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/MCAnalysis/MCAtom.h" 24 #include "llvm/MC/MCAnalysis/MCFunction.h" 25 #include "llvm/MC/MCAnalysis/MCModule.h" 26 #include "llvm/MC/MCAnalysis/MCModuleYAML.h" 27 #include "llvm/MC/MCAsmInfo.h" 28 #include "llvm/MC/MCContext.h" 29 #include "llvm/MC/MCDisassembler.h" 30 #include "llvm/MC/MCInst.h" 31 #include "llvm/MC/MCInstPrinter.h" 32 #include "llvm/MC/MCInstrAnalysis.h" 33 #include "llvm/MC/MCInstrInfo.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 <algorithm> 61 #include <cctype> 62 #include <cstring> 63 #include <system_error> 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(std::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 = nullptr) { 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 if (Obj->isCOFF()) { 172 const auto COFFObj = dyn_cast<COFFObjectFile>(Obj); 173 if (COFFObj->getArch() == Triple::thumb) 174 TheTriple.setTriple("thumbv7-windows"); 175 } 176 } 177 } else 178 TheTriple.setTriple(Triple::normalize(TripleName)); 179 180 // Get the target specific parser. 181 std::string Error; 182 const Target *TheTarget = TargetRegistry::lookupTarget(ArchName, TheTriple, 183 Error); 184 if (!TheTarget) { 185 errs() << ToolName << ": " << Error; 186 return nullptr; 187 } 188 189 // Update the triple name and return the found target. 190 TripleName = TheTriple.getTriple(); 191 return TheTarget; 192 } 193 194 // Write a graphviz file for the CFG inside an MCFunction. 195 // FIXME: Use GraphWriter 196 static void emitDOTFile(const char *FileName, const MCFunction &f, 197 MCInstPrinter *IP) { 198 // Start a new dot file. 199 std::string Error; 200 raw_fd_ostream Out(FileName, Error, sys::fs::F_Text); 201 if (!Error.empty()) { 202 errs() << "llvm-objdump: warning: " << Error << '\n'; 203 return; 204 } 205 206 Out << "digraph \"" << f.getName() << "\" {\n"; 207 Out << "graph [ rankdir = \"LR\" ];\n"; 208 for (MCFunction::const_iterator i = f.begin(), e = f.end(); i != e; ++i) { 209 // Only print blocks that have predecessors. 210 bool hasPreds = (*i)->pred_begin() != (*i)->pred_end(); 211 212 if (!hasPreds && i != f.begin()) 213 continue; 214 215 Out << '"' << (*i)->getInsts()->getBeginAddr() << "\" [ label=\"<a>"; 216 // Print instructions. 217 for (unsigned ii = 0, ie = (*i)->getInsts()->size(); ii != ie; 218 ++ii) { 219 if (ii != 0) // Not the first line, start a new row. 220 Out << '|'; 221 if (ii + 1 == ie) // Last line, add an end id. 222 Out << "<o>"; 223 224 // Escape special chars and print the instruction in mnemonic form. 225 std::string Str; 226 raw_string_ostream OS(Str); 227 IP->printInst(&(*i)->getInsts()->at(ii).Inst, OS, ""); 228 Out << DOT::EscapeString(OS.str()); 229 } 230 Out << "\" shape=\"record\" ];\n"; 231 232 // Add edges. 233 for (MCBasicBlock::succ_const_iterator si = (*i)->succ_begin(), 234 se = (*i)->succ_end(); si != se; ++si) 235 Out << (*i)->getInsts()->getBeginAddr() << ":o -> " 236 << (*si)->getInsts()->getBeginAddr() << ":a\n"; 237 } 238 Out << "}\n"; 239 } 240 241 void llvm::DumpBytes(StringRef bytes) { 242 static const char hex_rep[] = "0123456789abcdef"; 243 // FIXME: The real way to do this is to figure out the longest instruction 244 // and align to that size before printing. I'll fix this when I get 245 // around to outputting relocations. 246 // 15 is the longest x86 instruction 247 // 3 is for the hex rep of a byte + a space. 248 // 1 is for the null terminator. 249 enum { OutputSize = (15 * 3) + 1 }; 250 char output[OutputSize]; 251 252 assert(bytes.size() <= 15 253 && "DumpBytes only supports instructions of up to 15 bytes"); 254 memset(output, ' ', sizeof(output)); 255 unsigned index = 0; 256 for (StringRef::iterator i = bytes.begin(), 257 e = bytes.end(); i != e; ++i) { 258 output[index] = hex_rep[(*i & 0xF0) >> 4]; 259 output[index + 1] = hex_rep[*i & 0xF]; 260 index += 3; 261 } 262 263 output[sizeof(output) - 1] = 0; 264 outs() << output; 265 } 266 267 bool llvm::RelocAddressLess(RelocationRef a, RelocationRef b) { 268 uint64_t a_addr, b_addr; 269 if (error(a.getOffset(a_addr))) return false; 270 if (error(b.getOffset(b_addr))) return false; 271 return a_addr < b_addr; 272 } 273 274 static void DisassembleObject(const ObjectFile *Obj, bool InlineRelocs) { 275 const Target *TheTarget = getTarget(Obj); 276 // getTarget() will have already issued a diagnostic if necessary, so 277 // just bail here if it failed. 278 if (!TheTarget) 279 return; 280 281 // Package up features to be passed to target/subtarget 282 std::string FeaturesStr; 283 if (MAttrs.size()) { 284 SubtargetFeatures Features; 285 for (unsigned i = 0; i != MAttrs.size(); ++i) 286 Features.AddFeature(MAttrs[i]); 287 FeaturesStr = Features.getString(); 288 } 289 290 std::unique_ptr<const MCRegisterInfo> MRI( 291 TheTarget->createMCRegInfo(TripleName)); 292 if (!MRI) { 293 errs() << "error: no register info for target " << TripleName << "\n"; 294 return; 295 } 296 297 // Set up disassembler. 298 std::unique_ptr<const MCAsmInfo> AsmInfo( 299 TheTarget->createMCAsmInfo(*MRI, TripleName)); 300 if (!AsmInfo) { 301 errs() << "error: no assembly info for target " << TripleName << "\n"; 302 return; 303 } 304 305 std::unique_ptr<const MCSubtargetInfo> STI( 306 TheTarget->createMCSubtargetInfo(TripleName, "", FeaturesStr)); 307 if (!STI) { 308 errs() << "error: no subtarget info for target " << TripleName << "\n"; 309 return; 310 } 311 312 std::unique_ptr<const MCInstrInfo> MII(TheTarget->createMCInstrInfo()); 313 if (!MII) { 314 errs() << "error: no instruction info for target " << TripleName << "\n"; 315 return; 316 } 317 318 std::unique_ptr<const MCObjectFileInfo> MOFI(new MCObjectFileInfo); 319 MCContext Ctx(AsmInfo.get(), MRI.get(), MOFI.get()); 320 321 std::unique_ptr<MCDisassembler> DisAsm( 322 TheTarget->createMCDisassembler(*STI, Ctx)); 323 324 if (!DisAsm) { 325 errs() << "error: no disassembler for target " << TripleName << "\n"; 326 return; 327 } 328 329 330 if (Symbolize) { 331 std::unique_ptr<MCRelocationInfo> RelInfo( 332 TheTarget->createMCRelocationInfo(TripleName, Ctx)); 333 if (RelInfo) { 334 std::unique_ptr<MCSymbolizer> Symzer( 335 MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), 336 Obj)); 337 if (Symzer) 338 DisAsm->setSymbolizer(std::move(Symzer)); 339 } 340 } 341 342 std::unique_ptr<const MCInstrAnalysis> MIA( 343 TheTarget->createMCInstrAnalysis(MII.get())); 344 345 int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); 346 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter( 347 AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI)); 348 if (!IP) { 349 errs() << "error: no instruction printer for target " << TripleName 350 << '\n'; 351 return; 352 } 353 354 if (CFG || !YAMLCFG.empty()) { 355 std::unique_ptr<MCObjectDisassembler> OD( 356 new MCObjectDisassembler(*Obj, *DisAsm, *MIA)); 357 std::unique_ptr<MCModule> Mod(OD->buildModule(/* withCFG */ true)); 358 for (MCModule::const_atom_iterator AI = Mod->atom_begin(), 359 AE = Mod->atom_end(); 360 AI != AE; ++AI) { 361 outs() << "Atom " << (*AI)->getName() << ": \n"; 362 if (const MCTextAtom *TA = dyn_cast<MCTextAtom>(*AI)) { 363 for (MCTextAtom::const_iterator II = TA->begin(), IE = TA->end(); 364 II != IE; 365 ++II) { 366 IP->printInst(&II->Inst, outs(), ""); 367 outs() << "\n"; 368 } 369 } 370 } 371 if (CFG) { 372 for (MCModule::const_func_iterator FI = Mod->func_begin(), 373 FE = Mod->func_end(); 374 FI != FE; ++FI) { 375 static int filenum = 0; 376 emitDOTFile((Twine((*FI)->getName()) + "_" + 377 utostr(filenum) + ".dot").str().c_str(), 378 **FI, IP.get()); 379 ++filenum; 380 } 381 } 382 if (!YAMLCFG.empty()) { 383 std::string Error; 384 raw_fd_ostream YAMLOut(YAMLCFG.c_str(), Error, sys::fs::F_Text); 385 if (!Error.empty()) { 386 errs() << ToolName << ": warning: " << Error << '\n'; 387 return; 388 } 389 mcmodule2yaml(YAMLOut, *Mod, *MII, *MRI); 390 } 391 } 392 393 StringRef Fmt = Obj->getBytesInAddress() > 4 ? "\t\t%016" PRIx64 ": " : 394 "\t\t\t%08" PRIx64 ": "; 395 396 // Create a mapping, RelocSecs = SectionRelocMap[S], where sections 397 // in RelocSecs contain the relocations for section S. 398 std::error_code EC; 399 std::map<SectionRef, SmallVector<SectionRef, 1>> SectionRelocMap; 400 for (const SectionRef &Section : Obj->sections()) { 401 section_iterator Sec2 = Section.getRelocatedSection(); 402 if (Sec2 != Obj->section_end()) 403 SectionRelocMap[*Sec2].push_back(Section); 404 } 405 406 for (const SectionRef &Section : Obj->sections()) { 407 bool Text; 408 if (error(Section.isText(Text))) 409 break; 410 if (!Text) 411 continue; 412 413 uint64_t SectionAddr; 414 if (error(Section.getAddress(SectionAddr))) 415 break; 416 417 uint64_t SectSize; 418 if (error(Section.getSize(SectSize))) 419 break; 420 421 // Make a list of all the symbols in this section. 422 std::vector<std::pair<uint64_t, StringRef>> Symbols; 423 for (const SymbolRef &Symbol : Obj->symbols()) { 424 bool contains; 425 if (!error(Section.containsSymbol(Symbol, contains)) && contains) { 426 uint64_t Address; 427 if (error(Symbol.getAddress(Address))) 428 break; 429 if (Address == UnknownAddressOrSize) 430 continue; 431 Address -= SectionAddr; 432 if (Address >= SectSize) 433 continue; 434 435 StringRef Name; 436 if (error(Symbol.getName(Name))) 437 break; 438 Symbols.push_back(std::make_pair(Address, Name)); 439 } 440 } 441 442 // Sort the symbols by address, just in case they didn't come in that way. 443 array_pod_sort(Symbols.begin(), Symbols.end()); 444 445 // Make a list of all the relocations for this section. 446 std::vector<RelocationRef> Rels; 447 if (InlineRelocs) { 448 for (const SectionRef &RelocSec : SectionRelocMap[Section]) { 449 for (const RelocationRef &Reloc : RelocSec.relocations()) { 450 Rels.push_back(Reloc); 451 } 452 } 453 } 454 455 // Sort relocations by address. 456 std::sort(Rels.begin(), Rels.end(), RelocAddressLess); 457 458 StringRef SegmentName = ""; 459 if (const MachOObjectFile *MachO = dyn_cast<const MachOObjectFile>(Obj)) { 460 DataRefImpl DR = Section.getRawDataRefImpl(); 461 SegmentName = MachO->getSectionFinalSegmentName(DR); 462 } 463 StringRef name; 464 if (error(Section.getName(name))) 465 break; 466 outs() << "Disassembly of section "; 467 if (!SegmentName.empty()) 468 outs() << SegmentName << ","; 469 outs() << name << ':'; 470 471 // If the section has no symbols just insert a dummy one and disassemble 472 // the whole section. 473 if (Symbols.empty()) 474 Symbols.push_back(std::make_pair(0, name)); 475 476 477 SmallString<40> Comments; 478 raw_svector_ostream CommentStream(Comments); 479 480 StringRef Bytes; 481 if (error(Section.getContents(Bytes))) 482 break; 483 StringRefMemoryObject memoryObject(Bytes, SectionAddr); 484 uint64_t Size; 485 uint64_t Index; 486 487 std::vector<RelocationRef>::const_iterator rel_cur = Rels.begin(); 488 std::vector<RelocationRef>::const_iterator rel_end = Rels.end(); 489 // Disassemble symbol by symbol. 490 for (unsigned si = 0, se = Symbols.size(); si != se; ++si) { 491 uint64_t Start = Symbols[si].first; 492 uint64_t End; 493 // The end is either the size of the section or the beginning of the next 494 // symbol. 495 if (si == se - 1) 496 End = SectSize; 497 // Make sure this symbol takes up space. 498 else if (Symbols[si + 1].first != Start) 499 End = Symbols[si + 1].first - 1; 500 else 501 // This symbol has the same address as the next symbol. Skip it. 502 continue; 503 504 outs() << '\n' << Symbols[si].second << ":\n"; 505 506 #ifndef NDEBUG 507 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls(); 508 #else 509 raw_ostream &DebugOut = nulls(); 510 #endif 511 512 for (Index = Start; Index < End; Index += Size) { 513 MCInst Inst; 514 515 if (DisAsm->getInstruction(Inst, Size, memoryObject, 516 SectionAddr + Index, 517 DebugOut, CommentStream)) { 518 outs() << format("%8" PRIx64 ":", SectionAddr + Index); 519 if (!NoShowRawInsn) { 520 outs() << "\t"; 521 DumpBytes(StringRef(Bytes.data() + Index, Size)); 522 } 523 IP->printInst(&Inst, outs(), ""); 524 outs() << CommentStream.str(); 525 Comments.clear(); 526 outs() << "\n"; 527 } else { 528 errs() << ToolName << ": warning: invalid instruction encoding\n"; 529 if (Size == 0) 530 Size = 1; // skip illegible bytes 531 } 532 533 // Print relocation for instruction. 534 while (rel_cur != rel_end) { 535 bool hidden = false; 536 uint64_t addr; 537 SmallString<16> name; 538 SmallString<32> val; 539 540 // If this relocation is hidden, skip it. 541 if (error(rel_cur->getHidden(hidden))) goto skip_print_rel; 542 if (hidden) goto skip_print_rel; 543 544 if (error(rel_cur->getOffset(addr))) goto skip_print_rel; 545 // Stop when rel_cur's address is past the current instruction. 546 if (addr >= Index + Size) break; 547 if (error(rel_cur->getTypeName(name))) goto skip_print_rel; 548 if (error(rel_cur->getValueString(val))) goto skip_print_rel; 549 550 outs() << format(Fmt.data(), SectionAddr + addr) << name 551 << "\t" << val << "\n"; 552 553 skip_print_rel: 554 ++rel_cur; 555 } 556 } 557 } 558 } 559 } 560 561 static void PrintRelocations(const ObjectFile *Obj) { 562 StringRef Fmt = Obj->getBytesInAddress() > 4 ? "%016" PRIx64 : 563 "%08" PRIx64; 564 for (const SectionRef &Section : Obj->sections()) { 565 if (Section.relocation_begin() == Section.relocation_end()) 566 continue; 567 StringRef secname; 568 if (error(Section.getName(secname))) 569 continue; 570 outs() << "RELOCATION RECORDS FOR [" << secname << "]:\n"; 571 for (const RelocationRef &Reloc : Section.relocations()) { 572 bool hidden; 573 uint64_t address; 574 SmallString<32> relocname; 575 SmallString<32> valuestr; 576 if (error(Reloc.getHidden(hidden))) 577 continue; 578 if (hidden) 579 continue; 580 if (error(Reloc.getTypeName(relocname))) 581 continue; 582 if (error(Reloc.getOffset(address))) 583 continue; 584 if (error(Reloc.getValueString(valuestr))) 585 continue; 586 outs() << format(Fmt.data(), address) << " " << relocname << " " 587 << valuestr << "\n"; 588 } 589 outs() << "\n"; 590 } 591 } 592 593 static void PrintSectionHeaders(const ObjectFile *Obj) { 594 outs() << "Sections:\n" 595 "Idx Name Size Address Type\n"; 596 unsigned i = 0; 597 for (const SectionRef &Section : Obj->sections()) { 598 StringRef Name; 599 if (error(Section.getName(Name))) 600 return; 601 uint64_t Address; 602 if (error(Section.getAddress(Address))) 603 return; 604 uint64_t Size; 605 if (error(Section.getSize(Size))) 606 return; 607 bool Text, Data, BSS; 608 if (error(Section.isText(Text))) 609 return; 610 if (error(Section.isData(Data))) 611 return; 612 if (error(Section.isBSS(BSS))) 613 return; 614 std::string Type = (std::string(Text ? "TEXT " : "") + 615 (Data ? "DATA " : "") + (BSS ? "BSS" : "")); 616 outs() << format("%3d %-13s %08" PRIx64 " %016" PRIx64 " %s\n", i, 617 Name.str().c_str(), Size, Address, Type.c_str()); 618 ++i; 619 } 620 } 621 622 static void PrintSectionContents(const ObjectFile *Obj) { 623 std::error_code EC; 624 for (const SectionRef &Section : Obj->sections()) { 625 StringRef Name; 626 StringRef Contents; 627 uint64_t BaseAddr; 628 bool BSS; 629 if (error(Section.getName(Name))) 630 continue; 631 if (error(Section.getAddress(BaseAddr))) 632 continue; 633 if (error(Section.isBSS(BSS))) 634 continue; 635 636 outs() << "Contents of section " << Name << ":\n"; 637 if (BSS) { 638 uint64_t Size; 639 if (error(Section.getSize(Size))) 640 continue; 641 outs() << format("<skipping contents of bss section at [%04" PRIx64 642 ", %04" PRIx64 ")>\n", 643 BaseAddr, BaseAddr + Size); 644 continue; 645 } 646 647 if (error(Section.getContents(Contents))) 648 continue; 649 650 // Dump out the content as hex and printable ascii characters. 651 for (std::size_t addr = 0, end = Contents.size(); addr < end; addr += 16) { 652 outs() << format(" %04" PRIx64 " ", BaseAddr + addr); 653 // Dump line of hex. 654 for (std::size_t i = 0; i < 16; ++i) { 655 if (i != 0 && i % 4 == 0) 656 outs() << ' '; 657 if (addr + i < end) 658 outs() << hexdigit((Contents[addr + i] >> 4) & 0xF, true) 659 << hexdigit(Contents[addr + i] & 0xF, true); 660 else 661 outs() << " "; 662 } 663 // Print ascii. 664 outs() << " "; 665 for (std::size_t i = 0; i < 16 && addr + i < end; ++i) { 666 if (std::isprint(static_cast<unsigned char>(Contents[addr + i]) & 0xFF)) 667 outs() << Contents[addr + i]; 668 else 669 outs() << "."; 670 } 671 outs() << "\n"; 672 } 673 } 674 } 675 676 static void PrintCOFFSymbolTable(const COFFObjectFile *coff) { 677 const coff_file_header *header; 678 if (error(coff->getHeader(header))) 679 return; 680 681 for (unsigned SI = 0, SE = header->NumberOfSymbols; SI != SE; ++SI) { 682 const coff_symbol *Symbol; 683 StringRef Name; 684 if (error(coff->getSymbol(SI, Symbol))) 685 return; 686 687 if (error(coff->getSymbolName(Symbol, Name))) 688 return; 689 690 outs() << "[" << format("%2d", SI) << "]" 691 << "(sec " << format("%2d", int(Symbol->SectionNumber)) << ")" 692 << "(fl 0x00)" // Flag bits, which COFF doesn't have. 693 << "(ty " << format("%3x", unsigned(Symbol->Type)) << ")" 694 << "(scl " << format("%3x", unsigned(Symbol->StorageClass)) << ") " 695 << "(nx " << unsigned(Symbol->NumberOfAuxSymbols) << ") " 696 << "0x" << format("%08x", unsigned(Symbol->Value)) << " " 697 << Name << "\n"; 698 699 for (unsigned AI = 0, AE = Symbol->NumberOfAuxSymbols; AI < AE; ++AI, ++SI) { 700 if (Symbol->isSectionDefinition()) { 701 const coff_aux_section_definition *asd; 702 if (error(coff->getAuxSymbol<coff_aux_section_definition>(SI + 1, asd))) 703 return; 704 705 outs() << "AUX " 706 << format("scnlen 0x%x nreloc %d nlnno %d checksum 0x%x " 707 , unsigned(asd->Length) 708 , unsigned(asd->NumberOfRelocations) 709 , unsigned(asd->NumberOfLinenumbers) 710 , unsigned(asd->CheckSum)) 711 << format("assoc %d comdat %d\n" 712 , unsigned(asd->Number) 713 , unsigned(asd->Selection)); 714 } else if (Symbol->isFileRecord()) { 715 const coff_aux_file *AF; 716 if (error(coff->getAuxSymbol<coff_aux_file>(SI + 1, AF))) 717 return; 718 719 StringRef Name(AF->FileName, 720 Symbol->NumberOfAuxSymbols * COFF::SymbolSize); 721 outs() << "AUX " << Name.rtrim(StringRef("\0", 1)) << '\n'; 722 723 SI = SI + Symbol->NumberOfAuxSymbols; 724 break; 725 } else { 726 outs() << "AUX Unknown\n"; 727 } 728 } 729 } 730 } 731 732 static void PrintSymbolTable(const ObjectFile *o) { 733 outs() << "SYMBOL TABLE:\n"; 734 735 if (const COFFObjectFile *coff = dyn_cast<const COFFObjectFile>(o)) { 736 PrintCOFFSymbolTable(coff); 737 return; 738 } 739 for (const SymbolRef &Symbol : o->symbols()) { 740 StringRef Name; 741 uint64_t Address; 742 SymbolRef::Type Type; 743 uint64_t Size; 744 uint32_t Flags = Symbol.getFlags(); 745 section_iterator Section = o->section_end(); 746 if (error(Symbol.getName(Name))) 747 continue; 748 if (error(Symbol.getAddress(Address))) 749 continue; 750 if (error(Symbol.getType(Type))) 751 continue; 752 if (error(Symbol.getSize(Size))) 753 continue; 754 if (error(Symbol.getSection(Section))) 755 continue; 756 757 bool Global = Flags & SymbolRef::SF_Global; 758 bool Weak = Flags & SymbolRef::SF_Weak; 759 bool Absolute = Flags & SymbolRef::SF_Absolute; 760 761 if (Address == UnknownAddressOrSize) 762 Address = 0; 763 if (Size == UnknownAddressOrSize) 764 Size = 0; 765 char GlobLoc = ' '; 766 if (Type != SymbolRef::ST_Unknown) 767 GlobLoc = Global ? 'g' : 'l'; 768 char Debug = (Type == SymbolRef::ST_Debug || Type == SymbolRef::ST_File) 769 ? 'd' : ' '; 770 char FileFunc = ' '; 771 if (Type == SymbolRef::ST_File) 772 FileFunc = 'f'; 773 else if (Type == SymbolRef::ST_Function) 774 FileFunc = 'F'; 775 776 const char *Fmt = o->getBytesInAddress() > 4 ? "%016" PRIx64 : 777 "%08" PRIx64; 778 779 outs() << format(Fmt, Address) << " " 780 << GlobLoc // Local -> 'l', Global -> 'g', Neither -> ' ' 781 << (Weak ? 'w' : ' ') // Weak? 782 << ' ' // Constructor. Not supported yet. 783 << ' ' // Warning. Not supported yet. 784 << ' ' // Indirect reference to another symbol. 785 << Debug // Debugging (d) or dynamic (D) symbol. 786 << FileFunc // Name of function (F), file (f) or object (O). 787 << ' '; 788 if (Absolute) { 789 outs() << "*ABS*"; 790 } else if (Section == o->section_end()) { 791 outs() << "*UND*"; 792 } else { 793 if (const MachOObjectFile *MachO = 794 dyn_cast<const MachOObjectFile>(o)) { 795 DataRefImpl DR = Section->getRawDataRefImpl(); 796 StringRef SegmentName = MachO->getSectionFinalSegmentName(DR); 797 outs() << SegmentName << ","; 798 } 799 StringRef SectionName; 800 if (error(Section->getName(SectionName))) 801 SectionName = ""; 802 outs() << SectionName; 803 } 804 outs() << '\t' 805 << format("%08" PRIx64 " ", Size) 806 << Name 807 << '\n'; 808 } 809 } 810 811 static void PrintUnwindInfo(const ObjectFile *o) { 812 outs() << "Unwind info:\n\n"; 813 814 if (const COFFObjectFile *coff = dyn_cast<COFFObjectFile>(o)) { 815 printCOFFUnwindInfo(coff); 816 } else { 817 // TODO: Extract DWARF dump tool to objdump. 818 errs() << "This operation is only currently supported " 819 "for COFF object files.\n"; 820 return; 821 } 822 } 823 824 static void printPrivateFileHeader(const ObjectFile *o) { 825 if (o->isELF()) { 826 printELFFileHeader(o); 827 } else if (o->isCOFF()) { 828 printCOFFFileHeader(o); 829 } 830 } 831 832 static void DumpObject(const ObjectFile *o) { 833 outs() << '\n'; 834 outs() << o->getFileName() 835 << ":\tfile format " << o->getFileFormatName() << "\n\n"; 836 837 if (Disassemble) 838 DisassembleObject(o, Relocations); 839 if (Relocations && !Disassemble) 840 PrintRelocations(o); 841 if (SectionHeaders) 842 PrintSectionHeaders(o); 843 if (SectionContents) 844 PrintSectionContents(o); 845 if (SymbolTable) 846 PrintSymbolTable(o); 847 if (UnwindInfo) 848 PrintUnwindInfo(o); 849 if (PrivateHeaders) 850 printPrivateFileHeader(o); 851 } 852 853 /// @brief Dump each object file in \a a; 854 static void DumpArchive(const Archive *a) { 855 for (Archive::child_iterator i = a->child_begin(), e = a->child_end(); i != e; 856 ++i) { 857 ErrorOr<std::unique_ptr<Binary>> ChildOrErr = i->getAsBinary(); 858 if (std::error_code EC = ChildOrErr.getError()) { 859 // Ignore non-object files. 860 if (EC != object_error::invalid_file_type) 861 errs() << ToolName << ": '" << a->getFileName() << "': " << EC.message() 862 << ".\n"; 863 continue; 864 } 865 if (ObjectFile *o = dyn_cast<ObjectFile>(&*ChildOrErr.get())) 866 DumpObject(o); 867 else 868 errs() << ToolName << ": '" << a->getFileName() << "': " 869 << "Unrecognized file type.\n"; 870 } 871 } 872 873 /// @brief Open file and figure out how to dump it. 874 static void DumpInput(StringRef file) { 875 // If file isn't stdin, check that it exists. 876 if (file != "-" && !sys::fs::exists(file)) { 877 errs() << ToolName << ": '" << file << "': " << "No such file\n"; 878 return; 879 } 880 881 if (MachOOpt && Disassemble) { 882 DisassembleInputMachO(file); 883 return; 884 } 885 886 // Attempt to open the binary. 887 ErrorOr<std::unique_ptr<Binary>> BinaryOrErr = createBinary(file); 888 if (std::error_code EC = BinaryOrErr.getError()) { 889 errs() << ToolName << ": '" << file << "': " << EC.message() << ".\n"; 890 return; 891 } 892 std::unique_ptr<Binary> binary = std::move(BinaryOrErr.get()); 893 894 if (Archive *a = dyn_cast<Archive>(binary.get())) 895 DumpArchive(a); 896 else if (ObjectFile *o = dyn_cast<ObjectFile>(binary.get())) 897 DumpObject(o); 898 else 899 errs() << ToolName << ": '" << file << "': " << "Unrecognized file type.\n"; 900 } 901 902 int main(int argc, char **argv) { 903 // Print a stack trace if we signal out. 904 sys::PrintStackTraceOnErrorSignal(); 905 PrettyStackTraceProgram X(argc, argv); 906 llvm_shutdown_obj Y; // Call llvm_shutdown() on exit. 907 908 // Initialize targets and assembly printers/parsers. 909 llvm::InitializeAllTargetInfos(); 910 llvm::InitializeAllTargetMCs(); 911 llvm::InitializeAllAsmParsers(); 912 llvm::InitializeAllDisassemblers(); 913 914 // Register the target printer for --version. 915 cl::AddExtraVersionPrinter(TargetRegistry::printRegisteredTargetsForVersion); 916 917 cl::ParseCommandLineOptions(argc, argv, "llvm object file dumper\n"); 918 TripleName = Triple::normalize(TripleName); 919 920 ToolName = argv[0]; 921 922 // Defaults to a.out if no filenames specified. 923 if (InputFilenames.size() == 0) 924 InputFilenames.push_back("a.out"); 925 926 if (!Disassemble 927 && !Relocations 928 && !SectionHeaders 929 && !SectionContents 930 && !SymbolTable 931 && !UnwindInfo 932 && !PrivateHeaders) { 933 cl::PrintHelpMessage(); 934 return 2; 935 } 936 937 std::for_each(InputFilenames.begin(), InputFilenames.end(), 938 DumpInput); 939 940 return 0; 941 } 942