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