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 "Disassembler.h"
16#include "llvm/MC/MCAsmBackend.h"
17#include "llvm/MC/MCAsmInfo.h"
18#include "llvm/MC/MCContext.h"
19#include "llvm/MC/MCInstPrinter.h"
20#include "llvm/MC/MCInstrInfo.h"
21#include "llvm/MC/MCObjectFileInfo.h"
22#include "llvm/MC/MCParser/AsmLexer.h"
23#include "llvm/MC/MCRegisterInfo.h"
24#include "llvm/MC/MCSectionMachO.h"
25#include "llvm/MC/MCStreamer.h"
26#include "llvm/MC/MCSubtargetInfo.h"
27#include "llvm/MC/MCTargetAsmParser.h"
28#include "llvm/MC/MCTargetOptionsCommandFlags.h"
29#include "llvm/Support/CommandLine.h"
30#include "llvm/Support/Compression.h"
31#include "llvm/Support/FileUtilities.h"
32#include "llvm/Support/FormattedStream.h"
33#include "llvm/Support/Host.h"
34#include "llvm/Support/ManagedStatic.h"
35#include "llvm/Support/MemoryBuffer.h"
36#include "llvm/Support/PrettyStackTrace.h"
37#include "llvm/Support/Signals.h"
38#include "llvm/Support/SourceMgr.h"
39#include "llvm/Support/TargetRegistry.h"
40#include "llvm/Support/TargetSelect.h"
41#include "llvm/Support/ToolOutputFile.h"
42using namespace llvm;
43
44static cl::opt<std::string>
45InputFilename(cl::Positional, cl::desc("<input file>"), cl::init("-"));
46
47static cl::opt<std::string>
48OutputFilename("o", cl::desc("Output filename"),
49               cl::value_desc("filename"));
50
51static cl::opt<bool>
52ShowEncoding("show-encoding", cl::desc("Show instruction encodings"));
53
54static cl::opt<bool>
55CompressDebugSections("compress-debug-sections",
56                      cl::desc("Compress DWARF debug sections"));
57
58static cl::opt<bool>
59ShowInst("show-inst", cl::desc("Show internal instruction representation"));
60
61static cl::opt<bool>
62ShowInstOperands("show-inst-operands",
63                 cl::desc("Show instructions operands as parsed"));
64
65static cl::opt<unsigned>
66OutputAsmVariant("output-asm-variant",
67                 cl::desc("Syntax variant to use for output printing"));
68
69static cl::opt<bool>
70PrintImmHex("print-imm-hex", cl::init(false),
71            cl::desc("Prefer hex format for immediate values"));
72
73enum OutputFileType {
74  OFT_Null,
75  OFT_AssemblyFile,
76  OFT_ObjectFile
77};
78static cl::opt<OutputFileType>
79FileType("filetype", cl::init(OFT_AssemblyFile),
80  cl::desc("Choose an output file type:"),
81  cl::values(
82       clEnumValN(OFT_AssemblyFile, "asm",
83                  "Emit an assembly ('.s') file"),
84       clEnumValN(OFT_Null, "null",
85                  "Don't emit anything (for timing purposes)"),
86       clEnumValN(OFT_ObjectFile, "obj",
87                  "Emit a native object ('.o') file"),
88       clEnumValEnd));
89
90static cl::list<std::string>
91IncludeDirs("I", cl::desc("Directory of include files"),
92            cl::value_desc("directory"), cl::Prefix);
93
94static cl::opt<std::string>
95ArchName("arch", cl::desc("Target arch to assemble for, "
96                          "see -version for available targets"));
97
98static cl::opt<std::string>
99TripleName("triple", cl::desc("Target triple to assemble for, "
100                              "see -version for available targets"));
101
102static cl::opt<std::string>
103MCPU("mcpu",
104     cl::desc("Target a specific cpu type (-mcpu=help for details)"),
105     cl::value_desc("cpu-name"),
106     cl::init(""));
107
108static cl::list<std::string>
109MAttrs("mattr",
110  cl::CommaSeparated,
111  cl::desc("Target specific attributes (-mattr=help for details)"),
112  cl::value_desc("a1,+a2,-a3,..."));
113
114static cl::opt<Reloc::Model>
115RelocModel("relocation-model",
116             cl::desc("Choose relocation model"),
117             cl::init(Reloc::Default),
118             cl::values(
119            clEnumValN(Reloc::Default, "default",
120                       "Target default relocation model"),
121            clEnumValN(Reloc::Static, "static",
122                       "Non-relocatable code"),
123            clEnumValN(Reloc::PIC_, "pic",
124                       "Fully relocatable, position independent code"),
125            clEnumValN(Reloc::DynamicNoPIC, "dynamic-no-pic",
126                       "Relocatable external references, non-relocatable code"),
127            clEnumValEnd));
128
129static cl::opt<llvm::CodeModel::Model>
130CMModel("code-model",
131        cl::desc("Choose code model"),
132        cl::init(CodeModel::Default),
133        cl::values(clEnumValN(CodeModel::Default, "default",
134                              "Target default code model"),
135                   clEnumValN(CodeModel::Small, "small",
136                              "Small code model"),
137                   clEnumValN(CodeModel::Kernel, "kernel",
138                              "Kernel code model"),
139                   clEnumValN(CodeModel::Medium, "medium",
140                              "Medium code model"),
141                   clEnumValN(CodeModel::Large, "large",
142                              "Large code model"),
143                   clEnumValEnd));
144
145static cl::opt<bool>
146NoInitialTextSection("n", cl::desc("Don't assume assembly file starts "
147                                   "in the text section"));
148
149static cl::opt<bool>
150GenDwarfForAssembly("g", cl::desc("Generate dwarf debugging info for assembly "
151                                  "source files"));
152
153static cl::opt<std::string>
154DebugCompilationDir("fdebug-compilation-dir",
155                    cl::desc("Specifies the debug info's compilation dir"));
156
157static cl::opt<std::string>
158MainFileName("main-file-name",
159             cl::desc("Specifies the name we should consider the input file"));
160
161static cl::opt<bool> SaveTempLabels("save-temp-labels",
162                                    cl::desc("Don't discard temporary labels"));
163
164static cl::opt<bool> NoExecStack("no-exec-stack",
165                                 cl::desc("File doesn't need an exec stack"));
166
167enum ActionType {
168  AC_AsLex,
169  AC_Assemble,
170  AC_Disassemble,
171  AC_MDisassemble,
172};
173
174static cl::opt<ActionType>
175Action(cl::desc("Action to perform:"),
176       cl::init(AC_Assemble),
177       cl::values(clEnumValN(AC_AsLex, "as-lex",
178                             "Lex tokens from a .s file"),
179                  clEnumValN(AC_Assemble, "assemble",
180                             "Assemble a .s file (default)"),
181                  clEnumValN(AC_Disassemble, "disassemble",
182                             "Disassemble strings of hex bytes"),
183                  clEnumValN(AC_MDisassemble, "mdis",
184                             "Marked up disassembly of strings of hex bytes"),
185                  clEnumValEnd));
186
187static const Target *GetTarget(const char *ProgName) {
188  // Figure out the target triple.
189  if (TripleName.empty())
190    TripleName = sys::getDefaultTargetTriple();
191  Triple TheTriple(Triple::normalize(TripleName));
192
193  // Get the target specific parser.
194  std::string Error;
195  const Target *TheTarget = TargetRegistry::lookupTarget(ArchName, TheTriple,
196                                                         Error);
197  if (!TheTarget) {
198    errs() << ProgName << ": " << Error;
199    return nullptr;
200  }
201
202  // Update the triple name and return the found target.
203  TripleName = TheTriple.getTriple();
204  return TheTarget;
205}
206
207static std::unique_ptr<tool_output_file> GetOutputStream() {
208  if (OutputFilename == "")
209    OutputFilename = "-";
210
211  std::error_code EC;
212  auto Out = llvm::make_unique<tool_output_file>(OutputFilename, EC,
213                                                 sys::fs::F_None);
214  if (EC) {
215    errs() << EC.message() << '\n';
216    return nullptr;
217  }
218
219  return Out;
220}
221
222static std::string DwarfDebugFlags;
223static void setDwarfDebugFlags(int argc, char **argv) {
224  if (!getenv("RC_DEBUG_OPTIONS"))
225    return;
226  for (int i = 0; i < argc; i++) {
227    DwarfDebugFlags += argv[i];
228    if (i + 1 < argc)
229      DwarfDebugFlags += " ";
230  }
231}
232
233static std::string DwarfDebugProducer;
234static void setDwarfDebugProducer(void) {
235  if(!getenv("DEBUG_PRODUCER"))
236    return;
237  DwarfDebugProducer += getenv("DEBUG_PRODUCER");
238}
239
240static int AsLexInput(SourceMgr &SrcMgr, MCAsmInfo &MAI,
241                      raw_ostream &OS) {
242
243  AsmLexer Lexer(MAI);
244  Lexer.setBuffer(SrcMgr.getMemoryBuffer(SrcMgr.getMainFileID())->getBuffer());
245
246  bool Error = false;
247  while (Lexer.Lex().isNot(AsmToken::Eof)) {
248    AsmToken Tok = Lexer.getTok();
249
250    switch (Tok.getKind()) {
251    default:
252      SrcMgr.PrintMessage(Lexer.getLoc(), SourceMgr::DK_Warning,
253                          "unknown token");
254      Error = true;
255      break;
256    case AsmToken::Error:
257      Error = true; // error already printed.
258      break;
259    case AsmToken::Identifier:
260      OS << "identifier: " << Lexer.getTok().getString();
261      break;
262    case AsmToken::Integer:
263      OS << "int: " << Lexer.getTok().getString();
264      break;
265    case AsmToken::Real:
266      OS << "real: " << Lexer.getTok().getString();
267      break;
268    case AsmToken::String:
269      OS << "string: " << Lexer.getTok().getString();
270      break;
271
272    case AsmToken::Amp:            OS << "Amp"; break;
273    case AsmToken::AmpAmp:         OS << "AmpAmp"; break;
274    case AsmToken::At:             OS << "At"; break;
275    case AsmToken::Caret:          OS << "Caret"; break;
276    case AsmToken::Colon:          OS << "Colon"; break;
277    case AsmToken::Comma:          OS << "Comma"; break;
278    case AsmToken::Dollar:         OS << "Dollar"; break;
279    case AsmToken::Dot:            OS << "Dot"; break;
280    case AsmToken::EndOfStatement: OS << "EndOfStatement"; break;
281    case AsmToken::Eof:            OS << "Eof"; break;
282    case AsmToken::Equal:          OS << "Equal"; break;
283    case AsmToken::EqualEqual:     OS << "EqualEqual"; break;
284    case AsmToken::Exclaim:        OS << "Exclaim"; break;
285    case AsmToken::ExclaimEqual:   OS << "ExclaimEqual"; break;
286    case AsmToken::Greater:        OS << "Greater"; break;
287    case AsmToken::GreaterEqual:   OS << "GreaterEqual"; break;
288    case AsmToken::GreaterGreater: OS << "GreaterGreater"; break;
289    case AsmToken::Hash:           OS << "Hash"; break;
290    case AsmToken::LBrac:          OS << "LBrac"; break;
291    case AsmToken::LCurly:         OS << "LCurly"; break;
292    case AsmToken::LParen:         OS << "LParen"; break;
293    case AsmToken::Less:           OS << "Less"; break;
294    case AsmToken::LessEqual:      OS << "LessEqual"; break;
295    case AsmToken::LessGreater:    OS << "LessGreater"; break;
296    case AsmToken::LessLess:       OS << "LessLess"; break;
297    case AsmToken::Minus:          OS << "Minus"; break;
298    case AsmToken::Percent:        OS << "Percent"; break;
299    case AsmToken::Pipe:           OS << "Pipe"; break;
300    case AsmToken::PipePipe:       OS << "PipePipe"; break;
301    case AsmToken::Plus:           OS << "Plus"; break;
302    case AsmToken::RBrac:          OS << "RBrac"; break;
303    case AsmToken::RCurly:         OS << "RCurly"; break;
304    case AsmToken::RParen:         OS << "RParen"; break;
305    case AsmToken::Slash:          OS << "Slash"; break;
306    case AsmToken::Star:           OS << "Star"; break;
307    case AsmToken::Tilde:          OS << "Tilde"; break;
308    }
309
310    // Print the token string.
311    OS << " (\"";
312    OS.write_escaped(Tok.getString());
313    OS << "\")\n";
314  }
315
316  return Error;
317}
318
319static int AssembleInput(const char *ProgName, const Target *TheTarget,
320                         SourceMgr &SrcMgr, MCContext &Ctx, MCStreamer &Str,
321                         MCAsmInfo &MAI, MCSubtargetInfo &STI,
322                         MCInstrInfo &MCII, MCTargetOptions &MCOptions) {
323  std::unique_ptr<MCAsmParser> Parser(
324      createMCAsmParser(SrcMgr, Ctx, Str, MAI));
325  std::unique_ptr<MCTargetAsmParser> TAP(
326      TheTarget->createMCAsmParser(STI, *Parser, MCII, MCOptions));
327
328  if (!TAP) {
329    errs() << ProgName
330           << ": error: this target does not support assembly parsing.\n";
331    return 1;
332  }
333
334  Parser->setShowParsedOperands(ShowInstOperands);
335  Parser->setTargetParser(*TAP);
336
337  int Res = Parser->Run(NoInitialTextSection);
338
339  return Res;
340}
341
342int main(int argc, char **argv) {
343  // Print a stack trace if we signal out.
344  sys::PrintStackTraceOnErrorSignal();
345  PrettyStackTraceProgram X(argc, argv);
346  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
347
348  // Initialize targets and assembly printers/parsers.
349  llvm::InitializeAllTargetInfos();
350  llvm::InitializeAllTargetMCs();
351  llvm::InitializeAllAsmParsers();
352  llvm::InitializeAllDisassemblers();
353
354  // Register the target printer for --version.
355  cl::AddExtraVersionPrinter(TargetRegistry::printRegisteredTargetsForVersion);
356
357  cl::ParseCommandLineOptions(argc, argv, "llvm machine code playground\n");
358  MCTargetOptions MCOptions = InitMCTargetOptionsFromFlags();
359  TripleName = Triple::normalize(TripleName);
360  Triple TheTriple(TripleName);
361  setDwarfDebugFlags(argc, argv);
362
363  setDwarfDebugProducer();
364
365  const char *ProgName = argv[0];
366  const Target *TheTarget = GetTarget(ProgName);
367  if (!TheTarget)
368    return 1;
369
370  ErrorOr<std::unique_ptr<MemoryBuffer>> BufferPtr =
371      MemoryBuffer::getFileOrSTDIN(InputFilename);
372  if (std::error_code EC = BufferPtr.getError()) {
373    errs() << ProgName << ": " << EC.message() << '\n';
374    return 1;
375  }
376  MemoryBuffer *Buffer = BufferPtr->get();
377
378  SourceMgr SrcMgr;
379
380  // Tell SrcMgr about this buffer, which is what the parser will pick up.
381  SrcMgr.AddNewSourceBuffer(std::move(*BufferPtr), 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  std::unique_ptr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
388  assert(MRI && "Unable to create target register info!");
389
390  std::unique_ptr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(*MRI, TripleName));
391  assert(MAI && "Unable to create target asm info!");
392
393  if (CompressDebugSections) {
394    if (!zlib::isAvailable()) {
395      errs() << ProgName
396             << ": build tools with zlib to enable -compress-debug-sections";
397      return 1;
398    }
399    MAI->setCompressDebugSections(true);
400  }
401
402  // FIXME: This is not pretty. MCContext has a ptr to MCObjectFileInfo and
403  // MCObjectFileInfo needs a MCContext reference in order to initialize itself.
404  MCObjectFileInfo MOFI;
405  MCContext Ctx(MAI.get(), MRI.get(), &MOFI, &SrcMgr);
406  MOFI.InitMCObjectFileInfo(TripleName, RelocModel, CMModel, Ctx);
407
408  if (SaveTempLabels)
409    Ctx.setAllowTemporaryLabels(false);
410
411  Ctx.setGenDwarfForAssembly(GenDwarfForAssembly);
412  // Default to 4 for dwarf version.
413  unsigned DwarfVersion = MCOptions.DwarfVersion ? MCOptions.DwarfVersion : 4;
414  if (DwarfVersion < 2 || DwarfVersion > 4) {
415    errs() << ProgName << ": Dwarf version " << DwarfVersion
416           << " is not supported." << '\n';
417    return 1;
418  }
419  Ctx.setDwarfVersion(DwarfVersion);
420  if (!DwarfDebugFlags.empty())
421    Ctx.setDwarfDebugFlags(StringRef(DwarfDebugFlags));
422  if (!DwarfDebugProducer.empty())
423    Ctx.setDwarfDebugProducer(StringRef(DwarfDebugProducer));
424  if (!DebugCompilationDir.empty())
425    Ctx.setCompilationDir(DebugCompilationDir);
426  if (!MainFileName.empty())
427    Ctx.setMainFileName(MainFileName);
428
429  // Package up features to be passed to target/subtarget
430  std::string FeaturesStr;
431  if (MAttrs.size()) {
432    SubtargetFeatures Features;
433    for (unsigned i = 0; i != MAttrs.size(); ++i)
434      Features.AddFeature(MAttrs[i]);
435    FeaturesStr = Features.getString();
436  }
437
438  std::unique_ptr<tool_output_file> Out = GetOutputStream();
439  if (!Out)
440    return 1;
441
442  std::unique_ptr<buffer_ostream> BOS;
443  raw_pwrite_stream *OS = &Out->os();
444  std::unique_ptr<MCStreamer> Str;
445
446  std::unique_ptr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo());
447  std::unique_ptr<MCSubtargetInfo> STI(
448      TheTarget->createMCSubtargetInfo(TripleName, MCPU, FeaturesStr));
449
450  MCInstPrinter *IP = nullptr;
451  if (FileType == OFT_AssemblyFile) {
452    IP = TheTarget->createMCInstPrinter(Triple(TripleName), OutputAsmVariant,
453                                        *MAI, *MCII, *MRI);
454
455    // Set the display preference for hex vs. decimal immediates.
456    IP->setPrintImmHex(PrintImmHex);
457
458    // Set up the AsmStreamer.
459    MCCodeEmitter *CE = nullptr;
460    MCAsmBackend *MAB = nullptr;
461    if (ShowEncoding) {
462      CE = TheTarget->createMCCodeEmitter(*MCII, *MRI, Ctx);
463      MAB = TheTarget->createMCAsmBackend(*MRI, TripleName, MCPU);
464    }
465    auto FOut = llvm::make_unique<formatted_raw_ostream>(*OS);
466    Str.reset(TheTarget->createAsmStreamer(
467        Ctx, std::move(FOut), /*asmverbose*/ true,
468        /*useDwarfDirectory*/ true, IP, CE, MAB, ShowInst));
469
470  } else if (FileType == OFT_Null) {
471    Str.reset(TheTarget->createNullStreamer(Ctx));
472  } else {
473    assert(FileType == OFT_ObjectFile && "Invalid file type!");
474
475    if (!Out->os().supportsSeeking()) {
476      BOS = make_unique<buffer_ostream>(Out->os());
477      OS = BOS.get();
478    }
479
480    MCCodeEmitter *CE = TheTarget->createMCCodeEmitter(*MCII, *MRI, Ctx);
481    MCAsmBackend *MAB = TheTarget->createMCAsmBackend(*MRI, TripleName, MCPU);
482    Str.reset(TheTarget->createMCObjectStreamer(TheTriple, Ctx, *MAB, *OS, CE,
483                                                *STI, RelaxAll,
484                                                /*DWARFMustBeAtTheEnd*/ false));
485    if (NoExecStack)
486      Str->InitSections(true);
487  }
488
489  int Res = 1;
490  bool disassemble = false;
491  switch (Action) {
492  case AC_AsLex:
493    Res = AsLexInput(SrcMgr, *MAI, Out->os());
494    break;
495  case AC_Assemble:
496    Res = AssembleInput(ProgName, TheTarget, SrcMgr, Ctx, *Str, *MAI, *STI,
497                        *MCII, MCOptions);
498    break;
499  case AC_MDisassemble:
500    assert(IP && "Expected assembly output");
501    IP->setUseMarkup(1);
502    disassemble = true;
503    break;
504  case AC_Disassemble:
505    disassemble = true;
506    break;
507  }
508  if (disassemble)
509    Res = Disassembler::disassemble(*TheTarget, TripleName, *STI, *Str,
510                                    *Buffer, SrcMgr, Out->os());
511
512  // Keep output if no errors.
513  if (Res == 0) Out->keep();
514  return Res;
515}
516