1 //===-- llvm-objdump.cpp - Object file dumping utility for llvm -----------===// 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 program is a utility that works like binutils "objdump", that is, it 11 // dumps out a plethora of information about an object file depending on the 12 // flags. 13 // 14 // The flags and output of this program should be near identical to those of 15 // binutils objdump. 16 // 17 //===----------------------------------------------------------------------===// 18 19 #include "llvm-objdump.h" 20 #include "llvm/ADT/STLExtras.h" 21 #include "llvm/ADT/StringExtras.h" 22 #include "llvm/ADT/Triple.h" 23 #include "llvm/MC/MCAsmInfo.h" 24 #include "llvm/MC/MCAtom.h" 25 #include "llvm/MC/MCContext.h" 26 #include "llvm/MC/MCDisassembler.h" 27 #include "llvm/MC/MCFunction.h" 28 #include "llvm/MC/MCInst.h" 29 #include "llvm/MC/MCInstPrinter.h" 30 #include "llvm/MC/MCInstrAnalysis.h" 31 #include "llvm/MC/MCInstrInfo.h" 32 #include "llvm/MC/MCModule.h" 33 #include "llvm/MC/MCModuleYAML.h" 34 #include "llvm/MC/MCObjectDisassembler.h" 35 #include "llvm/MC/MCObjectFileInfo.h" 36 #include "llvm/MC/MCObjectSymbolizer.h" 37 #include "llvm/MC/MCRegisterInfo.h" 38 #include "llvm/MC/MCRelocationInfo.h" 39 #include "llvm/MC/MCSubtargetInfo.h" 40 #include "llvm/Object/Archive.h" 41 #include "llvm/Object/COFF.h" 42 #include "llvm/Object/MachO.h" 43 #include "llvm/Object/ObjectFile.h" 44 #include "llvm/Support/Casting.h" 45 #include "llvm/Support/CommandLine.h" 46 #include "llvm/Support/Debug.h" 47 #include "llvm/Support/FileSystem.h" 48 #include "llvm/Support/Format.h" 49 #include "llvm/Support/GraphWriter.h" 50 #include "llvm/Support/Host.h" 51 #include "llvm/Support/ManagedStatic.h" 52 #include "llvm/Support/MemoryBuffer.h" 53 #include "llvm/Support/MemoryObject.h" 54 #include "llvm/Support/PrettyStackTrace.h" 55 #include "llvm/Support/Signals.h" 56 #include "llvm/Support/SourceMgr.h" 57 #include "llvm/Support/TargetRegistry.h" 58 #include "llvm/Support/TargetSelect.h" 59 #include "llvm/Support/raw_ostream.h" 60 #include "llvm/Support/system_error.h" 61 #include <algorithm> 62 #include <cctype> 63 #include <cstring> 64 65 using namespace llvm; 66 using namespace object; 67 68 static cl::list<std::string> 69 InputFilenames(cl::Positional, cl::desc("<input object files>"),cl::ZeroOrMore); 70 71 static cl::opt<bool> 72 Disassemble("disassemble", 73 cl::desc("Display assembler mnemonics for the machine instructions")); 74 static cl::alias 75 Disassembled("d", cl::desc("Alias for --disassemble"), 76 cl::aliasopt(Disassemble)); 77 78 static cl::opt<bool> 79 Relocations("r", cl::desc("Display the relocation entries in the file")); 80 81 static cl::opt<bool> 82 SectionContents("s", cl::desc("Display the content of each section")); 83 84 static cl::opt<bool> 85 SymbolTable("t", cl::desc("Display the symbol table")); 86 87 static cl::opt<bool> 88 MachOOpt("macho", cl::desc("Use MachO specific object file parser")); 89 static cl::alias 90 MachOm("m", cl::desc("Alias for --macho"), cl::aliasopt(MachOOpt)); 91 92 cl::opt<std::string> 93 llvm::TripleName("triple", cl::desc("Target triple to disassemble for, " 94 "see -version for available targets")); 95 96 cl::opt<std::string> 97 llvm::ArchName("arch", cl::desc("Target arch to disassemble for, " 98 "see -version for available targets")); 99 100 static cl::opt<bool> 101 SectionHeaders("section-headers", cl::desc("Display summaries of the headers " 102 "for each section.")); 103 static cl::alias 104 SectionHeadersShort("headers", cl::desc("Alias for --section-headers"), 105 cl::aliasopt(SectionHeaders)); 106 static cl::alias 107 SectionHeadersShorter("h", cl::desc("Alias for --section-headers"), 108 cl::aliasopt(SectionHeaders)); 109 110 static cl::list<std::string> 111 MAttrs("mattr", 112 cl::CommaSeparated, 113 cl::desc("Target specific attributes"), 114 cl::value_desc("a1,+a2,-a3,...")); 115 116 static cl::opt<bool> 117 NoShowRawInsn("no-show-raw-insn", cl::desc("When disassembling instructions, " 118 "do not print the instruction bytes.")); 119 120 static cl::opt<bool> 121 UnwindInfo("unwind-info", cl::desc("Display unwind information")); 122 123 static cl::alias 124 UnwindInfoShort("u", cl::desc("Alias for --unwind-info"), 125 cl::aliasopt(UnwindInfo)); 126 127 static cl::opt<bool> 128 PrivateHeaders("private-headers", 129 cl::desc("Display format specific file headers")); 130 131 static cl::alias 132 PrivateHeadersShort("p", cl::desc("Alias for --private-headers"), 133 cl::aliasopt(PrivateHeaders)); 134 135 static cl::opt<bool> 136 Symbolize("symbolize", cl::desc("When disassembling instructions, " 137 "try to symbolize operands.")); 138 139 static cl::opt<bool> 140 CFG("cfg", cl::desc("Create a CFG for every function found in the object" 141 " and write it to a graphviz file")); 142 143 // FIXME: Does it make sense to have a dedicated tool for yaml cfg output? 144 static cl::opt<std::string> 145 YAMLCFG("yaml-cfg", 146 cl::desc("Create a CFG and write it as a YAML MCModule."), 147 cl::value_desc("yaml output file")); 148 149 static StringRef ToolName; 150 151 bool llvm::error(error_code EC) { 152 if (!EC) 153 return false; 154 155 outs() << ToolName << ": error reading file: " << EC.message() << ".\n"; 156 outs().flush(); 157 return true; 158 } 159 160 static const Target *getTarget(const ObjectFile *Obj = NULL) { 161 // Figure out the target triple. 162 llvm::Triple TheTriple("unknown-unknown-unknown"); 163 if (TripleName.empty()) { 164 if (Obj) { 165 TheTriple.setArch(Triple::ArchType(Obj->getArch())); 166 // TheTriple defaults to ELF, and COFF doesn't have an environment: 167 // the best we can do here is indicate that it is mach-o. 168 if (Obj->isMachO()) 169 TheTriple.setObjectFormat(Triple::MachO); 170 } 171 } else 172 TheTriple.setTriple(Triple::normalize(TripleName)); 173 174 // Get the target specific parser. 175 std::string Error; 176 const Target *TheTarget = TargetRegistry::lookupTarget(ArchName, TheTriple, 177 Error); 178 if (!TheTarget) { 179 errs() << ToolName << ": " << Error; 180 return 0; 181 } 182 183 // Update the triple name and return the found target. 184 TripleName = TheTriple.getTriple(); 185 return TheTarget; 186 } 187 188 // Write a graphviz file for the CFG inside an MCFunction. 189 // FIXME: Use GraphWriter 190 static void emitDOTFile(const char *FileName, const MCFunction &f, 191 MCInstPrinter *IP) { 192 // Start a new dot file. 193 std::string Error; 194 raw_fd_ostream Out(FileName, Error, sys::fs::F_Text); 195 if (!Error.empty()) { 196 errs() << "llvm-objdump: warning: " << Error << '\n'; 197 return; 198 } 199 200 Out << "digraph \"" << f.getName() << "\" {\n"; 201 Out << "graph [ rankdir = \"LR\" ];\n"; 202 for (MCFunction::const_iterator i = f.begin(), e = f.end(); i != e; ++i) { 203 // Only print blocks that have predecessors. 204 bool hasPreds = (*i)->pred_begin() != (*i)->pred_end(); 205 206 if (!hasPreds && i != f.begin()) 207 continue; 208 209 Out << '"' << (*i)->getInsts()->getBeginAddr() << "\" [ label=\"<a>"; 210 // Print instructions. 211 for (unsigned ii = 0, ie = (*i)->getInsts()->size(); ii != ie; 212 ++ii) { 213 if (ii != 0) // Not the first line, start a new row. 214 Out << '|'; 215 if (ii + 1 == ie) // Last line, add an end id. 216 Out << "<o>"; 217 218 // Escape special chars and print the instruction in mnemonic form. 219 std::string Str; 220 raw_string_ostream OS(Str); 221 IP->printInst(&(*i)->getInsts()->at(ii).Inst, OS, ""); 222 Out << DOT::EscapeString(OS.str()); 223 } 224 Out << "\" shape=\"record\" ];\n"; 225 226 // Add edges. 227 for (MCBasicBlock::succ_const_iterator si = (*i)->succ_begin(), 228 se = (*i)->succ_end(); si != se; ++si) 229 Out << (*i)->getInsts()->getBeginAddr() << ":o -> " 230 << (*si)->getInsts()->getBeginAddr() << ":a\n"; 231 } 232 Out << "}\n"; 233 } 234 235 void llvm::DumpBytes(StringRef bytes) { 236 static const char hex_rep[] = "0123456789abcdef"; 237 // FIXME: The real way to do this is to figure out the longest instruction 238 // and align to that size before printing. I'll fix this when I get 239 // around to outputting relocations. 240 // 15 is the longest x86 instruction 241 // 3 is for the hex rep of a byte + a space. 242 // 1 is for the null terminator. 243 enum { OutputSize = (15 * 3) + 1 }; 244 char output[OutputSize]; 245 246 assert(bytes.size() <= 15 247 && "DumpBytes only supports instructions of up to 15 bytes"); 248 memset(output, ' ', sizeof(output)); 249 unsigned index = 0; 250 for (StringRef::iterator i = bytes.begin(), 251 e = bytes.end(); i != e; ++i) { 252 output[index] = hex_rep[(*i & 0xF0) >> 4]; 253 output[index + 1] = hex_rep[*i & 0xF]; 254 index += 3; 255 } 256 257 output[sizeof(output) - 1] = 0; 258 outs() << output; 259 } 260 261 bool llvm::RelocAddressLess(RelocationRef a, RelocationRef b) { 262 uint64_t a_addr, b_addr; 263 if (error(a.getOffset(a_addr))) return false; 264 if (error(b.getOffset(b_addr))) return false; 265 return a_addr < b_addr; 266 } 267 268 static void DisassembleObject(const ObjectFile *Obj, bool InlineRelocs) { 269 const Target *TheTarget = getTarget(Obj); 270 // getTarget() will have already issued a diagnostic if necessary, so 271 // just bail here if it failed. 272 if (!TheTarget) 273 return; 274 275 // Package up features to be passed to target/subtarget 276 std::string FeaturesStr; 277 if (MAttrs.size()) { 278 SubtargetFeatures Features; 279 for (unsigned i = 0; i != MAttrs.size(); ++i) 280 Features.AddFeature(MAttrs[i]); 281 FeaturesStr = Features.getString(); 282 } 283 284 std::unique_ptr<const MCRegisterInfo> MRI( 285 TheTarget->createMCRegInfo(TripleName)); 286 if (!MRI) { 287 errs() << "error: no register info for target " << TripleName << "\n"; 288 return; 289 } 290 291 // Set up disassembler. 292 std::unique_ptr<const MCAsmInfo> AsmInfo( 293 TheTarget->createMCAsmInfo(*MRI, TripleName)); 294 if (!AsmInfo) { 295 errs() << "error: no assembly info for target " << TripleName << "\n"; 296 return; 297 } 298 299 std::unique_ptr<const MCSubtargetInfo> STI( 300 TheTarget->createMCSubtargetInfo(TripleName, "", FeaturesStr)); 301 if (!STI) { 302 errs() << "error: no subtarget info for target " << TripleName << "\n"; 303 return; 304 } 305 306 std::unique_ptr<const MCInstrInfo> MII(TheTarget->createMCInstrInfo()); 307 if (!MII) { 308 errs() << "error: no instruction info for target " << TripleName << "\n"; 309 return; 310 } 311 312 std::unique_ptr<MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI)); 313 if (!DisAsm) { 314 errs() << "error: no disassembler for target " << TripleName << "\n"; 315 return; 316 } 317 318 std::unique_ptr<const MCObjectFileInfo> MOFI; 319 std::unique_ptr<MCContext> Ctx; 320 321 if (Symbolize) { 322 MOFI.reset(new MCObjectFileInfo); 323 Ctx.reset(new MCContext(AsmInfo.get(), MRI.get(), MOFI.get())); 324 std::unique_ptr<MCRelocationInfo> RelInfo( 325 TheTarget->createMCRelocationInfo(TripleName, *Ctx.get())); 326 if (RelInfo) { 327 std::unique_ptr<MCSymbolizer> Symzer( 328 MCObjectSymbolizer::createObjectSymbolizer(*Ctx.get(), RelInfo, Obj)); 329 if (Symzer) 330 DisAsm->setSymbolizer(std::move(Symzer)); 331 } 332 } 333 334 std::unique_ptr<const MCInstrAnalysis> MIA( 335 TheTarget->createMCInstrAnalysis(MII.get())); 336 337 int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); 338 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter( 339 AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI)); 340 if (!IP) { 341 errs() << "error: no instruction printer for target " << TripleName 342 << '\n'; 343 return; 344 } 345 346 if (CFG || !YAMLCFG.empty()) { 347 std::unique_ptr<MCObjectDisassembler> OD( 348 new MCObjectDisassembler(*Obj, *DisAsm, *MIA)); 349 std::unique_ptr<MCModule> Mod(OD->buildModule(/* withCFG */ true)); 350 for (MCModule::const_atom_iterator AI = Mod->atom_begin(), 351 AE = Mod->atom_end(); 352 AI != AE; ++AI) { 353 outs() << "Atom " << (*AI)->getName() << ": \n"; 354 if (const MCTextAtom *TA = dyn_cast<MCTextAtom>(*AI)) { 355 for (MCTextAtom::const_iterator II = TA->begin(), IE = TA->end(); 356 II != IE; 357 ++II) { 358 IP->printInst(&II->Inst, outs(), ""); 359 outs() << "\n"; 360 } 361 } 362 } 363 if (CFG) { 364 for (MCModule::const_func_iterator FI = Mod->func_begin(), 365 FE = Mod->func_end(); 366 FI != FE; ++FI) { 367 static int filenum = 0; 368 emitDOTFile((Twine((*FI)->getName()) + "_" + 369 utostr(filenum) + ".dot").str().c_str(), 370 **FI, IP.get()); 371 ++filenum; 372 } 373 } 374 if (!YAMLCFG.empty()) { 375 std::string Error; 376 raw_fd_ostream YAMLOut(YAMLCFG.c_str(), Error, sys::fs::F_Text); 377 if (!Error.empty()) { 378 errs() << ToolName << ": warning: " << Error << '\n'; 379 return; 380 } 381 mcmodule2yaml(YAMLOut, *Mod, *MII, *MRI); 382 } 383 } 384 385 StringRef Fmt = Obj->getBytesInAddress() > 4 ? "\t\t%016" PRIx64 ": " : 386 "\t\t\t%08" PRIx64 ": "; 387 388 // Create a mapping, RelocSecs = SectionRelocMap[S], where sections 389 // in RelocSecs contain the relocations for section S. 390 error_code EC; 391 std::map<SectionRef, SmallVector<SectionRef, 1>> SectionRelocMap; 392 for (const SectionRef &Section : Obj->sections()) { 393 section_iterator Sec2 = Section.getRelocatedSection(); 394 if (Sec2 != Obj->section_end()) 395 SectionRelocMap[*Sec2].push_back(Section); 396 } 397 398 for (const SectionRef &Section : Obj->sections()) { 399 bool Text; 400 if (error(Section.isText(Text))) 401 break; 402 if (!Text) 403 continue; 404 405 uint64_t SectionAddr; 406 if (error(Section.getAddress(SectionAddr))) 407 break; 408 409 uint64_t SectSize; 410 if (error(Section.getSize(SectSize))) 411 break; 412 413 // Make a list of all the symbols in this section. 414 std::vector<std::pair<uint64_t, StringRef>> Symbols; 415 for (const SymbolRef &Symbol : Obj->symbols()) { 416 bool contains; 417 if (!error(Section.containsSymbol(Symbol, contains)) && contains) { 418 uint64_t Address; 419 if (error(Symbol.getAddress(Address))) 420 break; 421 if (Address == UnknownAddressOrSize) 422 continue; 423 Address -= SectionAddr; 424 if (Address >= SectSize) 425 continue; 426 427 StringRef Name; 428 if (error(Symbol.getName(Name))) 429 break; 430 Symbols.push_back(std::make_pair(Address, Name)); 431 } 432 } 433 434 // Sort the symbols by address, just in case they didn't come in that way. 435 array_pod_sort(Symbols.begin(), Symbols.end()); 436 437 // Make a list of all the relocations for this section. 438 std::vector<RelocationRef> Rels; 439 if (InlineRelocs) { 440 for (const SectionRef &RelocSec : SectionRelocMap[Section]) { 441 for (const RelocationRef &Reloc : RelocSec.relocations()) { 442 Rels.push_back(Reloc); 443 } 444 } 445 } 446 447 // Sort relocations by address. 448 std::sort(Rels.begin(), Rels.end(), RelocAddressLess); 449 450 StringRef SegmentName = ""; 451 if (const MachOObjectFile *MachO = dyn_cast<const MachOObjectFile>(Obj)) { 452 DataRefImpl DR = Section.getRawDataRefImpl(); 453 SegmentName = MachO->getSectionFinalSegmentName(DR); 454 } 455 StringRef name; 456 if (error(Section.getName(name))) 457 break; 458 outs() << "Disassembly of section "; 459 if (!SegmentName.empty()) 460 outs() << SegmentName << ","; 461 outs() << name << ':'; 462 463 // If the section has no symbols just insert a dummy one and disassemble 464 // the whole section. 465 if (Symbols.empty()) 466 Symbols.push_back(std::make_pair(0, name)); 467 468 469 SmallString<40> Comments; 470 raw_svector_ostream CommentStream(Comments); 471 472 StringRef Bytes; 473 if (error(Section.getContents(Bytes))) 474 break; 475 StringRefMemoryObject memoryObject(Bytes, SectionAddr); 476 uint64_t Size; 477 uint64_t Index; 478 479 std::vector<RelocationRef>::const_iterator rel_cur = Rels.begin(); 480 std::vector<RelocationRef>::const_iterator rel_end = Rels.end(); 481 // Disassemble symbol by symbol. 482 for (unsigned si = 0, se = Symbols.size(); si != se; ++si) { 483 uint64_t Start = Symbols[si].first; 484 uint64_t End; 485 // The end is either the size of the section or the beginning of the next 486 // symbol. 487 if (si == se - 1) 488 End = SectSize; 489 // Make sure this symbol takes up space. 490 else if (Symbols[si + 1].first != Start) 491 End = Symbols[si + 1].first - 1; 492 else 493 // This symbol has the same address as the next symbol. Skip it. 494 continue; 495 496 outs() << '\n' << Symbols[si].second << ":\n"; 497 498 #ifndef NDEBUG 499 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls(); 500 #else 501 raw_ostream &DebugOut = nulls(); 502 #endif 503 504 for (Index = Start; Index < End; Index += Size) { 505 MCInst Inst; 506 507 if (DisAsm->getInstruction(Inst, Size, memoryObject, 508 SectionAddr + Index, 509 DebugOut, CommentStream)) { 510 outs() << format("%8" PRIx64 ":", SectionAddr + Index); 511 if (!NoShowRawInsn) { 512 outs() << "\t"; 513 DumpBytes(StringRef(Bytes.data() + Index, Size)); 514 } 515 IP->printInst(&Inst, outs(), ""); 516 outs() << CommentStream.str(); 517 Comments.clear(); 518 outs() << "\n"; 519 } else { 520 errs() << ToolName << ": warning: invalid instruction encoding\n"; 521 if (Size == 0) 522 Size = 1; // skip illegible bytes 523 } 524 525 // Print relocation for instruction. 526 while (rel_cur != rel_end) { 527 bool hidden = false; 528 uint64_t addr; 529 SmallString<16> name; 530 SmallString<32> val; 531 532 // If this relocation is hidden, skip it. 533 if (error(rel_cur->getHidden(hidden))) goto skip_print_rel; 534 if (hidden) goto skip_print_rel; 535 536 if (error(rel_cur->getOffset(addr))) goto skip_print_rel; 537 // Stop when rel_cur's address is past the current instruction. 538 if (addr >= Index + Size) break; 539 if (error(rel_cur->getTypeName(name))) goto skip_print_rel; 540 if (error(rel_cur->getValueString(val))) goto skip_print_rel; 541 542 outs() << format(Fmt.data(), SectionAddr + addr) << name 543 << "\t" << val << "\n"; 544 545 skip_print_rel: 546 ++rel_cur; 547 } 548 } 549 } 550 } 551 } 552 553 static void PrintRelocations(const ObjectFile *Obj) { 554 StringRef Fmt = Obj->getBytesInAddress() > 4 ? "%016" PRIx64 : 555 "%08" PRIx64; 556 for (const SectionRef &Section : Obj->sections()) { 557 if (Section.relocation_begin() == Section.relocation_end()) 558 continue; 559 StringRef secname; 560 if (error(Section.getName(secname))) 561 continue; 562 outs() << "RELOCATION RECORDS FOR [" << secname << "]:\n"; 563 for (const RelocationRef &Reloc : Section.relocations()) { 564 bool hidden; 565 uint64_t address; 566 SmallString<32> relocname; 567 SmallString<32> valuestr; 568 if (error(Reloc.getHidden(hidden))) 569 continue; 570 if (hidden) 571 continue; 572 if (error(Reloc.getTypeName(relocname))) 573 continue; 574 if (error(Reloc.getOffset(address))) 575 continue; 576 if (error(Reloc.getValueString(valuestr))) 577 continue; 578 outs() << format(Fmt.data(), address) << " " << relocname << " " 579 << valuestr << "\n"; 580 } 581 outs() << "\n"; 582 } 583 } 584 585 static void PrintSectionHeaders(const ObjectFile *Obj) { 586 outs() << "Sections:\n" 587 "Idx Name Size Address Type\n"; 588 unsigned i = 0; 589 for (const SectionRef &Section : Obj->sections()) { 590 StringRef Name; 591 if (error(Section.getName(Name))) 592 return; 593 uint64_t Address; 594 if (error(Section.getAddress(Address))) 595 return; 596 uint64_t Size; 597 if (error(Section.getSize(Size))) 598 return; 599 bool Text, Data, BSS; 600 if (error(Section.isText(Text))) 601 return; 602 if (error(Section.isData(Data))) 603 return; 604 if (error(Section.isBSS(BSS))) 605 return; 606 std::string Type = (std::string(Text ? "TEXT " : "") + 607 (Data ? "DATA " : "") + (BSS ? "BSS" : "")); 608 outs() << format("%3d %-13s %08" PRIx64 " %016" PRIx64 " %s\n", i, 609 Name.str().c_str(), Size, Address, Type.c_str()); 610 ++i; 611 } 612 } 613 614 static void PrintSectionContents(const ObjectFile *Obj) { 615 error_code EC; 616 for (const SectionRef &Section : Obj->sections()) { 617 StringRef Name; 618 StringRef Contents; 619 uint64_t BaseAddr; 620 bool BSS; 621 if (error(Section.getName(Name))) 622 continue; 623 if (error(Section.getContents(Contents))) 624 continue; 625 if (error(Section.getAddress(BaseAddr))) 626 continue; 627 if (error(Section.isBSS(BSS))) 628 continue; 629 630 outs() << "Contents of section " << Name << ":\n"; 631 if (BSS) { 632 outs() << format("<skipping contents of bss section at [%04" PRIx64 633 ", %04" PRIx64 ")>\n", BaseAddr, 634 BaseAddr + Contents.size()); 635 continue; 636 } 637 638 // Dump out the content as hex and printable ascii characters. 639 for (std::size_t addr = 0, end = Contents.size(); addr < end; addr += 16) { 640 outs() << format(" %04" PRIx64 " ", BaseAddr + addr); 641 // Dump line of hex. 642 for (std::size_t i = 0; i < 16; ++i) { 643 if (i != 0 && i % 4 == 0) 644 outs() << ' '; 645 if (addr + i < end) 646 outs() << hexdigit((Contents[addr + i] >> 4) & 0xF, true) 647 << hexdigit(Contents[addr + i] & 0xF, true); 648 else 649 outs() << " "; 650 } 651 // Print ascii. 652 outs() << " "; 653 for (std::size_t i = 0; i < 16 && addr + i < end; ++i) { 654 if (std::isprint(static_cast<unsigned char>(Contents[addr + i]) & 0xFF)) 655 outs() << Contents[addr + i]; 656 else 657 outs() << "."; 658 } 659 outs() << "\n"; 660 } 661 } 662 } 663 664 static void PrintCOFFSymbolTable(const COFFObjectFile *coff) { 665 const coff_file_header *header; 666 if (error(coff->getHeader(header))) return; 667 int aux_count = 0; 668 const coff_symbol *symbol = 0; 669 for (int i = 0, e = header->NumberOfSymbols; i != e; ++i) { 670 if (aux_count--) { 671 // Figure out which type of aux this is. 672 if (symbol->isSectionDefinition()) { // Section definition. 673 const coff_aux_section_definition *asd; 674 if (error(coff->getAuxSymbol<coff_aux_section_definition>(i, asd))) 675 return; 676 outs() << "AUX " 677 << format("scnlen 0x%x nreloc %d nlnno %d checksum 0x%x " 678 , unsigned(asd->Length) 679 , unsigned(asd->NumberOfRelocations) 680 , unsigned(asd->NumberOfLinenumbers) 681 , unsigned(asd->CheckSum)) 682 << format("assoc %d comdat %d\n" 683 , unsigned(asd->Number) 684 , unsigned(asd->Selection)); 685 } else 686 outs() << "AUX Unknown\n"; 687 } else { 688 StringRef name; 689 if (error(coff->getSymbol(i, symbol))) return; 690 if (error(coff->getSymbolName(symbol, name))) return; 691 outs() << "[" << format("%2d", i) << "]" 692 << "(sec " << format("%2d", int(symbol->SectionNumber)) << ")" 693 << "(fl 0x00)" // Flag bits, which COFF doesn't have. 694 << "(ty " << format("%3x", unsigned(symbol->Type)) << ")" 695 << "(scl " << format("%3x", unsigned(symbol->StorageClass)) << ") " 696 << "(nx " << unsigned(symbol->NumberOfAuxSymbols) << ") " 697 << "0x" << format("%08x", unsigned(symbol->Value)) << " " 698 << name << "\n"; 699 aux_count = symbol->NumberOfAuxSymbols; 700 } 701 } 702 } 703 704 static void PrintSymbolTable(const ObjectFile *o) { 705 outs() << "SYMBOL TABLE:\n"; 706 707 if (const COFFObjectFile *coff = dyn_cast<const COFFObjectFile>(o)) { 708 PrintCOFFSymbolTable(coff); 709 return; 710 } 711 for (const SymbolRef &Symbol : o->symbols()) { 712 StringRef Name; 713 uint64_t Address; 714 SymbolRef::Type Type; 715 uint64_t Size; 716 uint32_t Flags = Symbol.getFlags(); 717 section_iterator Section = o->section_end(); 718 if (error(Symbol.getName(Name))) 719 continue; 720 if (error(Symbol.getAddress(Address))) 721 continue; 722 if (error(Symbol.getType(Type))) 723 continue; 724 if (error(Symbol.getSize(Size))) 725 continue; 726 if (error(Symbol.getSection(Section))) 727 continue; 728 729 bool Global = Flags & SymbolRef::SF_Global; 730 bool Weak = Flags & SymbolRef::SF_Weak; 731 bool Absolute = Flags & SymbolRef::SF_Absolute; 732 733 if (Address == UnknownAddressOrSize) 734 Address = 0; 735 if (Size == UnknownAddressOrSize) 736 Size = 0; 737 char GlobLoc = ' '; 738 if (Type != SymbolRef::ST_Unknown) 739 GlobLoc = Global ? 'g' : 'l'; 740 char Debug = (Type == SymbolRef::ST_Debug || Type == SymbolRef::ST_File) 741 ? 'd' : ' '; 742 char FileFunc = ' '; 743 if (Type == SymbolRef::ST_File) 744 FileFunc = 'f'; 745 else if (Type == SymbolRef::ST_Function) 746 FileFunc = 'F'; 747 748 const char *Fmt = o->getBytesInAddress() > 4 ? "%016" PRIx64 : 749 "%08" PRIx64; 750 751 outs() << format(Fmt, Address) << " " 752 << GlobLoc // Local -> 'l', Global -> 'g', Neither -> ' ' 753 << (Weak ? 'w' : ' ') // Weak? 754 << ' ' // Constructor. Not supported yet. 755 << ' ' // Warning. Not supported yet. 756 << ' ' // Indirect reference to another symbol. 757 << Debug // Debugging (d) or dynamic (D) symbol. 758 << FileFunc // Name of function (F), file (f) or object (O). 759 << ' '; 760 if (Absolute) { 761 outs() << "*ABS*"; 762 } else if (Section == o->section_end()) { 763 outs() << "*UND*"; 764 } else { 765 if (const MachOObjectFile *MachO = 766 dyn_cast<const MachOObjectFile>(o)) { 767 DataRefImpl DR = Section->getRawDataRefImpl(); 768 StringRef SegmentName = MachO->getSectionFinalSegmentName(DR); 769 outs() << SegmentName << ","; 770 } 771 StringRef SectionName; 772 if (error(Section->getName(SectionName))) 773 SectionName = ""; 774 outs() << SectionName; 775 } 776 outs() << '\t' 777 << format("%08" PRIx64 " ", Size) 778 << Name 779 << '\n'; 780 } 781 } 782 783 static void PrintUnwindInfo(const ObjectFile *o) { 784 outs() << "Unwind info:\n\n"; 785 786 if (const COFFObjectFile *coff = dyn_cast<COFFObjectFile>(o)) { 787 printCOFFUnwindInfo(coff); 788 } else { 789 // TODO: Extract DWARF dump tool to objdump. 790 errs() << "This operation is only currently supported " 791 "for COFF object files.\n"; 792 return; 793 } 794 } 795 796 static void printPrivateFileHeader(const ObjectFile *o) { 797 if (o->isELF()) { 798 printELFFileHeader(o); 799 } else if (o->isCOFF()) { 800 printCOFFFileHeader(o); 801 } 802 } 803 804 static void DumpObject(const ObjectFile *o) { 805 outs() << '\n'; 806 outs() << o->getFileName() 807 << ":\tfile format " << o->getFileFormatName() << "\n\n"; 808 809 if (Disassemble) 810 DisassembleObject(o, Relocations); 811 if (Relocations && !Disassemble) 812 PrintRelocations(o); 813 if (SectionHeaders) 814 PrintSectionHeaders(o); 815 if (SectionContents) 816 PrintSectionContents(o); 817 if (SymbolTable) 818 PrintSymbolTable(o); 819 if (UnwindInfo) 820 PrintUnwindInfo(o); 821 if (PrivateHeaders) 822 printPrivateFileHeader(o); 823 } 824 825 /// @brief Dump each object file in \a a; 826 static void DumpArchive(const Archive *a) { 827 for (Archive::child_iterator i = a->child_begin(), e = a->child_end(); i != e; 828 ++i) { 829 std::unique_ptr<Binary> child; 830 if (error_code EC = i->getAsBinary(child)) { 831 // Ignore non-object files. 832 if (EC != object_error::invalid_file_type) 833 errs() << ToolName << ": '" << a->getFileName() << "': " << EC.message() 834 << ".\n"; 835 continue; 836 } 837 if (ObjectFile *o = dyn_cast<ObjectFile>(child.get())) 838 DumpObject(o); 839 else 840 errs() << ToolName << ": '" << a->getFileName() << "': " 841 << "Unrecognized file type.\n"; 842 } 843 } 844 845 /// @brief Open file and figure out how to dump it. 846 static void DumpInput(StringRef file) { 847 // If file isn't stdin, check that it exists. 848 if (file != "-" && !sys::fs::exists(file)) { 849 errs() << ToolName << ": '" << file << "': " << "No such file\n"; 850 return; 851 } 852 853 if (MachOOpt && Disassemble) { 854 DisassembleInputMachO(file); 855 return; 856 } 857 858 // Attempt to open the binary. 859 ErrorOr<Binary *> BinaryOrErr = createBinary(file); 860 if (error_code EC = BinaryOrErr.getError()) { 861 errs() << ToolName << ": '" << file << "': " << EC.message() << ".\n"; 862 return; 863 } 864 std::unique_ptr<Binary> binary(BinaryOrErr.get()); 865 866 if (Archive *a = dyn_cast<Archive>(binary.get())) 867 DumpArchive(a); 868 else if (ObjectFile *o = dyn_cast<ObjectFile>(binary.get())) 869 DumpObject(o); 870 else 871 errs() << ToolName << ": '" << file << "': " << "Unrecognized file type.\n"; 872 } 873 874 int main(int argc, char **argv) { 875 // Print a stack trace if we signal out. 876 sys::PrintStackTraceOnErrorSignal(); 877 PrettyStackTraceProgram X(argc, argv); 878 llvm_shutdown_obj Y; // Call llvm_shutdown() on exit. 879 880 // Initialize targets and assembly printers/parsers. 881 llvm::InitializeAllTargetInfos(); 882 llvm::InitializeAllTargetMCs(); 883 llvm::InitializeAllAsmParsers(); 884 llvm::InitializeAllDisassemblers(); 885 886 // Register the target printer for --version. 887 cl::AddExtraVersionPrinter(TargetRegistry::printRegisteredTargetsForVersion); 888 889 cl::ParseCommandLineOptions(argc, argv, "llvm object file dumper\n"); 890 TripleName = Triple::normalize(TripleName); 891 892 ToolName = argv[0]; 893 894 // Defaults to a.out if no filenames specified. 895 if (InputFilenames.size() == 0) 896 InputFilenames.push_back("a.out"); 897 898 if (!Disassemble 899 && !Relocations 900 && !SectionHeaders 901 && !SectionContents 902 && !SymbolTable 903 && !UnwindInfo 904 && !PrivateHeaders) { 905 cl::PrintHelpMessage(); 906 return 2; 907 } 908 909 std::for_each(InputFilenames.begin(), InputFilenames.end(), 910 DumpInput); 911 912 return 0; 913 } 914