctx.setDwarfDebugFlags(StringRef(DwarfDebugFlags));
406
407  // Package up features to be passed to target/subtarget
408  std::string FeaturesStr;
409  if (MAttrs.size()) {
410    SubtargetFeatures Features;
411    for (unsigned i = 0; i != MAttrs.size(); ++i)
412      Features.AddFeature(MAttrs[i]);
413    FeaturesStr = Features.getString();
414  }
415
416  OwningPtr<tool_output_file> Out(GetOutputStream());
417  if (!Out)
418    return 1;
419
420  formatted_raw_ostream FOS(Out->os());
421  OwningPtr<MCStreamer> Str;
422
423  OwningPtr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo());
424  OwningPtr<MCSubtargetInfo>
425    STI(TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
426
427  if (FileType == OFT_AssemblyFile) {
428    MCInstPrinter *IP =
429      TheTarget->createMCInstPrinter(OutputAsmVariant, *MAI, *MCII, *MRI, *STI);
430    MCCodeEmitter *CE = 0;
431    MCAsmBackend *MAB = 0;
432    if (ShowEncoding) {
433      CE = TheTarget->createMCCodeEmitter(*MCII, *STI, Ctx);
434      MAB = TheTarget->createMCAsmBackend(TripleName);
435    }
436    Str.reset(TheTarget->createAsmStreamer(Ctx, FOS, /*asmverbose*/true,
437                                           /*useLoc*/ true,
438                                           /*useCFI*/ true,
439                                           /*useDwarfDirectory*/ true,
440                                           IP, CE, MAB, ShowInst));
441
442  } else if (FileType == OFT_Null) {
443    Str.reset(createNullStreamer(Ctx));
444  } else {
445    assert(FileType == OFT_ObjectFile && "Invalid file type!");
446    MCCodeEmitter *CE = TheTarget->createMCCodeEmitter(*MCII, *STI, Ctx);
447    MCAsmBackend *MAB = TheTarget->createMCAsmBackend(TripleName);
448    Str.reset(TheTarget->createMCObjectStreamer(TripleName, Ctx, *MAB,
449                                                FOS, CE, RelaxAll,
450                                                NoExecStack));
451  }
452
453  int Res = 1;
454  switch (Action) {
455  case AC_AsLex:
456    Res = AsLexInput(SrcMgr, *MAI, Out.get());
457    break;
458  case AC_Assemble:
459    Res = AssembleInput(ProgName, TheTarget, SrcMgr, Ctx, *Str, *MAI, *STI);
460    break;
461  case AC_Disassemble:
462    Res = Disassembler::disassemble(*TheTarget, TripleName, *STI, *Str,
463                                    *Buffer, SrcMgr, Out->os());
464    break;
465  case AC_EDisassemble:
466    Res =  Disassembler::disassembleEnhanced(TripleName, *Buffer, SrcMgr, Out->os());
467    break;
468  }
469
470  // Keep output if no errors.
471  if (Res == 0) Out->keep();
472  return Res;
473}
474